SKIN CANCER DETECTION USING DEEP LEARNING

Computer Science and Engineering

Human skin is the most exposed part of the human body which needs to be prevented from heat, light, dust and direct exposure to UV rays. Skin cancer is one of the dangerous diseases found recent days. The higher damage caused by the skin cancer is mostly due to the late identification of it. Some of the challenges that affects the success of skin cancer detection include small datasets or data scarcity problem, noisy data, imbalanced data, inconsistency in image sizes and resolutions, unavailability of data, reliability of labelled data and imbalance of skin cancer datasets. This content provides a data augmentation technique based on Synthetic Minority Oversampling Technique (SMOTE) to address the class imbalance problem in the given images. Then it is a challenging task to distinguish between malignant and benign skin lesions as they are alike in their physical appearances. This results in more unnecessary biopsies. To tackle this problem, we developed an enhanced image classification model which can act as a preliminary check before moving to a costlier biopsy. The proposed model can recognize 7 distinct types of skin lesions. Analyses have been performed on the HAM10000 dataset. The classification process is based on transfer learning using multiple pre-trained models, combined with class-weighted loss and augmentation of the data. Experimental analysis shows that the modified ResNet50 model is capable of identifying skin lesion images into one of the seven classes of image.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.