Functional Brain Imaging Reliably Predicts Bimanual Motor Skill Performance in a Standardized Surgical Task

Area of Science:

• Neuroscience
• Medical Engineering
• Machine Learning

Background:

• Objective assessment of bi-manual motor skills is crucial for professions like surgery.
• Current assessment methods lack objective metrics, particularly for surgical certification.

Purpose of the Study:

• To predict performance scores in standardized bi-manual motor tasks using functional near-infrared spectroscopy (fNIRS) data.
• To develop and validate a deep-learning framework, 'Brain-NET', for analyzing fNIRS data to assess surgical skills.

Main Methods:

• Utilized fNIRS data collected during a standardized bi-manual motor task.
• Developed a deep-learning framework named 'Brain-NET' for feature extraction from fNIRS signals.
• Validated the model's predictive accuracy using R-squared values and classification performance via ROC curves and AUC.

Main Results:

• The 'Brain-NET' framework accurately predicted bi-manual surgical motor skills from neuroimaging data with an R-squared value of 0.73.
• The model demonstrated strong classification ability with an Area Under the Curve (AUC) of 0.91.
• Functional near-infrared spectroscopy combined with deep learning shows potential for skill assessment.

Conclusions:

• fNIRS coupled with deep learning analysis offers a promising approach for objective bi-manual skill assessment.
• This method enables bedside, rapid, and cost-effective evaluation of motor skills, particularly relevant for surgical training and certification.
• The 'Brain-NET' framework provides a novel neuroimaging-based tool for quantifying surgical proficiency.