Phase-Dependent Passive Muscle Stretching Modulates B-Mode Ultrasound Classification of Parkinson's Disease.
Authors
Affiliations (3)
Affiliations (3)
- School of Biomedical Engineering, Medical School, Shenzhen University, Shenzhen, Guangdong, China.
- Department of Ultrasound, Beijing Tiantan Hospital, Capital Medical University, Beijing, China.
- School of Biomedical Engineering, Medical School, Shenzhen University, Shenzhen, Guangdong, China. Electronic address: [email protected].
Abstract
Parkinson's disease (PD) is a progressive neurodegenerative disorder characterized by tremor, rigidity and bradykinesia. Although central nervous system pathology has been extensively studied, peripheral musculoskeletal manifestations in PD remain comparatively understudied. In this exploratory study, the characteristics of the gastrocnemius muscle were assessed using B-mode ultrasound in 28 PD patients and 18 age-matched controls. Dynamic B-mode ultrasound videos were acquired during passive stretching induced by ankle rotation using a continuous passive motion device, while shear wave elastography was recorded concurrently to quantify muscle stiffness. The stretching sequence was divided into three phases-initial (low elasticity), transitional (increasing elasticity) and final (high elasticity)-with phase boundaries standardized using the shear wave elastography-derived stiffness progression. For classification, only B-mode images were used as model input, and a VGG19 model was trained separately for each phase to discriminate PD from the controls. Classification performance increased progressively across the phases, with accuracy improving from 52% in the initial phase to 73% in the final phase, indicating that PD-related muscle features become more discriminative under higher tension. PD muscles exhibit phase-dependent differences from healthy controls that are detectable on B-mode ultrasound. These results support the potential of phase-resolved peripheral musculoskeletal imaging during passive stretching as a complementary biomarker for PD-related neurodegenerative processes.