Effect of Thresholding Using Groundtruth Pixel Intensities on Lung Image Database Consortium-Image Database Resource Initiative Lung Cancer Dataset

摘要

Thresholding is a primitive and one of the common operations performed on images for segmentation. They are widely being used in processing of medical images too. When intensity based operations are performed on images for segmentation and separation of specific regions of interest, the discriminability of these intensity based operations are to be analyzed. Discriminability refers to the capacity of the operation to isolate through segmentation, the region of interest (ROI) from the rest of the region unambiguously without performing any other region specific operations such as template based or other complex operations. In this paper, we provide some of our efforts made to identify the discriminability of thresholding operation on the Lung Image Database Consortium-Image Database Resource Initiative (LIDC-IDRI) lung cancer dataset.