This project will use a large database of images of skin lesions and attempt to perform automatic recognition of melanomas (cancer). The first step will be to segment the lesions, i.e. locate the region of interest in the image. This will be done by using rules to combine several techniques. Next, the regions are described by their shape , colour and texture to identify whether the lesion is malignant or not.

The project could be based on techniques from the following paper:

H. Ganster, A. Pinz, et al., Automated melanoma recognition., IEEE Trans. on Medical Imaging, 2001.

Also, see my paper: J Yang, X Sun, J Liang, P.L. Rosin, "Clinical Skin Lesion Diagnosis using Representations Inspired by Dermatologist Criteria", Proc. Computer Vision and Pattern Recognition (CVPR), pp. 1258-1266. 2018.

And could use data from these papers:

The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions PAD-UFES-20: A skin lesion dataset composed of patient data and clinical images collected from smartphones