Computer-vision based method for quantifying rising from chair in Parkinson's disease patients

Morinan, Gareth; Peng, Yuwei; Rupprechter, Samuel; Weil, Rimona S; Leyland, Louise‐Ann; Foltynie, Thomas; Sibley, Krista; Baig, Fahd; Morgante, Francesca; Gilron, Ro’ee; Wilt, Robert; Starr, Philip A.; O’Keeffe, Jonathan

doi:10.1016/j.ibmed.2021.100046

Cited by 17 publications

(22 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In turn, improving assessment reliability yields better patient outcomes. The present study adds to the current body of literature exploring this exciting avenue for improving Parkinson’s disease evaluation and treatment, including work using similar pose estimation and machine learning approaches for extracting features of gait, ataxia, bradykinesia, tremor severity, and various human movement classifiers [ 14 , 17 , 39 – 43 ]. We unprecedently add to this contemporary work by demonstrating initial steps at applying markerless pose estimation and simplistic machine learning algorithms to automatically assess upper limb movement during DBS localization and implantation.…”

Section: Discussionmentioning

confidence: 99%

“…The subjective assessment of Parkinson's disease severity, denoted the Unified Parkinson's Disease Rating Scale (MDS-UPDRS), only exhibits moderate reliability [57,58]; this reliability is even poorer for evaluation of tremor, a hallmark of Parkinson's disease [59]. Multiple studies elucidate the promising nature of objective means like markerless motion tracking and machine learning algorithms at augmenting the reliability of movement disorder symptom evaluation [12,13,[15][16][17][18]. In turn, improving assessment reliability yields better patient outcomes.…”

Section: Discussionmentioning

confidence: 99%

“…This method produces interrater reliability concerns, steep learning curves, and likely misses important information that exists below the threshold of human detection. These challenges may be addressed with motion tracking and artificial intelligence technologies [12][13][14][15][16][17][18]. Contemporary advancements in computational capacity and machine learning offer such improvement towards a more objective, efficient, and automated assessment, especially useful for clinics with fewer resources and knowledgeable clinicians to effectively carry out and manage DBS implantation and adjustment.…”

Section: Introductionmentioning

confidence: 99%

“…Previous studies have employed two-dimensional pose estimation in conjunction with binary classification or principal component analysis to automatically extract features of gait aligning with clinical parameters and bradykinesia indicators to assess Parkinson's disease severity [39][40][41]. Recent groups have demonstrated utility of machine learning and markerless pose estimation approaches for sit-to-stand gait analysis [17], classification of ataxia severity [18], and hand movements [42] diagnosis of a myriad of neurodevelopmental disorders, with promising results. Others have used less-robust decision tree, discriminant analysis, and nearest-neighbor machine learning algorithms to automatically score the severity of tremor in Parkinson's disease patients [14,15,43]; as well as to facilitate DBS adjustment [44].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Automatic extraction of upper-limb kinematic activity using deep learning-based markerless tracking during deep brain stimulation implantation for Parkinson’s disease: A proof of concept study

Baker

Tekriwal²,

Felsen³

et al. 2022

PLoS ONE

View full text Add to dashboard Cite

Optimal placement of deep brain stimulation (DBS) therapy for treating movement disorders routinely relies on intraoperative motor testing for target determination. However, in current practice, motor testing relies on subjective interpretation and correlation of motor and neural information. Recent advances in computer vision could improve assessment accuracy. We describe our application of deep learning-based computer vision to conduct markerless tracking for measuring motor behaviors of patients undergoing DBS surgery for the treatment of Parkinson’s disease. Video recordings were acquired during intraoperative kinematic testing (N = 5 patients), as part of standard of care for accurate implantation of the DBS electrode. Kinematic data were extracted from videos post-hoc using the Python-based computer vision suite DeepLabCut. Both manual and automated (80.00% accuracy) approaches were used to extract kinematic episodes from threshold derived kinematic fluctuations. Active motor epochs were compressed by modeling upper limb deflections with a parabolic fit. A semi-supervised classification model, support vector machine (SVM), trained on the parameters defined by the parabolic fit reliably predicted movement type. Across all cases, tracking was well calibrated (i.e., reprojection pixel errors 0.016–0.041; accuracies >95%). SVM predicted classification demonstrated high accuracy (85.70%) including for two common upper limb movements, arm chain pulls (92.30%) and hand clenches (76.20%), with accuracy validated using a leave-one-out process for each patient. These results demonstrate successful capture and categorization of motor behaviors critical for assessing the optimal brain target for DBS surgery. Conventional motor testing procedures have proven informative and contributory to targeting but have largely remained subjective and inaccessible to non-Western and rural DBS centers with limited resources. This approach could automate the process and improve accuracy for neuro-motor mapping, to improve surgical targeting, optimize DBS therapy, provide accessible avenues for neuro-motor mapping and DBS implantation, and advance our understanding of the function of different brain areas.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automatic extraction of upper-limb kinematic activity using deep learning-based markerless tracking during deep brain stimulation implantation for Parkinson’s disease: A proof of concept study

Baker

Tekriwal²,

Felsen³

et al. 2022

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…Kelvin uses a peak detection algorithm [18] to identify local maxima (peaks) and minima (troughs), which typically correspond to the start and midpoint of a periodic action. For example, as the fingertapping signal was based on the distance between thumb and index finger tip, a peak would correspond to the two fingers being maximally apart, and a trough would correspond to the two fingers touching.…”

Section: Kelvin Analytic Processesmentioning

confidence: 99%

An Evaluation of KELVIN, an Artificial Intelligence Platform, as an Objective Assessment of the MDS UPDRS Part III

Sibley

Girges

Candelario

et al. 2022

JPD

Self Cite

View full text Add to dashboard Cite

Background: Parkinson’s disease severity is typically measured using the Movement Disorder Society Unified Parkinson’s disease rating scale (MDS-UPDRS). While training for this scale exists, users may vary in how they score a patient with the consequence of intra-rater and inter-rater variability. Objective: In this study we explored the consistency of an artificial intelligence platform compared with traditional clinical scoring in the assessment of motor severity in PD. Methods: Twenty-two PD patients underwent simultaneous MDS-UPDRS scoring by two experienced MDS-UPDRS raters and the two sets of accompanying video footage were also scored by an artificial intelligence video analysis platform known as KELVIN. Results: KELVIN was able to produce a summary score for 7 MDS-UPDRS part 3 items with good inter-rater reliability (Intraclass Correlation Coefficient (ICC) 0.80 in the OFF-medication state, ICC 0.73 in the ON-medication state). Clinician scores had exceptionally high levels of inter-rater reliability in both the OFF (0.99) and ON (0.94) medication conditions (possibly reflecting the highly experienced team). There was an ICC of 0.84 in the OFF-medication state and 0.31 in the ON-medication state between the mean Clinician and mean Kelvin scores for the equivalent 7 motor items, possibly due to dyskinesia impacting on the KELVIN scores. Conclusion: We conclude that KELVIN may prove useful in the capture and scoring of multiple items of MDS-UPDRS part 3 with levels of consistency not far short of that achieved by experienced MDS-UPDRS clinical raters, and is worthy of further investigation.

show abstract

On The Feasibility of Multimodal Dialog Based Remote Balance Assessment

Sukhdev,

Roesler,

Neumann

et al. 2024

Studies in Computational Intelligence

View full text Add to dashboard Cite

Computer-vision based method for quantifying rising from chair in Parkinson's disease patients

Cited by 17 publications

References 31 publications

Automatic extraction of upper-limb kinematic activity using deep learning-based markerless tracking during deep brain stimulation implantation for Parkinson’s disease: A proof of concept study

Automatic extraction of upper-limb kinematic activity using deep learning-based markerless tracking during deep brain stimulation implantation for Parkinson’s disease: A proof of concept study

An Evaluation of KELVIN, an Artificial Intelligence Platform, as an Objective Assessment of the MDS UPDRS Part III

On The Feasibility of Multimodal Dialog Based Remote Balance Assessment

Contact Info

Product

Resources

About