Lesion border detection using deep learning

Computer aided diagnosis of medical images can result in (better) detection in addition to early diagnosis of many symptoms to assist health physicians and therefore reducing the mortality rate. Realization of an efficient mobile device for automatic diagnosis of melanoma would greatly enhance the applicability of medical image classification scheme and make it useful in clinical contexts. In this paper, a deep learning method using convolutional neural networks (CNN) is proposed for border detection of skin lesions based on clinical images. Prepossessing of clinical and dermoscopy images has been common and necessary in the lesion segmentation realm; however, the result of the study shows that CNN can be used with relatively much less prepossessing algorithm compared with previous methods.