Project

Segmentation and tracking of ovarian tumours from ovarian CT images 

Ovarian tumours are very dangerous for women, and have to be detected early. While MRIs are the definitive way of tumour detection, MRIs are also expensive and rarer, especially in countries like India. Consequently, it is common to use CT images, but detecting tumours and segmenting their boundaries is a much harder job. Further, 

segmentation of tumours is very subjective, even when done by experts, so tracking them in the long term becomes a serious challenge. Ovarian tumours come in all kinds of shapes, sizes and locations, and can be difficult to detect and segment, especially in CT 

images that do not have great amount of detail. Consequently, we aim to detect and segment tumours in the ovarian region using deep learning techniques. Utilising the large amount of data available and high speed computing, it is possible to segment tumours in CT images using deep learning techniques.