Diagnosis of Invasive Ductal Carcinoma using image processing techniques

Invasive Ductal Carcinoma (IDC) is the most common type of breast cancer accounting for almost 22% of all female cancers in the world. Conventionally, IDC is diagnosed by a pathologist after observing changes in the nuclear morphology of cancer cells. In this paper, we propose an algorithm to diagnose IDC by the analysis and processing of cytology images using MATLAB. The algorithm quantifies nuclear morphological parameters like size, the regularity of the nuclear margin and level of chromatin clumping for nuclei present in each cytology smear by using a sequence of image processing steps. After quantification of these parameters, a comparison is made between normal and IDC images using which a threshold is set for these parameters. These thresholds are then used to make a diagnosis for IDC where the algorithm classifies a cytology image as either normal or cancer. The algorithm therefore finds applications in automated screening programs.