BME Frontiers / 2022 / Article / Tab 2

Research Article

Automated Segmentation and Connectivity Analysis for Normal Pressure Hydrocephalus

Table 2

Comparison of Dice scores for various ventricle and gray-white mass segmentation algorithms for CT scans. The scores are reported as . All methods are our implementations created for the purpose of comparison. The marked improvement in subarachnoid segmentation performance is critical for probability maps were generated from the training set for each fold.

MethodVentricleGray-white matterSubarachnoid

3D UNet+Prob. maps
3D UNet
RF+MCV
Random forestN/A
3D MGACN/A
3D MCVN/A