Is VLSM a valid tool for determining the functional anatomy of the brain? Usefulness of additional Bayesian network analysis

Abstract Objectives The ability of voxel-based lesion-symptom mapping (VLSM) to define the functional anatomy of the human brain has not been fully assessed. With a view to assessing VLSM's validity, the present study analyzed the technique's ability to determine the known clinical-anatomic correlates of hemiparesis in stroke patients. Design Lesions (damaged in at least 5 patients) associated with transformed limb motor score (after adjustment on lesion volume) at 6 months were examined in 272 patients using VLSM. The value of additional multivariable linear, logistic and Bayesian analyses was examined. Results We first checked that motor hemiparesis was fully accounted for by corticospinal tract (CST) lesions (sensitivity = 100%; p = 0.0001). Conventional VLSM analysis flagged up 2 regions corresponding to the CST, but also 8 regions located outside the CST. All 10 brain regions achieving statistical significance in the VLSM analysis were submitted to 3 additional analyses. The backward linear regression analysis selected 5 regions, one only corresponding to the CST (R 2 : 0.03, p = 0.0008). The logistic regression analysis selected correctly the CST (OR: 2.39, 95%CI: 1.44–3.96; 0.001). The Bayesian network analysis selected regions including the CST (in 92% of 3000 bootstrap replications) and identified the source of multicollinearity. These lesions evaluated by structural equation modeling resulted in an excellent fit (p-value = 0.228, chi/df = 1.19, RMSEA = 0.032, CFI = 0.999). Analyses of confusion factors showed that conventional VLSM analyses were strongly influenced by lesion frequency (R 2 = 0.377; p = 0.0001) and multicollinearity. Conclusions Conventional VLSM analyses are sensitive but weakened by a type I error due to the combined effects of multicollinearity and lesion frequency. We demonstrate that the addition of a Bayesian network analysis, and to a lesser extent of logistic regression, controlled for this type I error and constituted a reliable means of defining the functional anatomy of the motor system in stroke patients.