ORIGINAL RESEARCH
Evaluation of the Automated GBM Segmentation on the 3D Slicer Medical Image Computing Platform
Meryem Cansu ŞAHİN
Abstract
The aim of this study is to present the volume analysis of the segments obtained manually with the automatic segments created using the GrowCut plugin of the 3D Slicer program. Three physicians segmented 10 patients separately with manual segmentation and GrowCut plugin. The segmentations obtained from both methods were recorded as binary volumes and the agreement between the two was compared in the 3D Slicer program using dice similarity coefficient (DSC) and average hausdorff distance (AHD). Manually segmented volumes of the 10 patients included in the study ranged from 281.234 cc to 1330.57 cc. In addition, volumes ranged from 281.387 cc to 1332.54 cc in GrowCut segmentation. The lowest DSC was 87.133% in the segmentations of all patients, It was observed that the AHD values ranged between 0.00743 and 3.06570 mm. In our study, only brain tumors were examined. Studies on the segmentation of different cancer types will be planned in the 3D Slicer software.
Full text:
Cite as: SAHIN, Meryem Cansu. Evaluation of the Automated GBM Segmentation on the 3D Slicer Medical Image Computing Platform. Gelenbevi Scientific Research Journal, 2021; 1(1):8-13
