Review Special Issues

Artificial intelligence applied to neuroimaging data in Parkinsonian syndromes: Actuality and expectations

  • Received: 27 November 2020 Accepted: 25 January 2021 Published: 19 February 2021
  • Idiopathic Parkinson's Disease (iPD) is a common motor neurodegenerative disorder. It affects more frequently the elderly population, causing a significant emotional burden both for the patient and caregivers, due to the disease-related onset of motor and cognitive disabilities. iPD's clinical hallmark is the onset of cardinal motor symptoms such as bradykinesia, rest tremor, rigidity, and postural instability. However, these symptoms appear when the neurodegenerative process is already in an advanced stage. Furthermore, the greatest challenge is to distinguish iPD from other similar neurodegenerative disorders, "atypical parkinsonisms", such as Multisystem Atrophy, Progressive Supranuclear Palsy and Cortical Basal Degeneration, since they share many phenotypic manifestations, especially in the early stages. The diagnosis of these neurodegenerative motor disorders is essentially clinical. Consequently, the diagnostic accuracy mainly depends on the professional knowledge and experience of the physician. Recent advances in artificial intelligence have made it possible to analyze the large amount of clinical and instrumental information in the medical field. The application machine learning algorithms to the analysis of neuroimaging data appear to be a promising tool for identifying microstructural alterations related to the pathological process in order to explain the onset of symptoms and the spread of the neurodegenerative process. In this context, the search for quantitative biomarkers capable of identifying parkinsonian patients in the prodromal phases of the disease, of correctly distinguishing them from atypical parkinsonisms and of predicting clinical evolution and response to therapy represent the main goal of most current clinical research studies. Our aim was to review the recent literature and describe the current knowledge about the contribution given by machine learning applications to research and clinical management of parkinsonian syndromes.

    Citation: Annalisa Vitale, Rossella Villa, Lorenzo Ugga, Valeria Romeo, Arnaldo Stanzione, Renato Cuocolo. Artificial intelligence applied to neuroimaging data in Parkinsonian syndromes: Actuality and expectations[J]. Mathematical Biosciences and Engineering, 2021, 18(2): 1753-1773. doi: 10.3934/mbe.2021091

    Related Papers:

    [1] Jiyun Shen, Yiyi Xia, Yiming Lu, Weizhong Lu, Meiling Qian, Hongjie Wu, Qiming Fu, Jing Chen . Identification of membrane protein types via deep residual hypergraph neural network. Mathematical Biosciences and Engineering, 2023, 20(11): 20188-20212. doi: 10.3934/mbe.2023894
    [2] Zhengtao Xi, Tongqiang Liu, Haifeng Shi, Zhuqing Jiao . Hypergraph representation of multimodal brain networks for patients with end-stage renal disease associated with mild cognitive impairment. Mathematical Biosciences and Engineering, 2023, 20(2): 1882-1902. doi: 10.3934/mbe.2023086
    [3] Babak Khorsand, Abdorreza Savadi, Javad Zahiri, Mahmoud Naghibzadeh . Alpha influenza virus infiltration prediction using virus-human protein-protein interaction network. Mathematical Biosciences and Engineering, 2020, 17(4): 3109-3129. doi: 10.3934/mbe.2020176
    [4] Wenjun Xu, Zihao Zhao, Hongwei Zhang, Minglei Hu, Ning Yang, Hui Wang, Chao Wang, Jun Jiao, Lichuan Gu . Deep neural learning based protein function prediction. Mathematical Biosciences and Engineering, 2022, 19(3): 2471-2488. doi: 10.3934/mbe.2022114
    [5] Peter Hinow, Edward A. Rietman, Sara Ibrahim Omar, Jack A. Tuszyński . Algebraic and topological indices of molecular pathway networks in human cancers. Mathematical Biosciences and Engineering, 2015, 12(6): 1289-1302. doi: 10.3934/mbe.2015.12.1289
    [6] Yongyin Han, Maolin Liu, Zhixiao Wang . Key protein identification by integrating protein complex information and multi-biological features. Mathematical Biosciences and Engineering, 2023, 20(10): 18191-18206. doi: 10.3934/mbe.2023808
    [7] Jinmiao Song, Shengwei Tian, Long Yu, Qimeng Yang, Qiguo Dai, Yuanxu Wang, Weidong Wu, Xiaodong Duan . RLF-LPI: An ensemble learning framework using sequence information for predicting lncRNA-protein interaction based on AE-ResLSTM and fuzzy decision. Mathematical Biosciences and Engineering, 2022, 19(5): 4749-4764. doi: 10.3934/mbe.2022222
    [8] Zhihong Zhang, Yingchun Luo, Meiping Jiang, Dongjie Wu, Wang Zhang, Wei Yan, Bihai Zhao . An efficient strategy for identifying essential proteins based on homology, subcellular location and protein-protein interaction information. Mathematical Biosciences and Engineering, 2022, 19(6): 6331-6343. doi: 10.3934/mbe.2022296
    [9] Linlu Song, Shangbo Ning, Jinxuan Hou, Yunjie Zhao . Performance of protein-ligand docking with CDK4/6 inhibitors: a case study. Mathematical Biosciences and Engineering, 2021, 18(1): 456-470. doi: 10.3934/mbe.2021025
    [10] Haipeng Zhao, Baozhong Zhu, Tengsheng Jiang, Zhiming Cui, Hongjie Wu . Identification of DNA-protein binding residues through integration of Transformer encoder and Bi-directional Long Short-Term Memory. Mathematical Biosciences and Engineering, 2024, 21(1): 170-185. doi: 10.3934/mbe.2024008
  • Idiopathic Parkinson's Disease (iPD) is a common motor neurodegenerative disorder. It affects more frequently the elderly population, causing a significant emotional burden both for the patient and caregivers, due to the disease-related onset of motor and cognitive disabilities. iPD's clinical hallmark is the onset of cardinal motor symptoms such as bradykinesia, rest tremor, rigidity, and postural instability. However, these symptoms appear when the neurodegenerative process is already in an advanced stage. Furthermore, the greatest challenge is to distinguish iPD from other similar neurodegenerative disorders, "atypical parkinsonisms", such as Multisystem Atrophy, Progressive Supranuclear Palsy and Cortical Basal Degeneration, since they share many phenotypic manifestations, especially in the early stages. The diagnosis of these neurodegenerative motor disorders is essentially clinical. Consequently, the diagnostic accuracy mainly depends on the professional knowledge and experience of the physician. Recent advances in artificial intelligence have made it possible to analyze the large amount of clinical and instrumental information in the medical field. The application machine learning algorithms to the analysis of neuroimaging data appear to be a promising tool for identifying microstructural alterations related to the pathological process in order to explain the onset of symptoms and the spread of the neurodegenerative process. In this context, the search for quantitative biomarkers capable of identifying parkinsonian patients in the prodromal phases of the disease, of correctly distinguishing them from atypical parkinsonisms and of predicting clinical evolution and response to therapy represent the main goal of most current clinical research studies. Our aim was to review the recent literature and describe the current knowledge about the contribution given by machine learning applications to research and clinical management of parkinsonian syndromes.



    Gradual to catastrophic subsidence associated with karst sinkholes can result in loss of human life and property damage. Ground deformation associated with sinkholes frequently damages infrastructure such as homes, buildings, roadways and utilities. For example, the catastrophic-collapse sinkhole that developed in Nixa, Missouri, on August 13, 2006, swallowed a car, the garage it was parked in, and part of the adjoining house [1]. Sinkholes are also frequently associated with other hazardous processes and problems, such as the flooding of associated depressions by runoff and locally elevated water table levels [2], water leakage at dams and other hydraulic structures [3], and groundwater pollution [1,2,3]. Therefore, the investigation of the subsurface structure and developmental mechanisms of a sinkhole can enable assessors to predict the potential impact of a sinkhole and to develop applicable mitigation measures [4].

    The imaging of the subsurface structure of a sinkhole using geophysical techniques is frequently cost-effective and relatively reliable because there is often significant contrast between the physical properties of sinkhole fill, which consists of water, air and/or soil, and the adjacent strata. Geophysical methods that are commonly used for sinkhole investigation include seismic refraction [5], gravimetry [6], ground-penetrating radar [7,8,9,10], electrical resistivity tomography [11,12,13,14,15,16,17,18,19], and multichannel analysis of surface waves [20,21]. Electrical resistivity tomography (ERT) is commonly used in the state of Missouri to investigate the shallow subsurface (depths < 60 m) in karst terrain because the subsurface karst features are generally characterized by high resistivity value contrasts [22].

    The ERT tool can be used to generate 2D or 3D images depicting the distribution of electrical resistivity values in the subsurface. 2D ERT imaging has been proven to be a suitable technique to map and characterize sinkholes in karst terrain. However, in some situations, 2D ERT images are less accurate than desired because 2D ERT processing software cannot adequately compensate for the lateral variations in resistivity that occur outside of the vertical plane of the 2D ERT profile. True three-dimensional (3D) ERT resistivity data, which compensate for lateral resistivity changes that occur outside the vertical plane of the 2D traverses, can be acquired. This approach involves the placement of current electrodes on the nodes of a rectangular grid and the measuring of all the possible potentials. This approach is the most reliable, but it tends to be costly and time-consuming [23,24]. A cost-effective alternative to true 3D ERT imaging is the pseudo-3D ERT imaging technique. This involves the acquisition of parallel and/or orthogonal surface 2D ERT profiles and the processing of the same as though they were true 3D data [25,26]. This method has been studied in detail by many authors [27,28]. Their studies have demonstrated that 3D inversion of a relatively dense grid of 2D ERT data, while certainly a surrogate to a full 3D survey, in many cases is optimal given financial, instrumentation and logistic limitations. Therefore, use of the pseudo-3D resistivity imaging approach is currently widespread in geophysical practice [29,30,31,32].

    Conventional 2D electrical resistivity tomography data were acquired along 16 parallel traverses spaced at 6.1 m (20 ft) intervals across a karst sinkhole site (Figure 1) in Greene County Missouri [4]. The interpretation of these 2D ERT data, constrained by visual inspection, core hole and surface wave control, demonstrate that 2D ERT data can be used to effectively image the subsurface structure of the sinkhole despite the limitation that 2D ERT processing cannot compensate for lateral changes in resistivity that occur outside of the vertical plane of the 2D ERT profile. The results of the 2D ERT investigation indicated the sinkhole developed along a linear, vertical, N-S oriented prominent solution-widened joint and other additional less prominent but seemingly linear low resistivity anomalies.

    Figure 1.  Location and geological maps of the study area.

    In an effort to better elucidate the subsurface structure of the sinkhole and the nature of the linear low resistivity anomalies, the ERT data set was reprocessed using the pseudo-3D ERT imaging technique. The interpretation of the pseudo-3D ERT data set enabled the researchers to more reliably characterize the 3D subsurface structure of the sinkhole and to better elucidate the formation and development mechanisms of the sinkhole.

    The study area is located in Greene County (southwestern Missouri) on the western side of the Ozark Uplift. The rock layers regionally dip gently towards the west with minor faulting and folding. Bedrock in the study area is the Mississippian age Burlington-Keokuk Limestone (Figure 1). This formation is characterized by layers of limestone interbedded with thin layers of chert and the presence of chert nodules within the limestone.

    Uneven dissolution of the Burlington-Keokuk Limestone Formation has resulted in a highly irregular bedrock-overburden interface [33] characterized by the prominent bedrock knobs (pinnacles) bounded by deep troughs (grikes) (Figure 2). The thickness of overburden/residuum in Greene County study area varies from few millimeters to about 15 m.

    Figure 2.  Carbonate rock outcrops with knobs and troughs (Schultheis, 2013).

    According to Orndorff [34], the geological structures in southwestern Missouri appear to have controlled the development of karst. The faults in the study area are generally oriented northwest and northeast and the bedrock area is characterized by two nearly orthogonal joint sets that exhibit vertical dipping with general strike orientations of N 20°W and N 60°E [35].

    In addition to the ERT data, multichannel analysis of surface waves (MASW) data were acquired. The 1D shear-wave velocity profiles generated for each MASW data set and core hole control were used to estimate depth to top-of-rock and to constrain the interpretation of the ERT images.

    Sixteen parallel west-east oriented 2D ERT profiles spaced at 6.1 m (20 ft) intervals were acquired across and in proximity to an existing active sinkhole (Figure 3).

    Figure 3.  Orientation and location of acquired 2D ERT data (Yellow lines) and vertical ERT sections extracted from the Pseudo-3D ERT (Blue lines).

    The 2D ERT data were acquired using an AGI R-8 Supersting multi-channel resistivity system and a dipole-dipole array consisting of 168 electrodes. A dipole-dipole array was employed because it is generally considered to provide superior vertical and lateral resolution relative to other electrode arrays, particularly when imaging karst fractures and caves [2,11,36]. The electrodes were spaced at 1.5 m intervals with expectation the subsurface would be imaged to a depth of at least 30 m.

    During data acquisition, current injection and voltage measurements were repeated twice until every electrode has served for current injection. The parameters used during the acquisition were 1000 mA with time of measurement of 1.2 s. The ERT field data were processed and both 2D and pseudo-3D images were generated. The ERT data were initially transformed into two-dimensional resistivity images with the use of Res2DInv software [37]. For the inversion, the L-1 norm (robust) constraint was applied to provide well-contrasted resistivity values. A maximum of 7 iterations was taken in order to achieve a good resolution and low RMS (<10%). Due to the topographic change in the field finite element scheme was employed to discretize the model. The grid is distorted base on the Schwarz-Christoffel transformation method to represent topography.

    Pseudo-3D ERT images were generated using the 16 parallel ERT profiles spaced at 6.1 m, which is 4 times the electrode spacing. The maximum recommended spacing between profiles for a pseudo-3D survey for environmental and engineering investigations at high depth is 4 times the electrode spacing [28]. To generate the pseudo 3D ERT image, the 16 2D ERT data files were combined into a 3D ERT data file and inverted using the RES3DINV program [24]. 16 horizontal depth slices (labelled Layer-1 (L-1) to Layer-16 (L-16)) and 15 west-east oriented vertical sections (labelled: Section-1 (S-1) to Section-15 (S-15)) that show 2D ERT images were generated. Each of the west-east oriented vertical sections represents the average resistivity of a 6.1m (20 ft) wide zone. The alignments of the vertical sections is as follows: S-1 represents the average resistivity of the zone from 0 m to 6.1 m (20 ft); S-2 represents the average resistivity from 6.1 m (20 ft) to 12.2 m (40 ft); ….and; S-15 represents the average resistivity of the zone from 85.3 m (280 ft) to 91.4 m (300 ft) (Figure 3). The horizontal depth slices represent the resistivity of the subsurface from the near surface (Layer-1) to a depth of 29.7 m (Layer-16).

    MASW data were acquired at nine locations (Figure 3) using Seistronix RAS-24, a 24-channel engineering seismograph coupled to 24 geophones of 4.5 Hz at 2.5 ft intervals, and 9 kg (20 lb) sledgehammer as energy source. Each MASW data set was transformed into a 1D shear-wave velocity profile using Surfseis3 software. The National Earthquake Hazard Reduction Program (NEHRP) site classification chart for different geological material, International Building code, 2000, provided a basis for the classification of subsurface materials based on their shear wave velocity values. In this research MASW is used mainly to determine depth to top of bedrock and to complement the interpretation of ERT profiles.

    One core hole in close proximity to the sinkhole, CH-1 in Figure 3, was drilled to facilitate the correlation of the MASW data and ERT profiles to the actual subsurface geology. The drilling was advanced to the bedrock surface using 21.6 cm (8.5 inch) O.D. hollow-stem augers and bedrock was cored using HQ core barrels.

    The interpretation of the MASW data is consistent with NEHRP site classification chart and core hole ground truth. A shear wave velocity of less than 366 m/s (1200 feet/sec) is typically interpreted as the shear wave velocity of soil. A shear-wave velocity greater than 366 m/s (1200 feet/sec) is interpreted as rock and this is consistent with the ranges of shear wave velocity values given by other researchers [14,38]. Examples of 1-D shear wave velocity profiles from MASW 5 & 7 are presented in Figure 4.

    Figure 4.  1-D shear wave velocity profile of MASW 5 (left) and MASW 7 (right).

    The interpretation of the pseudo-3D ERT data was constrained by core hole data, MASW data and visual field inspection. Moist clayey soils in Missouri are typically characterized by resistivity of less than 45 Ω‐m; drier, typically near-surface soils can be characterized by higher resistivity. Relatively dry, highly to moderately weathered carbonate rock is typically characterized by resistivity values ranging from 125 Ω‐m to 600 Ω‐m; higher water content and/or the presence of piped clay can significantly reduce the bulk resistivity of fractured carbonate rock. Air-filled voids are typically characterized by very high resistivity values. The resistivity signature of a void on ERT data dependents on the conductivity of the surrounding material and depth/size/shape of the void. Figures 5 and 6 are two (S-2 and S-13) of the fifteen ERT sections, derived from the pseudo-3D ERT image, that show how consistent the correlation is with core hole and MASW data. From the core hole and the MASW data, the depth to top of rock corresponds to the resistivity contour value of 125 Ω‐m outlined in black color in the ERT profiles in Figures 5a and 6a. The low resistivity zones, zone-1 (Figure 5a) and zone-2 (Figure 6a) in the ERT profiles are interpreted as fractured zones with higher water and/or piped clay content. A summary of the correlation results of Figures 5 and 6 is presented in Table 1.

    Figure 5.  (a) Vertical ERT section (S-2), vertical and horizontal axes are in meters (b) 1-D shear wave velocity (ft/s) of MASW-5 tied at 61m mark of S-2, (c) 1-D shear wave velocity (ft/s) of MASW-4 tied at 183 m mark of S-2. Pink colored arrows locate depth to bedrock.
    Figure 6.  (a) Vertical ERT section (S-13), vertical and horizontal axes are in meters (b) 1-D shear wave velocity (ft/s) of MASW-8 tied at 73 m mark of S-13, (c) 1-D shear wave velocity (ft/s) of MASW-7 tied at 183 m mark of S-13. Pink colored arrows locate depth to bedrock.
    Table 1.  Depth to top-of-rock correlation of vertical ERT sections, MASW data and core hole.
    Tie points of core hole and MASW data at ERT sections (S-2 & S-13) Depth to top-of-rock
    Core hole ties at 100 m mark of ERT S-2 Core hole (CH-1) 2.3 m (7.5 ft)
    ERT S-2 2.1 m (6.9 ft)
    MASW-4 ties at 183 m mark of ERT S-2 MASW-4 5.8 m (19 ft)
    ERT S-2 6.1 m (20 ft)
    MASW-5 ties at 61 m mark of ERT S-2 MASW-5 6.7 m (22 ft)
    ERT S-2 5.8 m (19 ft)
    MASW-7 ties at 183 m mark of ERT S-13 MASW-7 4.3 m (14 ft)
    ERT S-13 4.6 m (15 ft)
    MASW-8 ties at 73 m mark of ERT S-13 MASW-8 4 m (13 ft)
    ERT S-13 3.7 m (10 ft)

     | Show Table
    DownLoad: CSV

    A sequence of 16 horizontal ERT depth slices (layer-1 to layer-16) which cover a total depth of 29.7 m are extracted from the 3D ERT image and presented in Figure 7a, b. The upper four depth slices (depth range from 0.0 m to 2.66 m) indicate that the subsurface material in this depth range has mainly low resistivity values (<600 Ω-m) except for some scattered zones of higher resistivity (>600 Ω-m). The slices from layer-7 (4.67–5.9 m) to layer-16 (25.4–29.7 m) show mainly higher resistivity (>600 Ω-m) zones (rock) and some linear low resistivity anomalies which are most likely moist clay-rich vertical joint sets.

    Figure 7.  (a) A sequence of sixteen depth slices and location of surface expression of the sinkhole (black circle on layer-1). (b) Sixteen depth slices with the location of the surface expression of the sinkhole and its vertical extrapolation to depth, and interpreted solution widened joint sets labeled as JS-1, JS-2, and JS-3. Layer numbers and corresponding depth ranges are given on the right side of the slices.

    Depth slices from layer-5 to layer-12 depict two linear prominently low resistivity anomalies trending N 60°E and W-E (Figure 7a, b, Figure 8). Depth slices from layer-12 to layer-16 also show two prominently low resistivity anomalies oriented N 60°E and S-N (Figure 7a, b, Figure 9), but the W-E oriented anomaly observed in layer-5 to layer-12 is less visible in the deeper slices, layer-12 to layer-16. The visibility of the linear low resistivity anomalies varies both laterally and vertically. For instance, the W-E oriented anomaly is not clearly visible on the deepest slices (layer 12 to layer 16), and the S-N oriented linear anomaly is more visible with depth. The most plausible explanation is that joint width and/or the volume of moisture and piped clay varies.

    Figure 8.  Two linear prominently low resistivity anomalies oriented NE and W-E, in depth slice Layer-10 (8.95–10.8 m).
    Figure 9.  Two linear prominently low resistivity anomalies oriented NE and S-N, in depth slice Layer-14 (18.3–21.6 m).

    Based on the interpretation of the horizontal ERT depth slices (Figure 7a), three solution-widened joint sets (Figure 7b) have been identified. These three joints sets are labeled as follows: N 60°E trending joint set (JS-1), E-W trending joint set (JS-2), and N-S trending joint set (JS-3) (Figure 7b). Furthermore, an overlay of the surface expression of the sinkhole and the 3D ERT depth slices (Figure 7b) shows that the sinkhole developed at the intersection of the three imaged joint sets.

    As can be seen in both the 2D ERT profile in Figure 10a and the horizontal ERT depth slices, layer-11 to layer-16, in Figure 10b, the low resistivity anomalies representing the joint sets often tend to be wider and have much lower resistivity with depth (at least to a depth of 30 m). This typical characteristic is most likely due to one or combination of the following phenomena; (1) higher moisture concentrations at depth, (2) the widening of vertical seepage pathways through fractured rocks, and (3) more extensive solution-widening of fractures at depth and increased concentrations of piped clay.

    Figure 10.  (a) 2D ERT profile (b) Layers of horizontal ERT depth slices showing the increase in width and the decrease in resistivity value with depth along a vertical low resistivity anomaly.

    Prominent joint sets frequently have surface expressions characterized by elongated depressions which are preferential pathways for surface water flow. This implies that the intersection of the three joint sets is likely a point where water ponded and then infiltrated to the subsurface. Therefore, from the pseudo-3D ERT investigation results and other related studies in the region [4,22], a 3D model depicting the formation and development process of the sinkhole is developed (Figure 11). The formation and development of the sinkhole involved the following stages of processes: (ⅰ) the higher susceptibility for weathering and erosion along solution-widened joint sets resulted in the development of elongated depressions and saddles as a surface expression of the joint sets which in turn makes the joint sets a preferential pathway for surface water flow, (ⅱ) ponding of water at the intersection of the three joint sets and followed by a subsequent infiltration and percolation of water, (ⅲ) piping of clay/fine-grained soils and associated subsidence and minor collapse of residual soils. A previous study [4] has indicated that the sinkhole is a subsidence sinkhole that developed overtime gradually. The soil piping and associated subsidence is a slow and gradual process that continues unless the source of water draining to the sinkhole is retained or blocked. Thus, this kind of sinkhole development can be mitigated using appropriate engineering technologies if the source of piping waters is shut-off.

    Figure 11.  3D model depicting the formation and development process of the sinkhole. JS-1, JS-2 and JS-3 represent joint sets identified from ERT interpretation. The vertical low resistivity zone at the intersection of the three joint sets is interpreted as a zone of preferential pathway for seepage and soil piping which results in the development the sinkhole.

    The pseudo-3D ERT method was used to image the subsurface structure of an active sinkhole in Greene County Missouri. A set of 16 west-east oriented 2D ERT profiles were acquired as an input for the pseudo-3D ERT inversion.

    The results from the pseudo-3D ERT inversion showed more reliable images of the subsurface structure of the sinkhole than the 2D ERT images. Based on the interpretation of the 3D ERT data, it is concluded that the sinkhole developed at the intersection of three vertical solution-widened joint sets.

    The interpreted increase in clay and moisture content with depth (at least to a depth of 30 m) along the linear joint sets implies increased joint width, increased moisture content and/or increased clay content. This conclusion is in agreement with the previous study [4] which stated, the surface sinkhole formed mainly due to soil piping process which resulted in gradual ground subsidence and some associated soil collapse. The soil piping and associated subsidence is a slow and gradual process that continues unless the source of water draining to the sinkhole is retained or blocked. Thus, this kind of sinkhole development can be mitigated or halted using appropriate engineering technologies if the source of piping waters is curtailed.

    The authors declare no conflict of interest.



    [1] A. H. Schapira, P. Jenner, Etiology and pathogenesis of Parkinson's disease, Mov.Disord., 26 (2011), 1049–1055. doi: 10.1002/mds.23732
    [2] W. Dauer, S. Przedborski, Parkinson's disease: Mechanisms and models, Neuron, 39 (2003), 889–909. doi: 10.1016/S0896-6273(03)00568-3
    [3] R. F. Pfeiffer, Non-motor symptoms in Parkinson's disease, Parkinson. Relat. Disord., 22 (2016), 119–122.
    [4] A. J. Hughes, S. E. Daniel, L. Kilford, A. J. Lees, Accuracy of clinical diagnosis of idiopathic Parkinson's disease: A clinico-pathological study of 100 cases, J. Neurol. Neurosurg. Psych., 55 (1992), 181–184. doi: 10.1136/jnnp.55.3.181
    [5] R. B. Postuma, D. Berg, M. Stern, W. Poewe, C. W. Olanow, W. Oertel, et al., MDS clinical diagnostic criteria for Parkinson's disease, Mov. Disord., 30 (2015), 1591–1601. doi: 10.1002/mds.26424
    [6] E. Tolosa, G. Wenning, W. Poewe, The diagnosis of Parkinson's disease, Lancet Neurol., 5 (2006), 75–86.
    [7] E. Hustad, A. H. Skogholt, K. Hveem, J. O. Aasly, The accuracy of the clinical diagnosis of Parkinson disease, The HUNT study, J. Neurol., 265 (2018), 2120–2124.
    [8] J. Levin, A. Kurz, T. Arzberger, A. Giese, G. U. Höglinger, The differential diagnosis and treatment of atypical Parkinsonism, Dtsch. Arztebl. Int., 113 (2016), 61–69.
    [9] A. M. Keener, Y. M. Bordelon, Parkinsonism, Semin. Neurol., 36 (2016), 330–334.
    [10] U. Saeed, A. E. Lang, M. Masellis, Neuroimaging advances in Parkinson's disease and atypical Parkinsonian syndromes, Front. Neurol., 11 (2020), 572976.
    [11] S. Mangesius, S. Mariotto, S. Ferrari, S. Jr Pereverzyev, H. Lerchner, L. Haider, et al., Novel decision algorithm to discriminate parkinsonism with combined blood and imaging biomarkers, Parkinson. Relat. Disord., 77 (2020), 57–63. doi: 10.1016/j.parkreldis.2020.05.033
    [12] A. Kaipainen, O. Jääskeläinen, Y. Liu, F. Haapalinna, N. Nykänen, R. Vanninen, et al., Cerebrospinal fluid and MRI biomarkers in neurodegenerative diseases: A retrospective memory clinic-based study, J. Alzheimers Dis., 75 (2020), 751–765. doi: 10.3233/JAD-200175
    [13] N. He, J. Langley, D. E. Huddleston, S. Chen, P. Huang, H. Ling, et al., Increased iron-deposition in lateral-ventral substantia nigra pars compacta: A promising neuroimaging marker for Parkinson's disease, Neuroimage Clin., 28 (2020), 102391.
    [14] H. Sjöström, T. Granberg, F. Hashim, E. Westman, P. Svenningsson, Automated brainstem volumetry can aid in the diagnostics of parkinsonian disorders, Parkinson. Relat. Disord., 79 (2020), 18–25. doi: 10.1016/j.parkreldis.2020.08.004
    [15] X. Liu, N. Wang, C. Chen, P. Y. Wu, S. Piao, D. Geng, et al., Swallow tail sign on susceptibility map-weighted imaging (SMWI) for disease diagnosing and severity evaluating in parkinsonism, Acta Radiol., (2020), 284185120920793.
    [16] M. Picillo, M. F. Tepedino, F. Abate, R. Erro, S. Ponticorvo, S. Tartaglione, et al., Midbrain MRI assessments in progressive supranuclear palsy subtypes, J. Neurol. Neurosurg. Psych., 9 (2020), 98–103.
    [17] A. Quattrone, A. Antonini, D. E. Vaillancourt, K. Seppi, R. Ceravolo, A. P. Strafella, et al., A New MRI measure to early differentiate progressive supranuclear palsy from De novo Parkinson's dDisease in clinical practice: An international study, Mov. Disord., (2020).
    [18] G. Arribarat, P. Péran, Quantitative MRI markers in Parkinson's disease and parkinsonian syndromes, Curr. Opin. Neurol., 33 (2020), 222–229.
    [19] N. Pyatigorskaya, L. Yahia-Cherif, R. Gaurav, C. Ewenczyk, C. Gallea, R. Valabregue, et al., Multimodal magnetic resonance imaging quantification of brain changes in progressive supranuclear palsy, Mov. Disord., 35 (2020), 161–170. doi: 10.1002/mds.27877
    [20] B. Xiao, N. He, Q. Wang, Z. Cheng, Y. Jiao, E. M. Haacke, et al., Quantitative susceptibility mapping based hybrid feature extraction for diagnosis of Parkinson's disease, Neuro. Clin., 24 (2019), 102070.
    [21] L. Chougar, N. Pyatigorskaya, B. Degos, D. Grabli, S. Lehéricy, The role of magnetic resonance imaging for the diagnosis of atypical parkinsonism, Front. Neurol., 11 (2020), 665.
    [22] A. Quattrone, A. Sarica, D. La Torre, M. Morelli, B. Vescio, S. Nigro, et al., Magnetic resonance imaging biomarkers distinguish normal pressure hydrocephalus from progressive supranuclear palsy, Mov. Disord., 35(2020), 1406–1415. doi: 10.1002/mds.28087
    [23] D. B. Archer, T. Mitchell, R. G. Burciu, J. Yang, S. Nigro, A. Quattrone, et al., Magnetic resonance Imaging and neurofilament light in the differentiation of Parkinsonism, Mov. Disord., 35 (2020), 1388–1395. doi: 10.1002/mds.28060
    [24] M. Bocchetta, J. E. Iglesias, V. Chelban, E. Jabbari, R. Lamb, L. L. Russell, et al., Automated brainstem segmentation detects differential involvement in atypical parkinsonian syndromes, J. Mov. Disord., 13 (2020), 39–46. doi: 10.14802/jmd.19030
    [25] P. Mahlknecht, A. Hotter, A. Hussl, R. Esterhammer, M. Schocke, K. Seppi, Significance of MRI in diagnosis and differential diagnosis of Parkinson's disease, Neurodegener. Dis., 7 (2010), 300-318. doi: 10.1159/000314495
    [26] R. De Micco, A. Russo, A. Tessitore, Structural MRI in idiopathic Parkinson's disease, Int. Rev. Neurobio.l, 141(2018), 405–438. doi: 10.1016/bs.irn.2018.08.011
    [27] S. B. Kotsiantis, I. D. Zaharakis, P. E. Pintelas, Machine learning: A review of classification and combining techniques, Artif Intell. Rev., 26 (2006), 159–190. doi: 10.1007/s10462-007-9052-3
    [28] J. M. Mateos-Pérez, M. Dadar, M. Lacalle-Aurioles, Y. Iturria-Medina, Y. Zeighami, A. C. Evans, Structural neuroimaging as clinical predictor: A review of machine learning applications, Neuroimage Clin., 20 (2018), 506–522. doi: 10.1016/j.nicl.2018.08.019
    [29] R. Cuocolo, M. Caruso, T. Perillo, L. Ugga, M. Petretta, Machine learning in oncology: A clinical appraisal, Cancer Lett., 481 (2020), 55–62. doi: 10.1016/j.canlet.2020.03.032
    [30] L. Agnello, A. Comelli, E. Ardizzone, S. Vitabile, Unsupervised tissue classification of brain MR images for voxel-based morphometry analysis, Int. J. Imag. Syst. Technol., 26 (2016), 136–150. doi: 10.1002/ima.22168
    [31] A. Comelli, A. Stefano, S. Bignardi, C. Coronnello, G. Russo, M. G. Sabini, et al., Tissue classification to support local active delineation of brain tumors, in Medical Image Understanding and Analysis. MIUA (eds. Y.Zheng, B.Williams, K.Chen), Springer, 1065 (2019).
    [32] C. Ricciardi, M. Amboni, C. Santis, G. Ricciardelli, G. Improta, G. Cesarelli, et al., Classifying patients affected by Parkinson's disease into freezers or non-freezers through machine learning, IEEE International Symposium on Medical Measurements and Applications, (2020), 1–6.
    [33] C. Ricciardi, M. Amboni, C. De Santis, G. Ricciardelli, G. Improta, G. D'Addio, et al., Machine learning can detect the presence of Mild cognitive impairment in patients affected by Parkinson's Disease, IEEE International Symposium on Medical Measurements and Applications, (2020), 1–6.
    [34] R. Prashanth, S. Dutta Roy, P. K. Mandal, S. Ghosh, High-accuracy detection of early Parkinson's disease through multimodal features and machine learning, Int. J. Med. Inform., 90 (2016), 13–21. doi: 10.1016/j.ijmedinf.2016.03.001
    [35] Ó. Peña-Nogales, T. M. Ellmore, R. De Luis-García, J. Suescun, M. C. Schiess, L. Giancardo, Longitudinal connectomes as a candidate progression marker for prodromal Parkinson's disease, Front. Neurosci., 12 (2019), 967.
    [36] Y. Liu, Y. Wang, C. Huang, D. Zeng, Estimating personalized diagnostic rules depending on individualized characteristics, Stat. Med., 36 (2017), 1099–1117. doi: 10.1002/sim.7182
    [37] S. Shinde, S. Prasad, Y. Saboo, R. Kaushick, J. Saini, P.K. Pal, et al., Predictive markers for Parkinson's disease using deep neural nets on neuromelanin sensitive MRI, Neuro. Clin., 22 (2019), 101748.
    [38] B. Peng, S. Wang, Z. Zhou, Y. Liu, B. Tong, T. Zhang, et al., A multilevel-ROI-features-based machine learning method for detection of morphometric biomarkers in Parkinson's disease, Neurosci. Lett., 651 (2017), 88–94. doi: 10.1016/j.neulet.2017.04.034
    [39] Y. Chen, G. Zhu, D. Liu, Y. Liu, T. Yuan, X. Zhang, et al., Brain morphological changes in hypokinetic dysarthria of Parkinson's disease and use of machine learning to predict severity, CNS Neurosci. Ther., 26 (2020), 711–719. doi: 10.1111/cns.13304
    [40] E. Glaab, J. P. Trezzi, A. Greuel, C. Jäger, Z. Hodak, A. Drzezga, et al., Integrative analysis of blood metabolomics and PET brain neuroimaging data for Parkinson's disease, Neurobiol. Dis., 124 (2019), 555–562. doi: 10.1016/j.nbd.2019.01.003
    [41] K. L. Poston, S. York Williams, K. Zhang, W. Cai, D. Everling, F. M. Tayim, et al., Compensatory neural mechanisms in cognitively unimpaired Parkinson disease, Ann. Neurol., 79 (2016), 448–463. doi: 10.1002/ana.24585
    [42] C. Salvatore, A. Cerasa, I. Castiglioni, F. Gallivanone, A. Augimeri, M. Lopez, et al., Machine learning on brain MRI data for differential diagnosis of Parkinson's disease and Progressive Supranuclear Palsy, J. Neurosci. Methods, 222 (2014), 230–237. doi: 10.1016/j.jneumeth.2013.11.016
    [43] M. M. Correia, T. Rittman, C. L. Barnes, I. T. Coyle-Gilchrist, B. Ghosh, L. E. Hughes, et al., Towards accurate and unbiased imaging-based differentiation of Parkinson's disease, progressive supranuclear palsy and corticobasal syndrome, Brain Commun., 2 (2020), fcaa051.
    [44] D. B. Archer, J. T. Bricker, W. T. Chu, R. G. Burciu, J. L. Mccracken, S. Lai, et al., Development and validation of the automated imaging differentiation in Parkinsonism (AID-P): A multi-site machine learning study, Lancet Digit. Health, 1 (2019), e222–e231.
    [45] G. Du, M. M. Lewis, S. Kanekar, N. W. Sterling, L. He, L. Kong, et al., Combined diffusion tensor imaging and apparent transverse relaxation rate differentiate Parkinson disease and atypical Parkinsonism, AJNR Am. J. Neuroradiol., 38 (2017), 966–972. doi: 10.3174/ajnr.A5136
    [46] N. Amoroso, M. La Rocca, A. Monaco, R. Bellotti, S. Tangaro, Complex networks reveal early MRI markers of Parkinson's disease, Med. Image Anal., 48 (2018), 12–24. doi: 10.1016/j.media.2018.05.004
    [47] Q. Gu, H. Zhang, M. Xuan, W. Luo, P. Huang, S. Xia, et al., Automatic classification on multi-modal MRI data for diagnosis of the postural instability and gait difficulty subtype of Parkinson's disease, J. Parkinsons Dis., 6(2016), 545–556. doi: 10.3233/JPD-150729
    [48] M. R. Salmanpour, M. Shamsaei, A. Saberi, I. S. Klyuzhin, J. Tang, V. Sossi, et al., Machine learning methods for optimal prediction of motor outcome in Parkinson's disease, Phys. Med., 69 (2020), 233–240. doi: 10.1016/j.ejmp.2019.12.022
    [49] S. Waninger, C. Berka, M. Stevanovic Karic, S. Korszen, P. D. Mozley, C. Henchcliffe, et al., Neurophysiological biomarkers of Parkinson's disease, J. Parkinsons Dis., 10 (2020), 471–480. doi: 10.3233/JPD-191844
    [50] J. Tang, B. Yang, M. P. Adams, N. N. Shenkov, I. S. Klyuzhin, S. Fotouhi, et al., Artificial neural network-based prediction of outcome in Parkinson's disease patients using DaTscan SPECT imaging features, Mol. Imag. Biol., 21 (2019), 1165–1173. doi: 10.1007/s11307-019-01334-5
    [51] Y. L. Chen, X. A. Zhao, S. H. Ng, C. S. Lu, Y. C. Lin, J. S. Cheng, et al., Prediction of the clinical severity of progressive supranuclear palsy by diffusion tensor imaging, J. Clin. Med., 9 (2019), 40.
    [52] Y. Chen, G. Zhu, D. Liu, Y. Liu, T. Yuan, X. Zhang, et al., The morphology of thalamic subnuclei in Parkinson's disease and the effects of machine learning on disease diagnosis and clinical evaluation, J. Neurol. Sci., 411(2020), 116721.
    [53] T. Chih-Chien, L. Yu-Chun, N. Shu-Hang, C. Yao-Liang, C. Jur-Shan, L. Chin-Song, et al., A method for the prediction of clinical outcome using diffusion magnetic resonance imaging: Application on Parkinson's disease, J. Clin. Med., 9 (2020), 647.
    [54] M. Peralta, J. S. H. Baxter, A. R. Khan, C. Haegelen, P. Jannin, Striatal shape alteration as a staging biomarker for Parkinson's Disease, Neuroimage Clin., 27 (2020), 102272.
    [55] H. Kaka, E. Zhang, N. Khan, Artificial intelligence and deep learning in neuroradiology: Exploring the new frontier, Can. Assoc. Radiol. J., (2020), 846537120954293.
    [56] A. W. Olthof, P. M. A. van Ooijen, M. H. Rezazade Mehrizi, Promises of artificial intelligence in neuroradiology: A systematic technographic review, Neuroradiology, 62 (2020), 1265–1278. doi: 10.1007/s00234-020-02424-w
    [57] T. C. Booth, M. Williams, A. Luis, J. Cardoso, K. Ashkan, H. Shuaib, Machine learning and glioma imaging biomarkers, Clin. Radiol., 75 (2020), 20–32. doi: 10.1016/j.crad.2019.07.001
    [58] H. Zhou, R. Hu, O. Tang, C. Hu, L. Tang, K. Chang, et al., Automatic machine learning to differentiate pediatric posterior fossa tumors on routine MR imaging, AJNR Am. J. Neuroradiol., 41 (2020), 1279–1285. doi: 10.3174/ajnr.A6621
    [59] S. J. Cho, L. Sunwoo, S. H. Baik, Y. J. Bae, B. S. Choi, J. H. Kim, Brain metastasis detection using machine learning: A systematic review and meta-analysis, Neuro. Oncol., (2020), noaa232.
    [60] H. M. R. Afzal, S. Luo, S. Ramadan, J. Lechner-Scott, The emerging role of artificial intelligence in multiple sclerosis imaging, Mult. Scler., (2020), 1352458520966298.
    [61] P. N. E. Young, M. Estarellas, E. Coomans, M. Srikrishna, H. Beaumont, A. Maass, et al., Imaging biomarkers in neurodegeneration: Current and future practices, Alzheimers Res. Ther., 12 (2020), 49.
    [62] A. N. Nielsen, D. M. Barch, S. E. Petersen, B. L. Schlaggar, D. J. Greene, Machine learning with neuroimaging: Evaluating its applications in psychiatry, Biol. Psych. Cogn. Neurosci. Neuroimag., 5 (2020), 791–798.
    [63] N. M. Murray, M. Unberath, G. D. Hager, F. K. Hui, Artificial intelligence to diagnose ischemic stroke and identify large vessel occlusions: A systematic review, J. Neuroint. Surg., 12 (2020), 156–164. doi: 10.1136/neurintsurg-2019-015135
    [64] S. Heinzel, D. Berg, T. Gasser, H. Chen, C. Yao, R. B. Postuma, MDS task force on the definition of Parkinson's disease. Update of the MDS research criteria for prodromal Parkinson's disease, Mov. Disord., 34 (2019), 1464–1470. doi: 10.1002/mds.27802
    [65] R. B. Postuma, D. Berg, Prodromal Parkinson's disease: The decade past, the decade to come, Mov. Disord., 34 (2019), 665–675.
    [66] D. Salat, A. J. Noyce, A. Schrag, E. Tolosa, Challenges of modifying disease progression in prediagnostic Parkinson's disease, Lancet Neurol., 15(2016), 637–648. doi: 10.1016/S1474-4422(16)00060-0
    [67] R. B. Postuma, D. Berg, Advances in markers of prodromal Parkinson disease, Nat. Rev. Neurol., 12 (2016), 622–634.
    [68] K. Marek, D. Jennings, S. Lasch, A. Siderowf, C. Tanner, T. Simuni, et al., Parkinson Progression Marker Initiative. The Parkinson Progression Marker Initiative (PPMI), Prog. Neurobiol., 95 (2011), 629–635. doi: 10.1016/j.pneurobio.2011.09.005
    [69] D. Frosini, M. Cosottini, D. Volterrani, R. Ceravolo, Neuroimaging in Parkinson's disease: Focus on substantia nigra and nigro-striatal projection, Curr. Opin. Neurol., 30 (2017), 416–426. doi: 10.1097/WCO.0000000000000463
    [70] G. L. Hedlund, A. G. Osborn, K. L. Salzman, Osborn's brain imaging, pathology and anatomy, Second edition, Elsevier, Philadelphia, 2018.
    [71] L. Ugga, R. Cuocolo, S. Cocozza, G. Pontillo, A. Elefante, M. Quarantelli, et al., Magnetic resonance parkinsonism indices and interpeduncular angle in idiopathic normal pressure hydrocephalus and progressive supranuclear palsy, Neuroradiology, 62 (2020), 1657–1665. doi: 10.1007/s00234-020-02500-1
    [72] J. Xu, M. Zhang, Use of magnetic resonance imaging and artificial intelligence in studies of diagnosis of Parkinson's disease, ACS Chem. Neurosci., 10 (2019), 2658–2667. doi: 10.1021/acschemneuro.9b00207
    [73] M. A. Thenganatt, J. Jankovic, Parkinson disease subtypes, JAMA Neurol., 71 (2014), 499–504.
    [74] D. Aleksovski, D. Miljkovic, D. Bravi, A. Antonini, Disease progression in Parkinson subtypes: The PPMI dataset, Neurol. Sci., 39 (2018), 1971–1976. doi: 10.1007/s10072-018-3522-z
  • This article has been cited by:

    1. Haozhe Yin, Kai Wang, Wenjie Zhang, Ying Zhang, Ruijia Wu, Xuemin Lin, Efficient Computation of Hyper-Triangles on Hypergraphs, 2024, 18, 2150-8097, 729, 10.14778/3712221.3712238
  • Reader Comments
  • © 2021 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(5234) PDF downloads(328) Cited by(10)

Figures and Tables

Tables(1)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog