Figure 1.
Gestures utilized in this study [35].
Citation: Cristiane Sá Roriz Fonteles, Richard H. Finnell, Timothy M. George, Raymond J. Harshbarger. Craniosynostosis: current conceptions and misconceptions[J]. AIMS Genetics, 2016, 3(1): 99-129. doi: 10.3934/genet.2016.1.99
| [1] | Fanqi Zeng, Wenli Geng, Ke An Liu, Boya Wang . Differential Harnack estimates for the semilinear parabolic equation with three exponents on $ \mathbb{R}^{n} $. Electronic Research Archive, 2025, 33(1): 142-157. doi: 10.3934/era.2025008 |
| [2] | Yony Raúl Santaria Leuyacc . Elliptic equations in $ \mathbb{R}^2 $ involving supercritical exponential growth. Electronic Research Archive, 2024, 32(9): 5341-5356. doi: 10.3934/era.2024247 |
| [3] | Ganglong Zhou . Group invariant solutions for the planar Schrödinger-Poisson equations. Electronic Research Archive, 2023, 31(11): 6763-6789. doi: 10.3934/era.2023341 |
| [4] | S. Bandyopadhyay, M. Chhetri, B. B. Delgado, N. Mavinga, R. Pardo . Maximal and minimal weak solutions for elliptic problems with nonlinearity on the boundary. Electronic Research Archive, 2022, 30(6): 2121-2137. doi: 10.3934/era.2022107 |
| [5] | Qingcong Song, Xinan Hao . Positive solutions for fractional iterative functional differential equation with a convection term. Electronic Research Archive, 2023, 31(4): 1863-1875. doi: 10.3934/era.2023096 |
| [6] | Youqian Bai, Yongkun Li . Almost periodic positive solutions of two generalized Nicholson's blowflies equations with iterative term. Electronic Research Archive, 2024, 32(5): 3230-3240. doi: 10.3934/era.2024148 |
| [7] | Yantong Guo, Quansheng Wu, Xiaofeng Wang . An extension of high-order Kou's method for solving nonlinear systems and its stability analysis. Electronic Research Archive, 2025, 33(3): 1566-1588. doi: 10.3934/era.2025074 |
| [8] | Yueying Liu, Mengping Sun, Zhen Wang, Xiangyun Lin, Cuihua Zhang . Nash equilibrium strategies for non-zero-sum differential games of SDEs with time-varying coefficient and infinite Markov jumps. Electronic Research Archive, 2025, 33(4): 2525-2542. doi: 10.3934/era.2025112 |
| [9] | Bidi Younes, Abderrahmane Beniani, Khaled Zennir, Zayd Hajjej, Hongwei Zhang . Global solution for wave equation involving the fractional Laplacian with logarithmic nonlinearity. Electronic Research Archive, 2024, 32(9): 5268-5286. doi: 10.3934/era.2024243 |
| [10] | Nelson Vieira, M. Manuela Rodrigues, Milton Ferreira . Time-fractional telegraph equation of distributed order in higher dimensions with Hilfer fractional derivatives. Electronic Research Archive, 2022, 30(10): 3595-3631. doi: 10.3934/era.2022184 |
Classifying upper limb gestures using multichannel surface electromyography (sEMG) poses a formidable challenge with implications for both diagnostic and therapeutic applications [1,2]. The inherent non-linear and stochastic nature of sEMG signals introduces a significant hurdle, complicating the precise categorization of upper limb gestures [3,4]. The complex interplay among various muscles, coupled with their complex signaling patterns, contributes to the complexity of extracting meaningful information from sEMG recordings [5,6]. This challenge is particularly pronounced in the context of employing sEMG signals for electromechanical hand prostheses, where precision and reliability are paramount [7,8]. The variability inherent in sEMG signals, influenced by factors such as muscle fatigue, electrode placement, and individual anatomical differences, adds layers of intricacy to the task of achieving accurate and robust gesture classification [9,10]. Effective classification methods for sEMG in diverse applications require innovative solutions to address the challenges posed by unique characteristics of upper limb gesture signals.
Machine learning and deep learning techniques have emerged as transformative tools in the domains of information processing, providing unprecedented capabilities for various tasks such as pattern recognition [11,12], feature extraction [13,14], and classification [15,16,17]. In the context of sEMG for gesture classification, numerous studies have explored the applicability of machine learning and deep learning techniques [18,19]. For instance, Saeed et al. [20] applied machine learning techniques to raw signals from the DB1 dataset, achieving accuracies of 85.41% and 91.14% using an artificial neural network (ANN) and linear discriminant analysis (LDA), respectively. Karnam et al. [21] achieved a classification accuracy of 88.8% on DB1 using K-nearest neighbours (KNN). Akmal et al. [22] explored training strategies for artificial neural networks (ANN), which emerged as a pivotal aspect in sEMG signal classification. This study analyzes twelve different training strategies, evaluating their performance on multiday EMG data. The results highlight the resilience of backpropagation and scaled conjugate gradient methods, providing valuable insights into optimal training approaches for efficient prosthetic control. SVM-based classification for prosthetic finger movements is investigated for real-time implementation [23]. Leveraging the stability and efficiency of SVM on a Raspberry Pi, this study achieved 78% classification accuracy. Inam et al. [24] explored gender-specific considerations in sEMG for upper-limb prosthetics. Evaluating EMG differences between males and females, this research employs an ANN for classification. While overall similarities are observed, certain features exhibit gender-specific variations, shedding light on the importance of tailored approaches for diverse user populations. In the domain of deep learning, Hu et al. [25] implemented a recurrent neural network (RNN) and a convolutional neural network (CNN) on sEMG signals from NinaPro DB1, attaining an accuracy of 87%. Pancholi et al. [26] applied a CNN model to various NinaPro datasets, achieving classification accuracies ranging from 81.67% to 99.11%. Cheng et al. [27] utilized the NinaPro DB1 dataset, applying CNN to sEMG-feature-images extracted from raw signals and achieving a classification accuracy of 82.54%. Additionally, Tong et al. [28] applied a CNN and long-short term memory (LSTM) based hybrid classifier on the NinaPro DB1 dataset, yielding an accuracy of 78.31%. [29] explored a concatenate feature fusion recurrent convolutional neural network (CFF-RCNN) to address this by introducing a concatenate feature fusion (CFF) strategy, achieving notable accuracies. CFF-RCNN surpasses reported results, achieving 88.87% on DB1, 99.51% on DB2, and 99.29% on DB4 with over 50 gestures. Qureshi et al. [30] introduced the efficient concatenated convolutional neural network (E2CNN) as a robust solution for real-time sEMG classification. By converting raw sEMG signals into Log-Mel spectrograms (LMS) and employing concatenation layers, E2CNN achieves high accuracy and response times for both non disabled and amputee subjects, positioning it as a potential candidate for prosthetic control in real-world scenarios. Another study [31] also explored improving myoelectric control in wearable prostheses using CNNs. By comparing multiday sEMG recordings, the proposed CNN exhibits superior accuracy for able-bodied and amputee subjects in within-day and between-day analyses. This research underscores the CNN's efficacy and computational efficiency, presenting a promising avenue for enhancing prosthetic hand control.
In the proposed study, we are introducing a dual-pathway convolutional neural network (DP-CNN) to classify sEMG signals from both healthy and amputee subjects. The novelty of this architecture resides in its ability to process Log-Mel spectrogram (LMS) images derived from raw multichannel EMG signals. Utilizing spectrograms instead of raw EMG signals has demonstrated enhanced performance across various studies in the field [26,30,31,32]. This trend is evident in multiple investigations where LMS significantly improved the efficiency of the classification system. LMS effectively captured the time-frequency characteristics of the EMG signals [31,33], and they are commonly used in various applications, including driver fatigue detection and EMG classification [31]. For instance, in a study focused on driver fatigue detection, a model based on LMS and a convolution recurrent neural network (CRNN) was proposed and demonstrated high accuracy in distinguishing between alert and fatigued states [33]. LMS also brings superior performance to the deep learning models. This is evidenced in a study in which a CNN achieved an impressive classification accuracy of over 90% on healthy and amputee subjects when applied to LMS-based images [30]. Furthermore, LMS images derived from EMG signals have been employed successfully in hybrid deep-learning methods to classify four different EMG-signal patterns, achieving significant classification results [34]. Moreover, LMS also offers a data augmentation method, which has been shown to significantly enhance the accuracy of deep learning models in classifying EMG signals [32].
The rationale behind this preference for spectrograms lies in their ability to capture intricate time-frequency characteristics inherent in EMG signals. Spectrograms, particularly LMS, provide a more comprehensive representation of the signal, highlighting nuanced patterns that might be obscured in raw EMG data. This richer representation aids in extracting deep information from the signals, contributing to increased effectiveness in classification tasks. EMG signals exhibit complex and dynamic frequency patterns that convey information about muscle contractions. The Mel-frequency bands in LMS provide an effective means of representing these intricate patterns, offering a more compact and discriminative representation of the signal. The logarithmic transformation in LMS compresses the higher frequencies, emphasizing the lower-frequency components. This is beneficial for EMG analysis, as the lower frequencies often contain valuable information related to muscle activities and gestures.
In alignment with these findings, the current study also adopts the use of spectrograms, specifically LMS, as the primary input for the proposed DP-CNN. The choice is grounded in the well-substantiated efficacy of spectrograms, particularly LMS, as demonstrated in previous literature. Spectrograms not only effectively capture time-frequency characteristics, but also offer superior performance to deep learning models. The LMS technique involves transforming raw EMG signals into a visual representation that emphasizes the spectral features relevant for gesture classification. By utilizing Mel-frequency bands, which are perceptually spaced to mimic human hearing, and applying a logarithmic scale, the LMS efficiently captures the frequency patterns within EMG signals.
As such, the adoption of LMS as the primary input in our proposed DP-CNN is grounded in evidence of its efficacy and superior performance in previous studies. By implementing this methodology, we aim to validate the performance of DP-CNN on LMS images extracted from sEMG signals, thereby offering a robust, reliable, and real-time solution for prosthetic control applications. The proposed DP-CNN architecture has been implemented on surface EMG (sEMG) of healthy and amputee subjects taken from NinaPro DB1, DB2, and DB3, respectively. The proposed DP-CNN is based on the CNN, and input is provided as a spectrogram images converted from raw EMG signals. The LMS images are obtained from the raw multichannel EMG signals available in NinaPro Databases 1 (DB1) and 3 (DB3). The contribution of this work is validating the performance of the DP-CNN implemented on LMS images extracted from bio-signals, in this case, EMG signals.
The main contributions of this paper are:
● Preprocessing raw EMG signals into Log-Mel spectrogram images, enhancing feature extraction.
● Development of a dual-pathway convolutional neural network (DP-CNN) combining convolutional and dense pathways for robust EMG signal classification.
● Extensive assessment of the proposed DP-CNN's effectiveness and generalizability across diverse datasets, including both healthy and amputee subjects.
● Thorough comparison of the DP-CNN's performance against prior works on the same dataset, providing insights into advancements.
● Benchmarking against pre-trained transfer learning models (AlexNet, MobileNet, VGG19, DenseNet121, ResNet50), showcasing the uniqueness and efficacy of the proposed approach.
The paper is organized as follows: in Section 2, we provide a detailed description of the methodology used in this study. We first discuss the datasets used and the pre-processing steps undertaken to prepare the data for analysis. We then provide the details of the proposed deep neural network (DNN) used for classification. Section 3 illustrates the results obtained from the proposed methodology and we discuss the implications of our findings and their potential applications. Section 4 concludes the paper.
In this study, we are using three datasets from a publicly available database: NinaPro Database 1 (DB1) [35], Database 2 (DB2) [36], and Database 3 (DB3) [37] for investigation and validation of the proposed methodology. The experiments were approved by the Ethics Commission of the state of Valais (Switzerland), the main place for data acquisition [35]. The details of each dataset are given below:
1) The first NinaPro database contains 10 repetitions of 52 different movements carried out by 27 intact subjects and serves as a standard dataset for the classification of myoelectric motions. The DB1 dataset includes three exercises categorized into (A) basic movements of the fingers, (B) isometric, isotonic hand configurations and basic wrist movements, and (C) grasping and functional movements. Ten Otto Bock MyoBock 13E200 electrodes are used to collect sEMG data; a Cyberglove 2 data glove is used to collect kinematic data. Each subject and exercise has a corresponding MATLAB file in the database that contains information about the subject, the exercise, the electrodes' sEMG signal, the 22 cyberglove sensors' uncalibrated signal, the subject's repeated movement, and the stimulus' repeated occurrence.
2) The DB2 dataset is composed of three types of exercises that are categorized into three groups: (A) basic movements of fingers and wrist, (B) grasping and functional movements, and (C) force patterns. To collect kinematic data, a dataglove (Cyberglove 2) and an accelerometer on the wrist were used, while a Delsys Trigno Wireless EMG system were utilized with 12 active double-differential wireless electrodes to record muscular activity. The sampling rate for sEMG signals is 2 kHz. Each exercise and subject have a synchronized MATLAB file that contains various variables, such as subject and exercise number, sEMG signal, kinematic information, inclinometer signal, movement repeated by the subject, and force recorded during the third exercise. Additionally, force sensor calibration values for the least and highest force are included in each file.
3) The third NinaPro database is a comprehensive resource for the development and evaluation of naturally-controlled non-invasive robotic hand prostheses. The experiment consists of the same three exercises as DB2: basic finger and wrist motions, grasping and functioning movements, and force patterns. The dataset includes 49 different movements (including rest) performed by 10 amputee participants with each movement repeated 6 times, and the movements were chosen from the hand taxonomy as well as literature on hand robotics. A Delsys Trigno Wireless EMG system with 12 active double-differential wireless electrodes was used to collect the muscular activity. The database provides one MATLAB file with synchronized variables for each exercise and subject, including subject and exercise number, sEMG signal, kinematic information, inclinometer signal, movement repeated by the subject, and force recorded during the third exercise. The collection also contains force sensor calibration information for the minimum and maximum force.
The number of repetitions of each gesture is ten in DB1, and six in DB2 and DB3. For this study, we randomly selected ten subjects from DB1 and DB2, five male and five female each; and ten from DB3 for validation of the proposed technique. We utilized the common movements from the three databases; exercise C of DB1 and exercise B of DB2 and DB3 are the only common gestures among the datasets. The exercise is composed of 23 hand gestures and is illustrated in Figure 1.
The data in this study was recorded using different numbers of channels in the three databases: 10 channels in DB1, and 12 channels in DB2 and DB3. The sampling rate for DB1 was 100 Hz, while for the DB2 and DB3 datasets, it was 2 kHz. The DB1 data was already shielded from power line noise; however, DB2 and DB3 are not shielded from power line noise [35]. Therefore, a 50 Hz second-order Butterworth notch filter was used for DB2 and DB3. Furthermore, DB1 data was filtered at 1 Hz using a second-order Butterworth filter [38]. In DB1, the signals available were root mean squared (RMS) values of raw signals, while in DB2 and DB3, raw EMG signals were available. To ensure optimal processing for EMG-based prosthetics, it is recommended to use signal segments with duration ranges from 150 ms to 250 ms [39,40]. Therefore, each raw EMG signal is divided into smaller segments with a duration of 200 ms with an overlapping increment of 50 ms. Since the number of repetitions are different in each dataset, we obtained different numbers of segmented signals. Specifically, we extracted $ 5750 \times 10 $ segmented signals from DB1, while we extracted 14,950 $ \times $ 12 segmented signals from DB2 and DB3 for each subject. Here, the $ 10 $ and $ 12 $ represent the number of channels in each signal.
After segmentation, each signal was converted into a Log-Mel spectrogram (LMS) image using the librosa library in Python. The LMS is a representation of the spectral content of a signal on a logarithmic frequency scale. By using LMS, we were able to analyze the frequency content of the segmented signals, which can provide more information for classification.
Let us define a segmented signal $ s_{w}(t) $ with length $ L $ and sampling frequency $ f_{s_{w}} $ in hertz. Its short-term fourier transform (STFT) $ S_{w} $ is then given by
|
$ Sw(ˉx,ˉy)=N−1∑t−0sw(t+xH)⋅w(t)⋅e−ι2πytτ. $
|
(2.1) |
Here, $ H \in \mathbb{N} $ represents the hop length, $ w:[0:\tau-1] \in \mathbb{R} $ is the Hann window defined as $ w = 0.5 - 0.5 \cos{(\frac{2 \pi t}{\tau-1})} $, where $ \tau \in \mathbb{N} $ is the length of $ w $, $ \bar{x} \in [0:\frac{L-\tau}{H}] $ denotes the time index, and $ \bar{y} \in [0:\frac{N}{2}] $ denotes the frequency index.
The short-term fourier transform spectrogram of $ S_{w} $ can be obtained as
|
$ SSTFT(˜x,˜y)=|Sw(ˉx,ˉy)|2. $
|
(2.2) |
The Mel spectrum and linear frequency are related by $ f_{mel} = 2959 \times \log_{10}(1+\frac{f}{700}) $. We can estimate the LMS using
|
$ SLM(ˊx,ˊy)=fc(x+1)∑f(y)=fc(x−1)log10(MFB(˜x,˜y)⋅SSTFT(˜x,˜y)). $
|
(2.3) |
Here, $ M_{FB}(\tilde{x}, \tilde{y}) $ is the Mel filter bank and can be estimated from
|
$ MFB(˜x,˜y)={f(˜y)−fc(˜x−1)fc(˜x)−fc(˜x−1)for fc(˜x−1)≤f(˜y)<fc(˜x)f(˜y)−fc(˜x+1)fc(˜x)−fc(˜x+1)for fc(˜x)≤f(˜y)<fc(˜x+1)0others. $
|
(2.4) |
Here, $ f(\tilde{y}) $ denotes the linear frequency, and $ f_{c}(\tilde{x}) = \tilde{x} \cdot \delta f_{mel} $ represents the center frequencies on the Mel-scale.
For each windowed signal in DB1, we convert it into an LMS individually. Then, this process is iteratively repeated for all ten channels, providing us with ten LMS images. We then combine these ten LMS vertically to form an input image, as shown in Figure 2(a). This process yields 5750 EMG images as LMS, which serve as the input dataset to the DP-CNN model for each subject in DB1. Similar to DB1, we apply the same technique to each windowed signal in DB2 and DB3, resulting in twelve LMS for each signal. These twelve LMS are combined vertically to form an image, as illustrated in Figure 2(b), (c). This process resulted in 14,950 EMG images as LMS for each subject in DB2 and DB3.
We propose a dual-pathway convolutional neural network (DP-CNN) with batch normalization and max-pooling functions along with dropout layers for classification of electromyogram (EMG) signals. The proposed model is designed to operate on LMS as input, with a fixed feature size of $ 224 \times 224 $. The DP-CNN is comprised of two pathways, namely the traditional convolutional pathway and a dense pathway, which are combined using a concatenation layer. The input to the DP-CNN is denoted as $ S_{LM}(\grave{x}, \grave{y}) $, and the convolution operation can be explained from [41] and is given as follows:
|
$ O=F(SLM(ˊx,ˊy)|a)=fn(f3(f2(SLMˊx,ˊy)|θ2)|θ3)θN). $
|
(2.5) |
Here, $ f_n $ represents the $ n $-th layer of the DP-CNN, and $ N $ is the total number of layers used, which in this study is set to 8. The parameter for the $ n $-th layer is denoted as $ \theta_1 = [X, b] $. We can express the convolutional layer operations in the following way:
|
$ O1=f1(SLM(ˊx,ˊy)SLM(ˊx,ˊy)n|θN)=h(SLM(ˊx,ˊy)l+b∗X). $
|
(2.6) |
Here, $ I_n $ represents the input of the $ n $th-layer, $ X $ is the corresponding filter, $ * $ denotes the valid convolution operation, $ h(\cdot) $ denotes the pointwise activation function, and $ b $ denotes the vector bias term.
In the proposed DP-CNN, the convolutional layers of the first pathway are estimated using
|
$ Ci,j=n∑k=1m∑l=1Wk,l⋅Ii+k−1,j+l−1+b, $
|
(2.7) |
where $ C_{i, j} $ is the value of the feature map for the $ i^{th} $ row and $ j^{th} $ column, $ W_{k, l} $ is the weight for the $ k^{th} $ row and $ l^{th} $ column of the filter, $ I_{i+k-1, j+l-1} $ is the value for the corresponding position in the input image, $ b $ is the bias, and $ n $ and $ m $ are the dimensions of the filter.
The activation function, rectified linear unit (ReLU), can be represented as
|
$ ReLU(x)=max(0,x). $
|
(2.8) |
The dense layer equation can be represented by
|
$ y=D⋅n∑i=1wi⋅xi+b, $
|
(2.9) |
where $ y $ is the output of the dense layer, $ w_i $ is the weight for the $ i^{th} $ unit, $ x_i $ is the input for the $ i^{th} $ unit, $ n $ is the number of units in the dense layer, and the dropout layer $ D $ is a binary mask with a probability of 0.2 to set the corresponding value to 0.
The concatenation function can be represented by
|
$ O=[fCL(fRL(I)),fDL(fFL(fCL(fRL(I)))], $
|
(2.10) |
where $ O $ is the final output of the concatenation function, and $ [\cdot, \cdot] $ represents the concatenation operation of two arrays.
Overall, the proposed DP-CNN can be represented by
|
$ O=[C1,1,C1,2,...,Cn,m,y1,y2,...,yk], $
|
(2.11) |
where
|
$ Ci,j=ReLU(n∑k=1m∑l=1Wk,l⋅Ii+k−1,j+l−1+b), $
|
(2.12) |
and
|
$ yi=ReLU(m∑j=1wj⋅xj+b)⋅D, $
|
(2.13) |
and $ x_j $ is the output of the previous dense layer.
The architecture of the DP-CNN model is shown in Figure 3. The primary advantage of the convolutional pathway is its ability to automatically learn and extract relevant features from input images without the need for manual feature engineering. This is accomplished through the use of convolutional and pooling layers, which learn local and global patterns in the input data. The model is able to learn increasingly complex representations of the input images by stacking multiple convolutional layers on top of each other.
On the other hand, the advantage of the dense pathway is its ability to detect global patterns and relationships in the input data. This is accomplished through the use of fully connected layers that learn to combine features from all parts of the input data. The model is able to learn increasingly complex representations of the input data by stacking multiple fully connected layers on top of each other. Another benefit of the dense pathway is its ability to handle input data of any size and shape, as long as it can be converted to a one-dimensional format. This makes the model suitable for a wide range of input data types, such as text or time series data, which may not have the same spatial structure as images. Furthermore, the use of dropout layers in the dense pathway helps to regularize the model and prevent overfitting. Dropout layers randomly set a fraction of the activations in the previous layer to zero during training, which helps to reduce co-adaptation between neurons and forces the model to learn more robust features.
The DP-CNN model can capture different types of information from the input images because it has two separate pathways for processing the input data. The convolutional pathway can extract spatial features from images by using convolutional and pooling layers, whereas the dense pathway can capture global patterns and relationships in input data by using fully connected layers. The model can make more accurate predictions by combining the outputs of both pathways.
The use of multiple pathways allows for greater regularization and reduces the risk of overfitting. By having two distinct pathways that process the input data in different ways, the model is less likely to memorize the training data and more likely to generalize well to new, previously unseen data. Furthermore, the use of dropout layers in the dense pathway provides additional regularization and helps to prevent overfitting.
The DP-CNN model can capture different types of information from the input images by using two separate pathways for processing the input data. The convolutional pathway can extract spatial features from images by using convolutional and pooling layers, whereas the dense pathway can capture global patterns and relationships in the input data by using fully connected layers. By combining the outputs of both pathways, the model is able to make more accurate predictions by leveraging both types of information.
The use of multiple pathways improves regularization and reduces the risk of overfitting. The model is less likely to memorize the training data and more likely to generalize well to new, unseen data by having two separate pathways that process the input data in different ways. Furthermore, the use of dropout layers in the dense pathway provides additional regularization and aids in the prevention of overfitting.
For training the proposed DP-CNN model, a total of 50 epochs are set and a batch size of 32 is used. The Adam optimizer is employed for optimization. The learning rate was set to $ 0.001 $. The input to the DP-CNN model is an image with a size of $ 224 \times 224 \times 3 $. Prior to feeding the images to the DP-CNN model, they are resized using a rescaling layer (RL), which scales the pixel values in the range of 0 to 1. This is done to ensure that the input data is within a suitable range for the neural network model. The rescaled input images are then provided as input to the DP-CNN model for training.
Convolutional pathway: The input images processed by the rescaling layer are fed to the convoultional pathway. Details of each layer are given below:
● Layer 1: The first layer is composed of a convolutional layer of $ 32 $ filters with a filter size of $ 9 \times 9 $, with ReLU and L2 regularization of $ 0.001 $. The convolutional layer is succeeded by a batch normalization layer. After that, a maximum pooling layer is used with a pool size of $ 4 \times 4 $.
● Layer 2: The second layer is similar to the first layer and is composed of same batch normalization and max-pooling layer, except the the number of filters are 48 and a filter size of $ 5 \times 5 $ is used.
● Layer 3: The third layer is made up of a convolutional layer of 48 filters with a filter size of $ 3 \times 3 $ with ReLU activations. A batch normalization layer is introduced after the convolutional layer, and then a max-pooling layer is used with a pool size of $ 2 \times 2 $.
Dense pathway: The same input images from the rescaling layer are converted to one-dimensional data using a flatten layer and then are fed to the dense pathway:
● Layers 1 & 2: The first layer in the second pathway is a fully connected dense layer composed of 24 units, and ReLU is used as the activation function. The second layer is a dropout layer with a rate of 0.2.
● Layers 3 & 4: The third and fourth layers are similar to the first and second layer, except the fully connected dense layers have 32 units.
● Layers 5 & 6: The fifth and sixth layers are similar to the above two layers, except the fully connected dense layers have 48 units.
Concatenation and classification layer: The data from the first and second pathways are combined using a concatenation layer. This layer allows the outputs of both pathways within a neural network to be combined into a single output. The last layer is composed of 23 units.
The performance of the proposed DP-CNN is evaluated on metrics based on multi-class classification. For this study, we are using mean accuracy, mean precision, mean recall, and mean F1-score along with their standard deviations [42].
Let $ C_{TP} $ be true positives and $ C_{TN} $ true negatives, while $ C_{FP} $ are false positives and $ C_{FN} $ false negatives. Multi-class classification accuracy can be estimated using
|
$ Accuracy=CTP+CTNCTP+CTN+CFP+CFN. $
|
(2.14) |
For binary-class classification $ n = 1 $, precision $ P_{n = 1}^{k} $ and recall $ R_{n = 1}^{k} $ for a particular class $ k $ can be calculated as follows:
|
$ Pkn=1=CkTPCkTP+CkFP, $
|
(2.15) |
|
$ Rkn=1=CkTPCkTP+CkFN. $
|
(2.16) |
For multi-class classification $ n = N $, precision $ P_{N}^{k} $ and recall $ R_{N}^{k} $ can be calculated as follows:
|
$ PkN=∑Kk=1Pkn=1K, $
|
(2.17) |
|
$ RkN=∑Kk=1Rkn=1K. $
|
(2.18) |
F1-score for multi-class classification is given as
|
$ F1−score=2×(PkN×RkN(PkN)−1+(RkN)−1). $
|
(2.19) |
Table 1 illustrates the developmental setup employed in this study.
| Components | Specifications |
| Processor | Intel (R) Xeon (R) CPU E5-2630 |
| RAM | 32 GB |
| GPU | NVIDIA Tesla T4 |
| Software | Ubuntu |
| CUDA 11.8 | |
| Python 3.11 | |
| Keras 2.90 | |
| TensorFlow 2.9.2 | |
| Matplotlib 3.6.1 | |
| Scikit-Learn 1.1.2 | |
| NumPy 1.23.4 | |
| Pandas 1.5.0 | |
| Seaborn 0.12.0 |
DownLoad:
CSV
We have evaluated the performance of the individual pathways of the proposed dual-pathway CNN, and compared the results of each pathway with the DP-CNN. For this purpose, we have only tested the performance on a single subject from each dataset. For all three datasets, the dense pathway and concatenation layer are removed and the performance is validated. Alternatively, in the second phase, the convolutional pathway along with the concatenation layer in DP-CNN are removed and the performance of only the dense path is validated. Finally, the performance of the complete DP-CNN is validated on a single subject from each dataset. Figure 4 illustrates the accuracies achieved by each pathway along with DP-CNN on one subject from all three datasets. It can be seen that neither pathway could achieve the desired accuracy compared to the complete DP-CNN.
We have evaluated the proposed DP-CNN on 30 subjects: 20 able-bodied subjects from DB1 and DB2, and 10 amputee subjects from DB3. To ensure diversity and balance, we randomly selected ten subjects from each of the two databases DB1 and DB2. In each database, we ensured that five males and five females were selected. Regarding DB3, we were able to select ten subjects out of the eleven available, but we could not include subject one due to the unavailability of the desired gestures in the dataset [35]. Additionally, subjects 7 and 8 had fewer electrodes, resulting in ten channels instead of twelve, but we still processed their data for our analysis. This approach allowed us to gather a representative sample that can help us draw accurate conclusions and insights from our research. The testing and training are done for each subject from both dataset. Each dataset is divided into a 70–30 split, where 70% is used for training and 30% for validation. This is done for each subject and the performance metrics are determined for each subject. Using the proposed DP-CNN, DB1, DB2, and DB3 are classified individually, and then the mean accuracy, mean precision, mean recall, and mean F1-score are determined from all subjects. Table 2 shows the accuracy, precision, recall, and F1-score for DB1 subjects. The proposed DP-CNN achieved a mean classification accuracy of 94.93 ± 1.71%, mean precision of 94.93 ± 1.71%, mean recall of 94.93 ± 1.71%, and mean F1-score of 94.93 ± 1.71% on the subjects of the DB1 dataset. Similarly, the proposed DP-CNN achieved a mean classification accuracy of 94.00 ± 3.56%, mean precision of 94.00 ± 3.56%, mean recall of 94.00 ± 3.56%, and mean F1-score of 94.00 ± 3.56% when applied to DB2 subjects. Table 3 illustrates the accuracy, precision, recall, and F1-score for DB2 subjects. Similarly, the proposed DP-CNN has achieved a mean classification accuracy of 85.36 ± 0.82%, mean precision of 85.35 ± 0.86%, mean recall of 85.34 ± 0.81%, and mean F1-score of 85.36 ± 0.82%, when applied to DB3 subjects. Table 4 illustrates the accuracy, precision, recall, and F1-score for DB3 subjects.
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 94.17% | 94.17% | 94.16% | 94.15% |
| 2 | 94.24% | 94.25% | 94.25% | 94.25% |
| 3 | 95.46% | 95.46% | 95.46% | 94.45% |
| 4 | 95.13% | 95.14% | 95.13% | 95.13% |
| 5 | 94.57% | 94.58% | 94.56% | 94.57% |
| 6 | 95.23% | 95.24% | 95.24% | 95.24% |
| 7 | 96.75% | 96.75% | 96.75% | 96.75% |
| 8 | 94.76% | 94.76% | 94.77% | 94.76% |
| 9 | 93.24% | 93.21% | 93.21% | 93.24% |
| 10 | 95.97% | 95.98% | 96.00% | 95.98% |
| Mean ± SD | 94.93 ± 1.71% | 94.93 ± 1.71% | 94.93 ± 1.71% | 94.93 ± 1.71% |
DownLoad:
CSV
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 96.27% | 96.27% | 96.19% | 96.25% |
| 2 | 93.95% | 93.95% | 93.95% | 93.95% |
| 3 | 91.39% | 91.39% | 91.36% | 91.32% |
| 4 | 97.25% | 97.25% | 97.26% | 97.95% |
| 5 | 92.63% | 92.65% | 92.62% | 92.63% |
| 6 | 91.00% | 91.01% | 91.00% | 90.29% |
| 7 | 96.96% | 96.99% | 96.98% | 96.98% |
| 8 | 92.67% | 92.67% | 92.67% | 92.67% |
| 9 | 95.17% | 95.17% | 95.17% | 95.17% |
| 10 | 92.75% | 92.72% | 92.72% | 92.72% |
| Mean ± SD | 94.00 ± 3.56% | 94.00 ± 3.56% | 94.00 ± 3.56% | 94.00 ± 3.56% |
DownLoad:
CSV
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 88.15% | 88.11% | 88.12% | 88.15% |
| 2 | 82.69% | 82.69% | 82.69% | 82.69% |
| 3 | 87.22% | 87.22% | 87.23% | 87.22% |
| 4 | 84.81% | 84.83% | 84.81% | 84.81% |
| 5 | 86.91% | 86.91% | 86.91% | 86.91% |
| 6 | 82.39% | 82.41% | 82.40% | 82.40% |
| 7 | 87.01% | 87.00% | 87.02% | 87.00% |
| 8 | 86.15% | 86.95% | 86.95% | 86.95% |
| 9 | 81.95% | 81.95% | 81.95% | 81.95% |
| 10 | 86.95% | 86.95% | 86.95% | 86.95% |
| Mean ± SD | 85.36 ± 0.82% | 85.35 ± 0.86% | 85.34 ± 0.81% | 85.36 ± 0.82% |
DownLoad:
CSV
In this study, we evaluated the performance of DP-CNN on LMS extracted from raw EMG signals of the publicly available NinaPro datasets. To assess the effectiveness of our approach, we compared the results with previous studies that utilized deep learning-based techniques on the same datasets. Although, to best of our knowledge, only one experiment has been conducted on the NinaPro database using Mel or Log-Mel spectrograms, we have compared our results with other studies that have employed any deep learning-based techniques on this database.
The proposed DP-CNN model was compared to earlier studies on the three databases: DB1, DB2, and DB3. Table 5 reveals that the DP-CNN model attained an accuracy of 94.93% on DB1, exceeding earlier research that ranged from 66.60% to 91.27%. On DB2, the DP-CNN model achieved the similar accuracy as on DB1, 94.00%, as shown in Table 6. Previous investigations achieved accuracies ranging from 60.27% to 89.45%. Finally, on DB3 the DP-CNN model attained an accuracy of 85.36%, as shown in Table 7, which is higher than earlier research, which ranged from 46.27% to 81.67%. These findings show that the suggested DP-CNN model is successful for gesture classification across all three datasets, with higher accuracy than earlier studies.
| Reference | Year | Dataset | Classifier | Accuracy |
| [38] | 2016 | DB1 | CNN | 66.60% |
| [43] | 2016 | DB1 | CNN | 77.80% |
| [44] | 2019 | DB1 | CNN | 85.00% |
| [28] | 2019 | DB1 | CNN-LSTM | 71.20% |
| [45] | 2019 | DB1 | Multi-view CNN | 88.20% |
| [46] | 2022 | DB1 | Deformable CNN | 81.80% |
| [47] | 2022 | DB1 | Dual-View CNN | 86.29% |
| [30] | 2023 | DB1 | E2CNN | 91.27% |
| This Work | 2023 | DB1 | DP-CNN | 94.93% |
DownLoad:
CSV
| Reference | Year | Dataset | Classifier | Accuracy |
| [35] | 2014 | DB2 | SVM | 60.27% |
| [48] | 2017 | DB2 | CNN | 60.31% |
| [45] | 2019 | DB2 | CNN | 83.70% |
| [26] | 2021 | DB2 | DLPR | 89.45% |
| [49] | 2022 | DB2 | CNN | 87.56% |
| This Work | 2023 | DB2 | DP-CNN | 94.00% |
DownLoad:
CSV
| Reference | Year | Dataset | Classifier | Accuracy |
| [35] | 2014 | DB3 | SVM | 46.27% |
| [48] | 2017 | DB3 | CNN | 73.31% |
| [45] | 2019 | DB3 | CNN | 64.30% |
| [26] | 2021 | DB3 | DLPR | 81.67% |
| [49] | 2022 | DB3 | CNN | 74.24% |
| This Work | 2023 | DB3 | DP-CNN | 85.36% |
DownLoad:
CSV
Table 8 shows the computational cost of our proposed model compared to other studies. When comparing our DP-CNN model to previous methods, we found that, although it achieves slightly lower accuracy in DB2 compared to CFF-RCNN [29], it has notable advantages in terms of computational efficiency. In DB2, where CFF-RCNN achieves an accuracy of 99.51%, our proposed DP-CNN maintains a competitive accuracy of 94.00%. However, DP-CNN outperforms CFF-RCNN in terms of training time, taking only 462.82 seconds compared to CFF-RCNN's 542.245 seconds. This represents a 14.56% reduction in training time, highlighting the efficiency of DP-CNN. Our proposed model also has a lower prediction time, further emphasizing its computational advantages.
| Reference | Year | Dataset | Classifier | Accuracy | Training Time | Prediction Time |
| [29] | 2022 | DB1 | CFF-RCNN | 88.87% | 1727.415 s | - |
| [29] | 2022 | DB2 | CFF-RCNN | 99.51% | 542.245 s | - |
| [30] | 2023 | DB1 | E2CNN | 91.27% | 38.062 s | 0.093 s |
| DB1 | 94.93% | 482.37 s | 0.863 s | |||
| This Work | 2023 | DB2 | DP-CNN | 94.00% | 462.82 s | 0.649 s |
| DB3 | 85.36% | 501.23 s | 0.895 s |
DownLoad:
CSV
Since the sEMG signals are converted to images, we have compared the performance of our proposed DP-CNN with pre-trained transfer learning models that has been trained on million of images. The transfer learning models utilized for comparison in this study include AlexNet, MobileNet, VGG19, DenseNet121, and ResNet50. The performance of the DP-CNN on each database (DB1, DB2, and DB3) was compared to these transfer learning models. As shown in Figure 5, the DP-CNN outperformed all other models on DB1 and DB2, achieving accuracies of 94.00% and 94.93%, respectively. However, on DB3, the DP-CNN achieved an accuracy of 85.36%, which was lower than the accuracy achieved by AlexNet (87.29%) and VGG19 (87.96%). The other models achieved lower accuracies on all three databases compared to the DP-CNN.
The proposed DP-CNN model's performance on the NinaPro datasets was examined in this work. The method involved converting sEMG signals to LMS and then classifying them with the DP-CNN model. To assess the proposed model's performance, the findings were compared to earlier studies that used deep learning-based techniques on the same datasets.
The DP-CNN architecture comprises two pathways, namely the convolutional and dense pathways. These pathways are capable of capturing local patterns in the EMG signals and global patterns and relationships in the signals, respectively. Through the integration of the outputs from both pathways, the model can enhance its predictive accuracy and improve its capacity to classify EMG signals. The utilization of batch normalization and dropout layers in both pathways serves to regularize the model and mitigate overfitting.
The suggested DP-CNN model's findings were compared to earlier studies on the NinaPro DB1, DB2, and DB3. The accuracy gained by each model was used in the comparison. The suggested DP-CNN model surpassed earlier studies in terms of accuracy across all three datasets, according to the results. On DB1, the DP-CNN model achieved an accuracy of 94.93%, the highest accuracy achieved on this dataset thus far. On DB2, the DP-CNN model achieved an accuracy of 94.00%, the highest accuracy ever achieved on this dataset. Finally, on DB3, the DP-CNN model attained an accuracy of 85.36%, the highest accuracy achieved on this dataset thus far. These findings show that the DP-CNN model performs well for gesture classification across all three datasets.
In this study, we addressed the challenges associated with machine and deep learning algorithms, especially their performance decline in the face of increased number of classes, diverse data collected over multiple days, and population differences. Recognizing the need for a robust learning system, we proposed and applied a dual-pathway convolutional neural network (DP-CNN) to diverse datasets featuring both able-bodied and amputee subjects. The DP-CNN operated on Log-Mel spectrogram-based images derived from surface electromyography signals obtained from NinaPro DB1 and DB3. The results were benchmarked against other CNN models implemented on the same datasets, revealing the superior performance of the proposed DP-CNN. In DB1, the DP-CNN achieved a remarkable mean classification accuracy of 94.93%, a substantial 28.33% increase from the baseline and a noteworthy improvement of 6.73% over the previous highest accuracy. Similar advancements were observed in DB2 and DB3, showcasing the model's consistent and robust performance across datasets. The architecture of the DP-CNN, featuring convolutional and dense pathways, played a pivotal role in capturing both local and global patterns within EMG signals. Integrating outputs from these pathways enhanced predictive accuracy and classification capabilities. The incorporation of batch normalization and dropout layers in both pathways further contributed to model regularization and mitigated overfitting. Comparisons with prior studies on NinaPro DB1, DB2, and DB3 demonstrated that the DP-CNN consistently outperformed earlier models in terms of accuracy. Achieving the highest accuracy on each dataset—94.93% on DB1, 94.00% on DB2, and 85.36% on DB3—the DP-CNN showcased its effectiveness in gesture classification. Additionally, a comparative analysis against pre-trained transfer learning models, including AlexNet, MobileNet, VGG19, DenseNet121, and ResNet50, highlighted the DP-CNN's supremacy in terms of accuracy on DB1 and DB2. Although, on DB3, it slightly lagged behind specific models, but the overall performance improvement in sEMG-based gesture detection was significant. The DP-CNN model, equipped with dual pathways, proved to be an effective solution for improving the accuracy and robustness of sEMG-based gesture classification. This study contributes valuable insights into advancing machine learning techniques for prosthetic control applications, emphasizing the practical significance of employing sophisticated architectures like the DP-CNN in real-world scenarios.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
This work was supported by the National Research Foundation of Korea (NRF) funded by the Korean government (MSIT) (No. RS-2023-00243034). This work was supported by the Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korean government (MSIT) (2021-0-00755, Dark data analysis technology for data scale and accuracy improvement). The authors also express their gratitude to Princess Nourah bint Abdulrahman University Researchers Supporting Project Number PNURSP2024R104, Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.
The authors declare there is no conflict of interest.
| [1] |
Slater BJ, Lenton KA, Kwan MD, et al. (2008) Cranial sutures: a brief review. Plast Reconstr Surg 121: 170e-178e. doi: 10.1097/01.prs.0000304441.99483.97
|
| [2] |
Twigg SR, Wilkie AO (2015) A genetic-pathophysiological framework for craniosynostosis. Am J Hum Genet 97: 359-377. doi: 10.1016/j.ajhg.2015.07.006
|
| [3] | Hamm JA, Robin NH (2015) Newborn craniofacial malformations. Orofacial clefting and craniosynostosis. Clin Perinotol 42: 321-336. |
| [4] |
Persing JA, Jane JA, Shaffrey M (1989) Virchow and the pathogenesis of craniosynostosis: a translation of his work. Plast Reconstr Surg 83: 738-742. doi: 10.1097/00006534-198904000-00025
|
| [5] |
Patel A, Terner J, Travieso R, et al. (2012) On Bernard Sarnat's 100th birthday: pathology and management of craniosynostosis. J Craniofac Surg 23: 105-112. doi: 10.1097/SCS.0b013e318240fb0d
|
| [6] |
Cohen MM Jr (2005) Editorial: perspectives on craniosynostosis. Am J Med Genet A 136A: 313-326. doi: 10.1002/ajmg.a.30757
|
| [7] |
Boulet SL, Rasmussen SA, Honein MA (2008) A population-based study of craniosynostosis in metropolitan Atlanta, 1989-2003. Am J Med Genet A 146A: 984-991. doi: 10.1002/ajmg.a.32208
|
| [8] |
Lee HQ, Hutson JM, Wray AC, et al. (2012) Changing epidemiology of nonsyndromic craniosynostosis and revisiting the risk factors. J Craniofac Surg 23: 1245-1251. doi: 10.1097/SCS.0b013e318252d893
|
| [9] | Chim H, Manjila S, Cohen AR, et al. (2011) Molecular signaling in pathogenesis of craniosynostosis: the role of fibroblast growth factor and transforming growth factor-b. Neurosurg Focus 31: E7. |
| [10] | Kabbani H, Raghuveer TS (2004) Craniosynostosis. Am Fam Physician 69: 2863-2870. |
| [11] |
Kimonis V, Gold JA, Hoffman TL, et al. (2007) Genetics of craniosynostosis. Semin Pediatr Neurol 14: 150-161. doi: 10.1016/j.spen.2007.08.008
|
| [12] |
French LR, Jackson IT, Melton LJ (1990) A population-based study of craniosynostosis. J Clin Epidemiol 43: 69-73. doi: 10.1016/0895-4356(90)90058-W
|
| [13] |
Alazami AM, Seidahmed MZ, Alzahrani F, et al. (2014) Novel IFT122 mutation associated with impaired ciliogenesis and cranioectodermal dysplasia. Mol Genet Genomic Med 2: 103-106. doi: 10.1002/mgg3.44
|
| [14] |
Currarino G (2007) Sagittal synostosis in X-linked hypophosphatemic rickets and related diseases. Pediatr Radiol 37: 805-812. doi: 10.1007/s00247-007-0503-4
|
| [15] | Adès LC, Sullivan K, Biggin A, et al. (2006) FBN1, TGFBR1, and the Marfan-craniosynostosis/mental retardation disorders revisited. Am J Med Genet A 140: 1047-1058. |
| [16] |
Iqbal Z, Cejudo-Martin P, de Brouwer A, et al. (2010) Disruption of the podosome adaptor protein TKS4 (SH3PXD2B) causes the skeletal dysplasia, eye, and cardiac abnormalities of Frank-Ter Haar Syndrome. Am J Hum Genet 86: 254-261. doi: 10.1016/j.ajhg.2010.01.009
|
| [17] |
Agrawal D, Steinbok P, Cochrane DD (2006) Diagnosis of isolated sagittal synostosis: are radiographic studies necessary? Childs Nerv Syst 22: 375-378. doi: 10.1007/s00381-005-1243-0
|
| [18] |
Levi B, Wan DC, Wong VW, et al. (2012) Cranial suture biology: from pathways to patient care. J Craniofac Surg 23: 13-19. doi: 10.1097/SCS.0b013e318240c6c0
|
| [19] |
Van der Meulen J (2012) Metopic synostosis. Childs Nerv Syst 28: 1359-1367. doi: 10.1007/s00381-012-1803-z
|
| [20] |
Yee ST, Fearon JA, Gosain AK, et al. (2015) Classification and Management of Metopic Craniosynostosis. J Craniofac Surg 26: 1812-1817. doi: 10.1097/SCS.0000000000001866
|
| [21] |
Senarath-Yapa K, Chung MT, McArdle A, et al. (2012) Craniosynostosis: molecular pathways and future pharmacologic therapy. Organogenesis 8: 103-113. doi: 10.4161/org.23307
|
| [22] |
Jehee FS, Johnson D, Alonso LG, et al. (2005) Molecular screening for microdeletions at 9p22-p24 and 11q23-q24 in a large cohort of patients with trigonocephaly. Clin Genet 67: 503-510. doi: 10.1111/j.1399-0004.2005.00438.x
|
| [23] | Hoischen A, van Bon BW, Rodriguez-Santiago B, et al. (2011) De novo nonsense mutations in ASXL1 cause Bohring-Opitz syndrome. Nat Genet43: 729-731. |
| [24] |
Malpuech G, Demeocq F, Palcoux JB, et al. (1983) A previously undescribed autosomal recessive multiple congenital anomalies/mental retardation (MCA/MR) syndrome with growth failure, lip/palate cleft(s), and urogenital anomalies. Am J Med Genet 16: 475-480. doi: 10.1002/ajmg.1320160405
|
| [25] |
Vissers LE, Cox TC, Maga AM, et al. (2011) Heterozygous mutations of FREM1 are associated with an increased risk of isolated metopic craniosynostosis in humans and mice. PLoS Genet 7: e1002278. doi: 10.1371/journal.pgen.1002278
|
| [26] | McDonald-McGinn DM, Feret H, Nah HD, et al. (2010) Metopic craniosynostosis due to mutations in GLI3: A novel association. Am J Med GenetA 152A: 1654-1660. |
| [27] | Twigg SR, Lloyd D, Jenkins D, et al. (2012) Mutations in multidomain protein MEGF8 identify a Carpenter syndrome subtype associated with defective lateralization. Am J Hum Genet91: 897-905. |
| [28] |
Wilkie AO, Byren JC, Hurst JA, et al. (2010) Prevalence and complications of single-gene and chromosomal disorders in craniosynostosis. Pediatrics 126: e391-400. doi: 10.1542/peds.2009-3491
|
| [29] |
Lajeunie E, Le Merrer M, Bonaïti-Pellie C, et al. (1995) Genetic study of nonsyndromic coronal craniosynostosis. Am J Med Genet 55: 500-504. doi: 10.1002/ajmg.1320550422
|
| [30] |
Ziyadeh J, Le Merrer M, Robert M, et al. (2013) Mucopolysaccharidosis type I and craniosynostosis. Acta Neurochir (Wien) 155: 1973-1976. doi: 10.1007/s00701-013-1831-9
|
| [31] | Villarreal DD, Villarreal H, Paez AM, et al. (2013) A patient with a unique frameshift mutation in GPC3, causing Simpson-Golabi-Behmel syndrome, presenting with craniosynostosis, penoscrotal hypospadias, and a large prostatic utricle. Am J Med Genet A 161A: 3121-3125. |
| [32] |
Shahinian HK, Jaekle R, Suh RH, et al. (1998) Obstetrical factors governing the etiopathogenesis of lambdoid synostosis. Am J Perinatol 15: 281-286. doi: 10.1055/s-2007-993943
|
| [33] |
Czerwinski M, Kolar JC, Fearon JA (2011) Complex craniosynostosis. Plast Reconstr Surg 128: 955-961. doi: 10.1097/PRS.0b013e3182268ca6
|
| [34] |
Chai Y, Maxson RE Jr (2006) Recent advances in craniofacial morphogenesis. Dev Dyn 235: 2353-2375. doi: 10.1002/dvdy.20833
|
| [35] |
Schweizer G, Ayer-LeLièvre C, Le Douarin NM (1983) Restrictions in developmental capacities in the dorsal root ganglia during the course of development. Cell Differ 13:191-200. doi: 10.1016/0045-6039(83)90089-1
|
| [36] |
Spears R, Svoboda KKH (2005) Growth factors and signaling proteins in craniofacial development. Semin Orthod 11: 184-198. doi: 10.1053/j.sodo.2005.07.003
|
| [37] | Delannet M, Duband JL (1992) Transforming growth factor-beta control of cell-substratum adhesion during avian neural crest cell emigration in vitro. Development 116: 275-287. |
| [38] |
Baird A (1994) Fibroblast growth factors: activities and significance of nonneurotrophin neurotrophic growth factors. Curr Opin Neurobiol 4: 78-86. doi: 10.1016/0959-4388(94)90035-3
|
| [39] | Morrison-Graham K, Schatteman GC, Bork T, et al. (1992) A PDGF receptor mutation in the mouse (Patch) perturbs the development of a non-neuronal subset of neural crest-derived cells. Development 115: 133-142. |
| [40] |
Bodine PV (2008) Wnt signalling controlo f bone cell apoptosis. Cell Res 18: 248-253. doi: 10.1038/cr.2008.13
|
| [41] |
Noden DM (1983) The role of the neural crest in patterning of avian cranial skeletal, connective, and muscle tissues. Dev Biol 96: 144-165. doi: 10.1016/0012-1606(83)90318-4
|
| [42] |
Deckelbaum RA, Holmes G, Zhao Z, et al. (2012) Regulation of cranial morphogenesis and cell fate at the neural crest-mesoderm boundary by engrailed 1. Development 139: 1346-1358. doi: 10.1242/dev.076729
|
| [43] |
Twigg SR, Kan R, Babbs C, et al. (2004) Mutations of ephrin-B1 (EFNB1), a marker of tissue boundary formation, cause craniofrontonasal syndrome. Proc Natl Acad Sci U S A 101: 8652 - 8657. doi: 10.1073/pnas.0402819101
|
| [44] |
Merrill AE, Bochukova EG, Brugger SM, et al. (2006) Cell mixing at a neural crest-mesoderm boundary and deficient ephrin-Eph signaling in the pathogenesis of craniosynostosis. Hum Mol Genet 15: 1319-1328. doi: 10.1093/hmg/ddl052
|
| [45] | Noden DM (1978) The control of avian cephalic neural crest cytodifferentiation. I. Skeletal and connective tissues. Dev Biol 67: 296-312. |
| [46] | Noden DM (1988) Interactions and fates of avian craniofacial mesenchyme. Development 103: 121-140. |
| [47] | Le Lievre CS (1978) Participation of neural crest-derived cells in the genesis of the skull in birds. J Embryol Exp Morphol 47: 17-37. |
| [48] | Couly GF, Coltey PM, Le Douarin NM (1993) The triple origin of skull in higher vertebrates: a study in quail-chick chimeras. Development 117: 409-429. |
| [49] |
Jiang X, Iseki S, Maxson RE, et al. (2002) Tissue origins and interactions in the mammalian skull vault. Dev Biol 241: 106-116. doi: 10.1006/dbio.2001.0487
|
| [50] |
Nie X (2005) Cranial base in craniofacial development: developmental features, influence on facial growth, anomaly, and molecular basis. Acta Odontol Scand 63:127-135. doi: 10.1080/00016350510019847
|
| [51] | Sasaki T, Ito Y, Bringas P Jr, et al. (2006) TGF (beta)-mediated FGF signaling is crucial for regulating cranial neural crest cell proliferation during frontal bone development. Development 133: 371-381. |
| [52] |
Bernard S, Loukas M, Rizk E, et al. (2015) The human occipital bone: review and update on its embryology and molecular development. Childs Nerv Syst 31: 2217-2223. doi: 10.1007/s00381-015-2870-8
|
| [53] |
Mishina Y, Snider TN (2014) Neural crest cell signaling pathways critical to cranial bone development and pathology. Exp Cell Res 325: 138-147. doi: 10.1016/j.yexcr.2014.01.019
|
| [54] |
Zhao H, Feng J, Ho TV, et al. (2015) The suture provides a niche for mesenchymal stem cells of craniofacial bones. Nat Cell Biol 17: 386-396. doi: 10.1038/ncb3139
|
| [55] |
Ogle RC, Tholpady SS, McGlynn KA, et al. (2004) Regulation of cranial suture morphogenesis. Cells Tissues Organs 176: 54-66. doi: 10.1159/000075027
|
| [56] |
Opperman LA, Rawlins JT (2005) The extracellular matrix environment in suture morphogenesis and growth. Cells Tissues Organs 181: 127-135. doi: 10.1159/000091374
|
| [57] | Manzanares MC, Goret-Nicaise M, Dhem A (1988) Metopic sutural closure in the human skull. J Anat 161: 203-215. |
| [58] |
Sahar DE, Longaker MT, Quarto N (2005) Sox9 neural crest determinant gene controls patterning and closure of the posterior frontal cranial suture. Dev Biol 280: 344-361. doi: 10.1016/j.ydbio.2005.01.022
|
| [59] | Opperman LA (2000) Cranial sutures as intramembranous bone growth sites. Dev Dyn 219: 472-485. |
| [60] |
Gagan JR, Tholpady SS, Ogle RC (2007) Cellular dynamics and tissue interactions of the dura mater during head development. Birth Defects Res C Embryo Today 81: 297-304. doi: 10.1002/bdrc.20104
|
| [61] |
Richtsmeier JT, Flaherty K (2013) Hand in glove: brain and skull in development and dysmorphogenesis. Acta Neuropathol 125: 469-489. doi: 10.1007/s00401-013-1104-y
|
| [62] |
Mao JJ, Nah HD (2004) Growth and development: hereditary and mechanical modulations. Am J Orthod Dentofacial Orthop 125: 676-689. doi: 10.1016/j.ajodo.2003.08.024
|
| [63] | Todd TW, Lyon DW (1925) Cranial suture closure: Its progress and age relationship. Part II. Ectocranial closure in adult males of white stock. Am J Phys Anthropol 8: 23-45. |
| [64] | Ting MC, Wu NL, Roybal PG, et al. (2009) EphA4 as an effector of Twist1 in the guidance of osteogenic precursor cells during calvarial bone growth and in craniosynostosis. Development 136: 855-864. |
| [65] |
Greenwald JA, Mehrara BJ, Spector JA, et al. (2000) Regional differentiation of cranial suture-associated dura mater in vivo and in vitro: implications for suture fusion and patency. J Bone Miner Res 15: 2413-2430. doi: 10.1359/jbmr.2000.15.12.2413
|
| [66] |
Zimmerman B, Moegelin A, de Souza P, et al. (1998) Morphology of the development of the sagittal suture of mice. Anat Embryol (Berl) 197:155-165. doi: 10.1007/s004290050127
|
| [67] | Senarath-Yapa K, Li S, Meyer NP, et al. (2013) Integration of multiple signaling pathways determines differences in the osteogenic potential and tissue regeneration of neural crest-derived and mesoderm-derived calvarial bones. Int J Mol Sci 14: 5978-5997. |
| [68] | Quarto N, Wan DC, Kwan MD, et al. (2010) Origin matters: differences in embryonic tissue origin and Wnt signaling determine the osteogenic potential and healing capacity of frontal and parietal calvarial bones. J Bone Miner Res 25: 1680-1694. |
| [69] | Cohen Jr MM (1993) Sutural biology and the correlates of craniosynostosis. Am J Med Genet 47: 581-616. |
| [70] | Pritchard JJ, Scott JH, Girgis FG (1956) The structure and development of cranial and facial sutures. J Anat 90: 73 - 86. |
| [71] | Koskinen-Moffett LK, Moffett BC Jr, Graham JM Jr (1982) Cranial synostosis and intra-uterine compression: a developmental study of human sutures. Prog Clin Biol Res 101: 365-378. |
| [72] |
Kokich VG (1976) Age changes in the human frontozygomatic suture from 20 to 95 years. Am J Orthod 69: 411-430. doi: 10.1016/0002-9416(76)90209-8
|
| [73] |
Morriss-Kay GM, Wilkie AO (2005) Growth of the normal skull vault and its alteration in craniosynostosis: insights from human genetics and experimental studies. J Anat 207: 637-653. doi: 10.1111/j.1469-7580.2005.00475.x
|
| [74] |
Nayak SR, Krishnamurthy A, Kumar SJM, et al. (2007) The mendosal suture of the occipital bone: occurrence in Indian population, embryology and clinical significance. Surg Radiol Anat 29: 329-332. doi: 10.1007/s00276-007-0216-2
|
| [75] |
Kirmi O, Lo SJ, Johnson D, et al. (2009) Craniosynostosis: a radiological and surgical perspective. Semin Ultrasound CT MR 30: 492-512. doi: 10.1053/j.sult.2009.08.002
|
| [76] |
Vu HL, Panchal J, Parker EE, et al. (2001) The timing of physiologic closure of the metopic suture: a review of 159 patients using reconstructed 3D CT scans of the craniofacial region. J Craniofac Surg 12: 527-532. doi: 10.1097/00001665-200111000-00005
|
| [77] |
Park SS, Beyer RP, Smyth MD, et al. (2015) Osteoblast differentiation profiles define sex specific gene expression patterns in craniosynostosis. Bone 76: 169-176. doi: 10.1016/j.bone.2015.03.001
|
| [78] | Kim HJ, Rice DP, Kettunen PJ, et al. (1998) FGF-, BMP- and Shh-mediated signalling pathways in the regulation of cranial suture morphogenesis and calvarial bone development. Development 125: 1241-1251. |
| [79] |
Mansukhani A, Ambrosetti D, Holmes G, et al. (2005) Sox2 induction by FGF and FGFR2 activating mutations inhibits Wnt signaling and osteoblast differentiation. J Cell Biol 168: 1065-1076. doi: 10.1083/jcb.200409182
|
| [80] |
Liu B, Yu HM, Hsu W (2007) Craniosynostosis caused by Axin2 deficiency is mediated through distinct functions of beta-catenin in proliferation and differentiation. Dev Biol 301: 298-308. doi: 10.1016/j.ydbio.2006.10.018
|
| [81] |
McMahon JA, Takada S, Zimmerman LB, et al. (1998) Noggin-mediated antagonism of BMP signaling is required for growth and patterning of the neural tube and somite. Genes Dev 12: 1438-1452. doi: 10.1101/gad.12.10.1438
|
| [82] |
Rosen V (2006) BMP and BMP inhibitors in bone. Ann N Y Acad Sci 1068: 19-25. doi: 10.1196/annals.1346.005
|
| [83] |
Warren SM, Brunet LJ, Harland RM, et al. (2003) The BMP antagonist noggin regulates cranial suture fusion. Nature 422: 625-629. doi: 10.1038/nature01545
|
| [84] |
Rice DP, Kim HJ, Thesleff I (1999) Apoptosis in murine calvarial bone and suture development. Eur J Oral Sci 107: 265-275. doi: 10.1046/j.0909-8836.1999.eos107406.x
|
| [85] |
Agresti M, Gosain AK (2005) Detection of apoptosis in fusing versus nonfusing mouse cranial sutures. J Craniofac Surg 16: 572-578. doi: 10.1097/01.scs.0000168759.61947.7f
|
| [86] |
Opperman LA, Sweeney TM, Redmon J, et al. (1993) Tissue interactions with underlying dura mater inhibit osseous obliteration of developing cranial sutures. Dev Dyn 198: 312-322. doi: 10.1002/aja.1001980408
|
| [87] | Opperman LA, Passarelli RW, Morgan EP, et al. (1995) Cranial sutures require tissue interactions with dura mater to resist osseous obliteration in vitro. J Bone Miner Res 10: 1978-1987. |
| [88] |
Tholpady SS, Freyman TF, Chachra D, et al. (2007) Tensional forces influence gene expression and sutural state of rat calvariae in vitro. Plast Reconstr Surg 120: 601-611. doi: 10.1097/01.prs.0000270284.69632.6b
|
| [89] |
Mao JJ (2002) Mechanobiology of craniofacial sutures. J Dent Res 81: 810-816. doi: 10.1177/154405910208101203
|
| [90] |
Kopher RA, Mao JJ (2003) Suture growth modulated by the oscillatory component of micromechanical strain. J Bone Min Res 18: 521-528. doi: 10.1359/jbmr.2003.18.3.521
|
| [91] |
Mao JJ, Wang X, Kopher RA, et al. (2003) Strain induced osteogenesis in the cranial suture upon controlled delivery of low-frequency cyclic forces. Front Biosci 8: A10-A17. doi: 10.2741/917
|
| [92] |
Ingber DE, Tensegrity I (2003) Cell structure and hierarchical systems biology. J Cell Sci 116: 1157-1173. doi: 10.1242/jcs.00359
|
| [93] | Carinci P, Becchetti E, Bodo M (2000) Role of the extracellular matrix and growth factors in skull morphogenesis and in the pathogenesis of craniosynostosis. Int J Dev Biol 44: 715-723. |
| [94] |
Salmivirta M, Jalkanen M (1995) Syndecan family of cell surface proteoglycans: developmentally regulated receptors for extracellular effector molecules. Experientia 51: 863-872. doi: 10.1007/BF01921737
|
| [95] |
Bidwell JP, Pavalko FM (2010) The Load-Bearing Mechanosome Revisited. Clin Rev Bone Miner Metab 8: 213-223. doi: 10.1007/s12018-010-9075-1
|
| [96] | Carinci P, Becchetti E, Baroni T, et al. (2007) Extracellular matrix and growth factors in the pathogenesis of some craniofacial malformations. Eur J Histochem 51: 105-116. |
| [97] | Bodo M, Carinci F, Baroni T, et al. (1997) Apert's syndrome: differential in vitro production of matrix macromolecules and its regulation by interleukins. Eur J Clin Invest 27: 36-42. |
| [98] |
Bodo M, Carinci F, Baroni T, et al. (1998) Interleukin pattern of Apert fibroblasts in vitro. Eur J Cell Biol 75: 383-388. doi: 10.1016/S0171-9335(98)80072-1
|
| [99] |
Locci P, Baroni T, Pezzetti F, et al. (1999) Differential in vitro phenotype pattern, transforming growth factorbeta(1) activity and mRNA expression of transforming growth factor-beta(1) in Apert osteoblasts. Cell Tissue Res 297: 475-483. doi: 10.1007/s004410051374
|
| [100] |
Bodo M, Baroni T, Carinci F, et al. (1999) A regulatory role of fibroblast growth factor in the expression of decorin, biglycan, betaglycan and syndecan in osteoblasts from patients with Crouzon's syndrome. Eur J Cell Biol 78: 323-330. doi: 10.1016/S0171-9335(99)80066-1
|
| [101] |
Su N, Jin M, Chen L (2014) Role of FGF/FGFR signaling in skeletal development and homeostasis: learning from mouse models. Bone Res 2: 14003. doi: 10.1038/boneres.2014.3
|
| [102] |
Liu YH, Tang Z, Kundu RK, et al. (1999) Msx2 gene dosage influences the number of proliferative osteogenic cells in growth centers of the developing murine skull: a possible mechanism for MSX2-mediated craniosynostosis in humans. Dev Biol 205: 260-274. doi: 10.1006/dbio.1998.9114
|
| [103] |
Cohen MM Jr (1997) Transforming growth factor beta s and fibroblast growth factors and their receptors: role in sutural biology and craniosynostosis. J Bone Miner Res 12: 322-331. doi: 10.1359/jbmr.1997.12.3.322
|
| [104] |
Jabs EW, Müller U, Li X, et al. (1993) A mutation in the homeodomain of the human MSX2 gene in a family affected with autosomal dominant craniosynostosis. Cell 75: 443-450. doi: 10.1016/0092-8674(93)90379-5
|
| [105] |
Warman ML, Mulliken JB, Hayward PG, et al. (1993) Newly recognized autosomal dominant craniosynostotic syndrome. Am J Med Genet 46: 444-449. doi: 10.1002/ajmg.1320460420
|
| [106] |
Ma L, Golden S, Wu L, et al. (1996) The molecular basis of Boston-type craniosynostosis: the Pro148-->His mutation in the N-terminal arm of the MSX2 homeodomain stabilizes DNA binding without altering nucleotide sequence preferences. Hum Mol Genet 5: 1915-1920. doi: 10.1093/hmg/5.12.1915
|
| [107] | Wilkie AO, Tang Z, Elanko N, et al. (2000) Functional haploinsufficiency of the human homeobox gene MSX2 causes defects in skull ossification. Nat Genet 24: 387-390. |
| [108] |
Alappat S, Zhang ZY, Chen YP (2003) Msx homeobox gene family and craniofacial development. Cell Res 13: 429-442. doi: 10.1038/sj.cr.7290185
|
| [109] | MacKenzie A, Ferguson MW, Sharpe PT (1992) Expression patterns of the homeobox gene, Hox-8, in the mouse embryo suggest a role in specifying tooth initiation and shape. Development 115: 403-420. |
| [110] | Jowett AK, Vainio S, Ferguson MW, et al. (1993) Epithelial-mesenchymal interactions are required for msx 1 and msx 2 gene expression in the developing murine molar tooth. Development 117: 461-470. |
| [111] |
Catron KM, Wang H, Hu G, et al. (1996) Comparison of MSX-1 and MSX-2 suggests a molecular basis for functional redundancy. Mech Dev 55: 185-199. doi: 10.1016/0925-4773(96)00503-5
|
| [112] |
Zhang H, Catron KM, Abate-Shen C (1996) A role for the Msx-1 homeodomain in transcriptional regulation: residues in the N-terminal arm mediate TATA binding protein interaction and transcriptional repression. Proc Natl Acad Sci U S A 93: 1764-1769. doi: 10.1073/pnas.93.5.1764
|
| [113] |
Newberry EP, Latifi T, Battaile JT, et al. (1997) Structure-function analysis of Msx2-mediated transcriptional suppression. Biochemistry 36: 10451-10462. doi: 10.1021/bi971008x
|
| [114] |
Catron KM, Zhang H, Marshall SC, et al. (1995) Transcriptional repression by Msx-1 does not require homeodomain DNA-binding sites. Mol Cell Biol 15: 861-871. doi: 10.1128/MCB.15.2.861
|
| [115] |
Wang W, Chen X, Xu H, et al. (1996) Msx3: a novel murine homologue of the Drosophila msh homeobox gene restricted to the dorsal embryonic central nervous system. Mech Dev 58: 203-215. doi: 10.1016/S0925-4773(96)00562-X
|
| [116] | MacKenzie A, Ferguson MW, Sharpe PT (1991) Hox-7 expression during murine craniofacial development. Development 113: 601-611. |
| [117] |
Satokata I, Maas R (1994) Msx1 deficient mice exhibit cleft palate and abnormalities of craniofacial and tooth development. Nat Genet 6: 348-356. doi: 10.1038/ng0494-348
|
| [118] |
Turner N, Grose R (2010) Fibroblast growth factor signalling: from development to cancer. Nat Rev Cancer 10: 116-129. doi: 10.1038/nrc2780
|
| [119] |
Isacchi A, Bergonzoni L, Sarmientos P (1990) Complete sequence of a human receptor for acidic and basic fibroblast growth factors. Nucleic Acids Res 18: 1906. doi: 10.1093/nar/18.7.1906
|
| [120] |
Pasquale EB, Singer SJ (1989) Identification of a developmentally regulated protein-tyrosine kinase by using anti-phosphotyrosine antibodies to screen a cDNA expression library. Proc Natl Acad Sci USA 86: 5449-5453. doi: 10.1073/pnas.86.14.5449
|
| [121] |
Pasquale EB (1990) A distinctive family of embryonic protein-tyrosine kinase receptors. Proc Natl Acad Sci USA 87: 5812-5816. doi: 10.1073/pnas.87.15.5812
|
| [122] | Partanen J, Makela TP, Eerola E, et al. (1991) FGFR-4, a novel acidic fibroblast growth factor receptor with a distinct expression pattern. EMBO J 10: 1347-1354. |
| [123] |
Wiedemann M, Trueb B (2000) Characterization of a novel protein (FGFRL1) from human cartilage related to FGF receptors. Genomics 69: 275-279. doi: 10.1006/geno.2000.6332
|
| [124] |
Sleeman M, Fraser J, McDonald M, et al. (2001) Identification of a new fibroblast growth factor receptor, FGFR5. Gene 271: 171-182. doi: 10.1016/S0378-1119(01)00518-2
|
| [125] |
Trueb B (2011) Biology of FGFRL1, the fifth fibroblast growth factor receptor. Cell Mol Life Sci 68: 951-964. doi: 10.1007/s00018-010-0576-3
|
| [126] |
Teven CM, Farina EM, Rivas J, et al. (2014) Fibroblast growth factor (FGF) signaling in development and skeletal diseases. Genes Dis 1: 199-213. doi: 10.1016/j.gendis.2014.09.005
|
| [127] |
Beenken A, Mohammadi M (2009) The FGF family: biology, pathophysiology and therapy. Nat Rev Drug Discov 8: 235-253. doi: 10.1038/nrd2792
|
| [128] |
Green PJ, Walsh FS, Doherty P (1996) Promiscuity of fibroblast growth factor receptors. Bioessays 18: 639-646. doi: 10.1002/bies.950180807
|
| [129] | Iseki S, Wilkie AO, Heath JK, et al. (1997) Fgfr2 and osteopontin domains in the developing skull vault are mutually exclusive and can be altered by locally applied FGF2. Development 124: 3375-3384. |
| [130] |
Delezoide AL, Benoist-Lasselin C, Legeai-Mallet L, et al. (1998) Spatio-temporal expression of FGFR 1, 2 and 3 genes during human embryo-fetal ossification. Mech Dev 77: 19-30. doi: 10.1016/S0925-4773(98)00133-6
|
| [131] |
Gosain AK, Recinos RF, Agresti M, et al. (2004) TGF-beta1, FGF-2, and receptor mRNA expression in suture mesenchyme and dura versus underlying brain in fusing and nonfusing mouse cranial sutures. Plast Reconstr Surg 113: 1675-1684. doi: 10.1097/01.PRS.0000117362.33347.43
|
| [132] |
Crossley PH, Minowada G, McArthur CA, et al. (1996) Roles for FGF8 in the induction, initiation, and maintenance of chick limb development. Cell 84: 127-136. doi: 10.1016/S0092-8674(00)80999-X
|
| [133] | Sarkar S, Petiot A, Copp A, et al. (2001) FGF2 promotes skeletogenic differentiation of cranial neural crest cells. Development 128: 2143-2152. |
| [134] |
Jabs EW, Li X, Scott AF, et al. (1994) Jackson-Weiss and Crouzon syndromes are allelic with mutations in fibroblast growth factor receptor 2. Nat Genet 8: 275-279. doi: 10.1038/ng1194-275
|
| [135] |
Muenke M, Schell U, Hehr A, et al. (1994) A common mutation in the fibroblast growth factor receptor 1 gene in Pfeiffer syndrome. Nat Genet 8: 269-274. doi: 10.1038/ng1194-269
|
| [136] |
Reardon W, Winter RM, Rutland P, et al. (1994) Mutations in the fibroblast growth factor receptor 2 gene cause Crouzon syndrome. Nat Genet 8: 98-103. doi: 10.1038/ng0994-98
|
| [137] |
Shiang R, Thompson LM, Zhu YZ, et al. (1994) Mutations in the transmembrane domain of FGFR3 cause the most common genetic form of dwarfism, achondroplasia. Cell 78: 335-342. doi: 10.1016/0092-8674(94)90302-6
|
| [138] | Iseki S, Wilkie AO, Morriss-Kay GM (1999) Fgfr1 and Fgfr2 have distinct differentiation- and proliferation-related roles in the developing mouse skull vault. Development 126: 5611-5620. |
| [139] | Kan SH, Elanko N, Johnson D, et al. (2002) Genomic screening of fibroblast growth-factor receptor 2 reveals a wide spectrum of mutations in patients with syndromic craniosynostosis. Am J Hum Genet 70:472-86. |
| [140] |
Anderson J, Burns HD, Enriquez-Harris P, et al. (1998) Apert syndrome mutations in fibroblast growth factor receptor 2 exhibit increased affinity for FGF ligand. Hum Mol Genet 7: 1475-1483. doi: 10.1093/hmg/7.9.1475
|
| [141] |
Yu K, Herr AB, Waksman G, et al. (2000) Loss of fibroblast growth factor receptor 2 ligand-binding specificity in Apert syndrome. Proc Natl Acad Sci U S A 97:14536-1441. doi: 10.1073/pnas.97.26.14536
|
| [142] |
Neilson KM, Friesel RE (1995) Constitutive activation of fibroblast growth factor receptor-2 by a point mutation associated with Crouzon syndrome. J Biol Chem 270: 26037-26040. doi: 10.1074/jbc.270.44.26037
|
| [143] |
Robertson SC, Meyer AN, Hart KC, et al. (1998) Activating mutations in the extracellular domain of the fibroblast growth factor receptor 2 function by disruption of the disulfide bond in the third immunoglobulin-like domain. Proc Natl Acad Sci U S A 95: 4567-4572. doi: 10.1073/pnas.95.8.4567
|
| [144] |
Oldridge M, Zackai EH, McDonald-McGinn DM, et al. (1999) De novo alu-element insertions in FGFR2 identify a distinct pathological basis for Apert syndrome. Am J Hum Genet 64: 446-461. doi: 10.1086/302245
|
| [145] |
Wilkie AOM, Slaney SF, Oldridge M, et al. (1995) Apert syndrome results from localized mutations of FGFR2 and is allelic with Crouzon syndrome. Nat Genet 9: 165-172. doi: 10.1038/ng0295-165
|
| [146] |
Przylepa KA, Paznekas W, Zhang M, et al. (1996) Fibroblast growth factor receptor 2 mutations in Beare-Stevenson cutis gyrata syndrome. Nat Genet 13: 492-494. doi: 10.1038/ng0896-492
|
| [147] | Li X, Lewanda AF, Eluma F, et al. (1994) Two craniosynostotic syndrome loci, Crouzon and Jackson-Weiss, map to chromosome 10q23-q26. Genomics 22: 418-424. |
| [148] |
Dionne CA, Modi WS, Crumley G, et al. (1992) BEK, a receptor for multiple members of the fibroblast growth factor (FGF) family, maps to human chromosome 10q25.3→q26. Cytogenet Cell Genet 60: 34-36. doi: 10.1159/000133290
|
| [149] |
McGillivray G, Savarirayan R, Cox T C, et al. (2005) Familial scaphocephaly syndrome caused by a novel mutation in the FGFR2 tyrosine kinase domain. J Med Genet 42: 656-662. doi: 10.1136/jmg.2004.027888
|
| [150] |
Robin NH, Feldman GJ, Mitchell HF, et al. (1994) Linkage of Pfeiffer syndrome to chromosome 8 centromere and evidence for genetic heterogeneity. Hum Mol Genet 3: 2153-2158. doi: 10.1093/hmg/3.12.2153
|
| [151] |
Zhou YX, Xu X, Chen L, et al. (2000) A Pro250Arg substitution in mouse Fgfr1 causes increased expression of Cbfa1 and premature fusion of calvarial sutures. Hum Mol Genet 9: 2001-2008. doi: 10.1093/hmg/9.13.2001
|
| [152] | Schell U, Hehr A, Feldman GJ, et al. (1995) Mutations in FGFR1 and FGFR2 cause familial and sporadic Pfeiffer syndrome. Hum Mol Genet 4: 323-328. |
| [153] |
Lajeunie E, Ma HW, Bonaventure J, et al. (1995) FGFR2 mutations in Pfeiffer syndrome. Nat Genet 9: 108. doi: 10.1038/ng0295-108
|
| [154] |
Rutland P, Pulleyn LJ, Reardon W, et al. (1995) Identical mutations in the FGFR2 gene cause both Pfeiffer and Crouzon syndrome phenotypes. Nat Genet 9: 173-176. doi: 10.1038/ng0295-173
|
| [155] |
Cunningham ML, Seto ML, Ratisoontorn C, et al. (2007) Syndromic craniosynostosis: from history to hydrogen bonds. Orthod Craniofacial Res 10: 67-81. doi: 10.1111/j.1601-6343.2007.00389.x
|
| [156] |
Merrill AE, Sarukhanov A, Krejci P, et al. (2012) Bent bone dysplasia-FGFR2 type, a distinct skeletal disorder, has deficient canonical FGF signaling. Am J Hum Genet 90: 550-557. doi: 10.1016/j.ajhg.2012.02.005
|
| [157] |
White KE, Cabral JM, Davis SI, et al. (2005) Mutations that cause osteoglophonic dysplasia define novel roles for FGFR1 in bone elongation. Am J Hum Genet 76: 361-367. doi: 10.1086/427956
|
| [158] | Farrow EG, Davis SI, Mooney SD, et al. (2006) Extended mutational analyses of FGFR1 in osteoglophonic dysplasia. Am J Med Genet A 140: 537-539. |
| [159] | Ibrahimi OA, Zhang F, Eliseenkova AV, et al. (2004) Proline to arginine mutations in FGF receptors 1 and 3 result in Pfeiffer and Muenke craniosynostosis syndromes through enhancement of FGF binding affinity. Hum Mol Genet 13: 69-78. |
| [160] |
Cunningham ML, Seto ML, Ratisoontorn C, et al. (2007) Syndromic craniosynostosis: from history to hydrogen bonds. Orthod Craniofac Res 10: 67-81. doi: 10.1111/j.1601-6343.2007.00389.x
|
| [161] |
El Ghouzzi V, Lajeunie E, Le Merrer M, et al. (1999) Mutations within or upstream of the basic helix-loop-helix domain of the TWIST gene are specific to Saethre-Chotzen syndrome. Europ J Hum Genet 7: 27-33. doi: 10.1038/sj.ejhg.5200240
|
| [162] | Gripp KW, Zackai EH, Stolle CA (2000) Mutations in the human TWIST gene. Hum Mutat 15: 150-155, 2000. |
| [163] |
Paznekas WA, Cunningham ML, Howard TD, et al. (1998) Genetic heterogeneity of Saethre-Chotzen syndrome, due to TWIST and FGFR mutations. Am J Hum Genet 62: 1370-1380. doi: 10.1086/301855
|
| [164] |
Johnson D, Iseki S, Wilkie AO, et al. (2000) Expression patterns of Twist and Fgfr1, -2 and -3 in the developing mouse coronal suture suggest a key role for twist in suture initiation and biogenesis. Mech Dev 91: 341-345. doi: 10.1016/S0925-4773(99)00278-6
|
| [165] | Rice DP, Aberg T, Chan Y, et al. (2000) Integration of FGF and TWIST in calvarial bone and suture development. Development 127: 1845-1855. |
| [166] | Kress W, Schropp C, Lieb G, et al. (2006) Saethre-Chotzen syndrome caused by TWIST 1 gene mutations: functional differentiation from Muenke coronal synostosis syndrome. Eur J Hum Genet 14: 39-48. |
| [167] |
Sharma VP, Fenwick AL, Brockop MS, et al. (2013) Mutations in TCF12, encoding a basic helix-loop-helix partner of TWIST1, are a frequent cause of coronal craniosynostosis. Nat Genet 45: 304-307. doi: 10.1038/ng.2531
|
| [168] | Di Rocco F, Baujat G, Arnaud E, et al. (2014) Clinical spectrum and outcomes in families with coronal synostosis and TCF12 mutations. Eur J Hum Genet 22: 1413-1416. |
| [169] | Quintero-Rivera F, Robson CD, Reiss RE, et al. (2006) Intracranial anomalies detected by imaging studies in 30 patients with Apert syndrome. Am J Med Genet A 140: 1337-1338. |
| [170] |
Meyers GA, Orlow SJ, Munro IR, et al. (1995) Fibroblast growth factor receptor 3 (FGFR3) transmembrane mutation in Crouzon syndrome with acanthosis nigricans. Nat Genet 11: 462-464. doi: 10.1038/ng1295-462
|
| [171] |
Arnaud-Lopez L, Fragoso R, Mantilla-Capacho J, et al. (2007) Crouzon with acanthosis nigricans: further delineation of the syndrome. Clin Genet 72: 405-410. doi: 10.1111/j.1399-0004.2007.00884.x
|
| [172] | Cohen MM Jr (1999) Let's call it 'Crouzonodermoskeletal syndrome' so we won't be prisoners of our own conventional terminology. Am J Med Genet 84: 74. |
| [173] | Cohen MM Jr, Kreiborg S (1992) Birth prevalence studies of the Crouzon syndrome: comparison of direct and indirect methods. Clin Genet 41: 12-15. |
| [174] | Fernandes MB, Maximino LP, Perosa GB, et al. (2016) Apert and Crouzon syndromes-cognitive development, brain abnormalities, and molecular aspects. Am J Med Genet Ain press. |
| [175] |
Cohen MM Jr (1993) Pfeiffer syndrome update, clinical subtypes, and guidelines for differential diagnosis. Am J Med Genet 45: 300-307. doi: 10.1002/ajmg.1320450305
|
| [176] |
Vogels A, Fryns JP (2006) Pfeiffer syndrome. Orphanet J Rare Dis 1: 19. doi: 10.1186/1750-1172-1-19
|
| [177] |
Jenkins D, Seelow D, Jehee FS, et al. (2007) RAB23 mutations in Carpenter syndrome imply an unexpected role for hedgehog signaling in cranial-suture development and obesity. Am J Hum Genet 80: 1162-1170. doi: 10.1086/518047
|
| [178] |
Bellus GA, Gaudenz K, Zackai EH, et al. (1996) Identical mutations in three diferente fibroblast growth fator receptor genes in autosomal dominant craniosynostosis syndromes. Nat Genet 14: 174-176. doi: 10.1038/ng1096-174
|
| [179] | Muenke M, Gripp KW, McDonald-McGinn DM, et al. (1997) A unique point mutation in the fibroblast growth factor receptor 3 gene (FGFR3) defines a new craniosynostosis syndrome. Am J Hum Genet 60: 555-564. |
| [180] | Kruszka P, Addissie YA, Yarnell CM, et al. (2016) Muenke syndrome: An international multicenter natural history study. Am J Med Genet A in press. |
| [181] | Czeizel AE, Elek C, Susanszky E (1993) Birth prevalence study of Apert syndrome. Am J Med Genet 45: 392. |
| [182] | Orphanet report series, Prevalence of rare diseases: Bibliographic data. July 2015, Number 1. Available from:http://www.orpha.net/orphacom/cahiers/docs/GB/Prevalence_of_rare_diseases_by_alphabetical_list.pdf |
| [183] | Carpenter syndrome. Available from: http://www.orpha.net/consor/cgi-bin/OC_Exp.php?Lng=GB&Expert=65759 |
Figures(3) / Tables(2)
Cristiane Sá Roriz Fonteles, Richard H. Finnell, Timothy M. George, Raymond J. Harshbarger. Craniosynostosis: current conceptions and misconceptions[J]. AIMS Genetics, 2016, 3(1): 99-129. doi: 10.3934/genet.2016.1.99
| Components | Specifications |
| Processor | Intel (R) Xeon (R) CPU E5-2630 |
| RAM | 32 GB |
| GPU | NVIDIA Tesla T4 |
| Software | Ubuntu |
| CUDA 11.8 | |
| Python 3.11 | |
| Keras 2.90 | |
| TensorFlow 2.9.2 | |
| Matplotlib 3.6.1 | |
| Scikit-Learn 1.1.2 | |
| NumPy 1.23.4 | |
| Pandas 1.5.0 | |
| Seaborn 0.12.0 |
DownLoad:
CSV
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 94.17% | 94.17% | 94.16% | 94.15% |
| 2 | 94.24% | 94.25% | 94.25% | 94.25% |
| 3 | 95.46% | 95.46% | 95.46% | 94.45% |
| 4 | 95.13% | 95.14% | 95.13% | 95.13% |
| 5 | 94.57% | 94.58% | 94.56% | 94.57% |
| 6 | 95.23% | 95.24% | 95.24% | 95.24% |
| 7 | 96.75% | 96.75% | 96.75% | 96.75% |
| 8 | 94.76% | 94.76% | 94.77% | 94.76% |
| 9 | 93.24% | 93.21% | 93.21% | 93.24% |
| 10 | 95.97% | 95.98% | 96.00% | 95.98% |
| Mean ± SD | 94.93 ± 1.71% | 94.93 ± 1.71% | 94.93 ± 1.71% | 94.93 ± 1.71% |
DownLoad:
CSV
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 96.27% | 96.27% | 96.19% | 96.25% |
| 2 | 93.95% | 93.95% | 93.95% | 93.95% |
| 3 | 91.39% | 91.39% | 91.36% | 91.32% |
| 4 | 97.25% | 97.25% | 97.26% | 97.95% |
| 5 | 92.63% | 92.65% | 92.62% | 92.63% |
| 6 | 91.00% | 91.01% | 91.00% | 90.29% |
| 7 | 96.96% | 96.99% | 96.98% | 96.98% |
| 8 | 92.67% | 92.67% | 92.67% | 92.67% |
| 9 | 95.17% | 95.17% | 95.17% | 95.17% |
| 10 | 92.75% | 92.72% | 92.72% | 92.72% |
| Mean ± SD | 94.00 ± 3.56% | 94.00 ± 3.56% | 94.00 ± 3.56% | 94.00 ± 3.56% |
DownLoad:
CSV
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 88.15% | 88.11% | 88.12% | 88.15% |
| 2 | 82.69% | 82.69% | 82.69% | 82.69% |
| 3 | 87.22% | 87.22% | 87.23% | 87.22% |
| 4 | 84.81% | 84.83% | 84.81% | 84.81% |
| 5 | 86.91% | 86.91% | 86.91% | 86.91% |
| 6 | 82.39% | 82.41% | 82.40% | 82.40% |
| 7 | 87.01% | 87.00% | 87.02% | 87.00% |
| 8 | 86.15% | 86.95% | 86.95% | 86.95% |
| 9 | 81.95% | 81.95% | 81.95% | 81.95% |
| 10 | 86.95% | 86.95% | 86.95% | 86.95% |
| Mean ± SD | 85.36 ± 0.82% | 85.35 ± 0.86% | 85.34 ± 0.81% | 85.36 ± 0.82% |
DownLoad:
CSV
| Reference | Year | Dataset | Classifier | Accuracy |
| [38] | 2016 | DB1 | CNN | 66.60% |
| [43] | 2016 | DB1 | CNN | 77.80% |
| [44] | 2019 | DB1 | CNN | 85.00% |
| [28] | 2019 | DB1 | CNN-LSTM | 71.20% |
| [45] | 2019 | DB1 | Multi-view CNN | 88.20% |
| [46] | 2022 | DB1 | Deformable CNN | 81.80% |
| [47] | 2022 | DB1 | Dual-View CNN | 86.29% |
| [30] | 2023 | DB1 | E2CNN | 91.27% |
| This Work | 2023 | DB1 | DP-CNN | 94.93% |
DownLoad:
CSV
| Reference | Year | Dataset | Classifier | Accuracy |
| [35] | 2014 | DB2 | SVM | 60.27% |
| [48] | 2017 | DB2 | CNN | 60.31% |
| [45] | 2019 | DB2 | CNN | 83.70% |
| [26] | 2021 | DB2 | DLPR | 89.45% |
| [49] | 2022 | DB2 | CNN | 87.56% |
| This Work | 2023 | DB2 | DP-CNN | 94.00% |
DownLoad:
CSV
| Reference | Year | Dataset | Classifier | Accuracy |
| [35] | 2014 | DB3 | SVM | 46.27% |
| [48] | 2017 | DB3 | CNN | 73.31% |
| [45] | 2019 | DB3 | CNN | 64.30% |
| [26] | 2021 | DB3 | DLPR | 81.67% |
| [49] | 2022 | DB3 | CNN | 74.24% |
| This Work | 2023 | DB3 | DP-CNN | 85.36% |
DownLoad:
CSV
| Reference | Year | Dataset | Classifier | Accuracy | Training Time | Prediction Time |
| [29] | 2022 | DB1 | CFF-RCNN | 88.87% | 1727.415 s | - |
| [29] | 2022 | DB2 | CFF-RCNN | 99.51% | 542.245 s | - |
| [30] | 2023 | DB1 | E2CNN | 91.27% | 38.062 s | 0.093 s |
| DB1 | 94.93% | 482.37 s | 0.863 s | |||
| This Work | 2023 | DB2 | DP-CNN | 94.00% | 462.82 s | 0.649 s |
| DB3 | 85.36% | 501.23 s | 0.895 s |
DownLoad:
CSV
| Components | Specifications |
| Processor | Intel (R) Xeon (R) CPU E5-2630 |
| RAM | 32 GB |
| GPU | NVIDIA Tesla T4 |
| Software | Ubuntu |
| CUDA 11.8 | |
| Python 3.11 | |
| Keras 2.90 | |
| TensorFlow 2.9.2 | |
| Matplotlib 3.6.1 | |
| Scikit-Learn 1.1.2 | |
| NumPy 1.23.4 | |
| Pandas 1.5.0 | |
| Seaborn 0.12.0 |
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 94.17% | 94.17% | 94.16% | 94.15% |
| 2 | 94.24% | 94.25% | 94.25% | 94.25% |
| 3 | 95.46% | 95.46% | 95.46% | 94.45% |
| 4 | 95.13% | 95.14% | 95.13% | 95.13% |
| 5 | 94.57% | 94.58% | 94.56% | 94.57% |
| 6 | 95.23% | 95.24% | 95.24% | 95.24% |
| 7 | 96.75% | 96.75% | 96.75% | 96.75% |
| 8 | 94.76% | 94.76% | 94.77% | 94.76% |
| 9 | 93.24% | 93.21% | 93.21% | 93.24% |
| 10 | 95.97% | 95.98% | 96.00% | 95.98% |
| Mean ± SD | 94.93 ± 1.71% | 94.93 ± 1.71% | 94.93 ± 1.71% | 94.93 ± 1.71% |
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 96.27% | 96.27% | 96.19% | 96.25% |
| 2 | 93.95% | 93.95% | 93.95% | 93.95% |
| 3 | 91.39% | 91.39% | 91.36% | 91.32% |
| 4 | 97.25% | 97.25% | 97.26% | 97.95% |
| 5 | 92.63% | 92.65% | 92.62% | 92.63% |
| 6 | 91.00% | 91.01% | 91.00% | 90.29% |
| 7 | 96.96% | 96.99% | 96.98% | 96.98% |
| 8 | 92.67% | 92.67% | 92.67% | 92.67% |
| 9 | 95.17% | 95.17% | 95.17% | 95.17% |
| 10 | 92.75% | 92.72% | 92.72% | 92.72% |
| Mean ± SD | 94.00 ± 3.56% | 94.00 ± 3.56% | 94.00 ± 3.56% | 94.00 ± 3.56% |
| Subject | Accuracy | Precision | Recall | F1-score |
| 1 | 88.15% | 88.11% | 88.12% | 88.15% |
| 2 | 82.69% | 82.69% | 82.69% | 82.69% |
| 3 | 87.22% | 87.22% | 87.23% | 87.22% |
| 4 | 84.81% | 84.83% | 84.81% | 84.81% |
| 5 | 86.91% | 86.91% | 86.91% | 86.91% |
| 6 | 82.39% | 82.41% | 82.40% | 82.40% |
| 7 | 87.01% | 87.00% | 87.02% | 87.00% |
| 8 | 86.15% | 86.95% | 86.95% | 86.95% |
| 9 | 81.95% | 81.95% | 81.95% | 81.95% |
| 10 | 86.95% | 86.95% | 86.95% | 86.95% |
| Mean ± SD | 85.36 ± 0.82% | 85.35 ± 0.86% | 85.34 ± 0.81% | 85.36 ± 0.82% |
| Reference | Year | Dataset | Classifier | Accuracy |
| [38] | 2016 | DB1 | CNN | 66.60% |
| [43] | 2016 | DB1 | CNN | 77.80% |
| [44] | 2019 | DB1 | CNN | 85.00% |
| [28] | 2019 | DB1 | CNN-LSTM | 71.20% |
| [45] | 2019 | DB1 | Multi-view CNN | 88.20% |
| [46] | 2022 | DB1 | Deformable CNN | 81.80% |
| [47] | 2022 | DB1 | Dual-View CNN | 86.29% |
| [30] | 2023 | DB1 | E2CNN | 91.27% |
| This Work | 2023 | DB1 | DP-CNN | 94.93% |
| Reference | Year | Dataset | Classifier | Accuracy |
| [35] | 2014 | DB2 | SVM | 60.27% |
| [48] | 2017 | DB2 | CNN | 60.31% |
| [45] | 2019 | DB2 | CNN | 83.70% |
| [26] | 2021 | DB2 | DLPR | 89.45% |
| [49] | 2022 | DB2 | CNN | 87.56% |
| This Work | 2023 | DB2 | DP-CNN | 94.00% |
| Reference | Year | Dataset | Classifier | Accuracy |
| [35] | 2014 | DB3 | SVM | 46.27% |
| [48] | 2017 | DB3 | CNN | 73.31% |
| [45] | 2019 | DB3 | CNN | 64.30% |
| [26] | 2021 | DB3 | DLPR | 81.67% |
| [49] | 2022 | DB3 | CNN | 74.24% |
| This Work | 2023 | DB3 | DP-CNN | 85.36% |
| Reference | Year | Dataset | Classifier | Accuracy | Training Time | Prediction Time |
| [29] | 2022 | DB1 | CFF-RCNN | 88.87% | 1727.415 s | - |
| [29] | 2022 | DB2 | CFF-RCNN | 99.51% | 542.245 s | - |
| [30] | 2023 | DB1 | E2CNN | 91.27% | 38.062 s | 0.093 s |
| DB1 | 94.93% | 482.37 s | 0.863 s | |||
| This Work | 2023 | DB2 | DP-CNN | 94.00% | 462.82 s | 0.649 s |
| DB3 | 85.36% | 501.23 s | 0.895 s |
