Objective: In this study, we initially tested the reliability and validity of a Kinect-based system developed by our team with a cohort of small sample size.

Background: Postural abnormalities are common disabling complications in patients with Parkinson’s Disease (PD) leading to pain, balance dysfunction and walking difficulties. At present, there is no recognized objective and quantitative evaluation method for postural abnormalities in PD.

Method: Fifteen individuals with PD were enrolled. An Azure Kinect depth camera based system was used to collect the postural images of PD patients during bipedal quiet standing, after which eight quantified coronal and sagittal features (F1-F8) reflecting the severity of postural abnormalities were automatically obtained by algorithms. Two professionally trained doctors (namely doctor1 and doctor2) individually measured F1 to F8 as previously defined with the free NeuroPostureApp© (https://www.neuroimaging.uni-kiel.de/NeuroPostureApp/). The intraclass correlation coefficient (ICC) was used to evaluate the human-human consistency and human-machine consistency of the testing results.

Results: There was moderate to high consistency between the results of doctor1 and doctor2 in F1, F3, F4, F5 and F6. Moreover, there was moderate to high consistency between the results of the system and doctor1 in F1, F2, F4, F5 and F7 and there was moderate to high consistency between the results of the system and doctor2 in F1, F4, and F5. However, the consistency between the results of the system and two doctors in F3 and F8 was very low.

Conclusion: The Kinect-based system for evaluating postural abnormalities in PD was generally reliable and valid and a study with a much larger sample size is needed to further verify this finding.

« Back to 2022 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/preliminary-verification-of-a-kinect-based-system-for-evaluating-postural-abnormalities-in-patients-with-parkinsons-disease/