Towards a Comprehensive Solution for a Vision-Based Digitized Neurological Examination.

Area of Science:

• Neurology
• Biomedical Engineering
• Computer Vision

Background:

• Healthcare systems face neurologist shortages, increasing the need for remote and digital care solutions.
• Current digital neurological assessment tools are limited in scope and application.
• Digitally recorded neurological exam data is crucial for improving patient care via telehealth and in-person visits.

Purpose of the Study:

• To introduce Digitized Neurological Examination (DNE), an accessible vision-based system for recording and quantifying neurological exams.
• To expand the options for recording neurological exam biomarkers and their clinical applications.
• To enable healthcare providers and individuals at home to capture detailed neurological movement data.

Main Methods:

• Developed a smartphone/tablet application (DNE) for video capture of standardized neurological tests.
• Implemented 2D/3D human-body pose estimation to extract kinematic and spatio-temporal features from recordings.
• Designed a modular software architecture supporting additional test integrations and a web server for data visualization.

Main Results:

• DNE quantifies clinically relevant movement features, enabling objective documentation and tracking of changes over time.
• Evaluated on a dataset of 21 subjects with normal and simulated impaired movements.
• Achieved classification accuracy exceeding 90% for upper-limb tests and 80% for stand-up and walk tests using machine learning models.

Conclusions:

• DNE provides a versatile and accurate method for objective neurological assessment using readily available technology.
• The system supports remote patient monitoring and can help bridge the gap caused by neurologist shortages.
• The quantified movement data from DNE offers valuable insights for clinical decision-making and patient management.