Malignant pleural mesothelioma segmentation from thoracic CT scans

In this paper, a texture-based segmentation method of the Malignant Pleural Mesothelioma from thoracic CT scans is presented. For the texture analysis part, we have used an automatic sampling and a manual sampling to extract statistical features from the MPM texture. For the segmentation stage, the method iterates the whole CT volume and selects pixels satisfying the extracted statistical criteria. The assessment of the proposed method showed an acceptable degree of similarity rate (J=0.73) between the ground truth and the generated MPM volume.