How Statistical Methods, Hemispheric Data and Masking Approaches Shape Probabilistic Sweet Spots in Deep Brain...

Area of Science:

• Neuroscience
• Medical Technology
• Computational Biology

Background:

• Probabilistic mapping is crucial for identifying optimal Deep Brain Stimulation (DBS) targets, known as Probabilistic Sweet Spots (PSS).
• The accuracy and reliability of PSS identification are sensitive to various workflow parameters, including statistical methods and data handling.
• Understanding these influences is key to standardizing practices and enhancing the clinical translation of DBS.

Purpose of the Study:

• To investigate how different methodological choices impact the stability and spatial consistency of PSS.
• To evaluate the effect of varying sample sizes on PSS computation using different statistical approaches.
• To assess the influence of hemispheric data handling and masking parameters on PSS outcomes.

Main Methods:

• Analysis of intraoperative stimulation test data from 36 Parkinson's Disease patients.
• Computation of PSS across increasing sample sizes using Bayesian t-test (BAYES), Logistic Regression Model (LRM), Wilcoxon test with FDR correction (WFDR), and Wilcoxon test with permutation correction (WPERM).
• Assessment of PSS stability and consistency based on statistical tests, hemispheric data handling, and masking parameters.

Main Results:

• The Bayesian t-test demonstrated superior robustness, especially with small to intermediate sample sizes.
• WFDR and LRM methods required larger cohorts (∼25-30 patients) for stabilization; WPERM consistently underperformed.
• Hemispheric stability was higher when analyzed separately, indicating potential asymmetries; masking parameters primarily affected PSS volume, not stability patterns.

Conclusions:

• The choice of statistical test significantly influences PSS outcomes, with the Bayesian t-test recommended for smaller cohorts.
• Separate analysis of brain hemispheres is advised to prevent masking clinically relevant asymmetries.
• Standardizing probabilistic mapping through careful consideration of these parameters enhances reliability for clinical translation.