パーキンソン病における運動野「ハンドノブ」指標と症状の片側優位性のマッピング：幅比較から機械学習まで

Parkinson's disease typically shows unilateral symptom dominance. Even though general neurodegeneration is well-documented, side-specific cortex affection has not been exhaustively studied.

Specifically, we hypothesize that symptom hemidominance in Parkinson's disease might be associated with motor cortex "hand knob" size. To that end, we retrospectively analyzed 47 Parkinson's patients via structured morphometry, that had undergone standardized MRI for later deep brain stimulation surgery. Additionally, 25 control patients imaged with the same technical setup but without Parkinson's disease were investigated.

Main dimensions of the "hand knob" (e.g. width and length) and compound measures (e.g. area and volume) were registered. Data analysis ranged from classical statistics for e.g. width comparison to advanced machine learning algorithms for integrated multivariable computation (e.g. t-SNE, Random Forest, Neural Networks and Naïve Bayes).

Left "hand knob" width appeared smaller in patients with right symptom hemidominance in exploratory univariable analyses (p = 0.026, uncorrected). In the control group this was not the case (p = 0.235). Complementary "hand knob" computations showed similar outcomes. All machine learning models concordantly suggested that integrated multivariable "hand knob" metrics can predict symptom side differences.

Symptom hemidominance and "hand knob" parameters seem to possibly be connected. Specific cortical motor morphology dynamics of the "hand knob" might exist in hemidominant Parkinson's - for which we provide hitherto overlooked, indicative evidence based on a simple yet fundamental imaging inquiry complemented with advanced machine learning. The hypothesis-generating insights presented here shall inspire further research of cortical motor structures in Parkinson's.