ADVANCED IMAGE PROCESSING AND DEEP LEARNING FOR SKIN CANCER SCREENING
Keywords:
Skin Cancer Screening, Disease Detection, CNN, HAM10000 dataset, Adam RMSprop, XAI, Image Processing, Deep Learning.Abstract
Due to the limited availability of resources, skin cancer is one of the most quickly spreading diseases
in the globe. Identification of skin cancer through an accurate diagnosis is essential for a preventative approach
in general. Dermatologists struggle to detect skin cancer at an early stage, and in recent years, both supervised
and unsupervised learning tasks have made extensive use of deep learning. One of these models, Convolutional
Neural Networks (CNN), has surpassed all others in object detection and classification tests. The dataset is
screened from MNIST: HAM10000 which consists of seven different types of skin lesions with the sample size of
10015 is used for the experimentation. The data pre-processing techniques like sampling, dull razor and
segmentation using autoencoder and decoder is employed. Utilizing the HAM10000 dataset and an optimized
CNN, the study identify seven types of skin cancer. Two optimization functions (Adam and RMSprop) and three
activation functions (Relu, Swish, and Tanh) were used to train the model. In addition, an XAI-based skin lesion
classification system with Grad-CAM and Grad-CAM++ was developed to explain the model's decisions. This
system can assist physicians in making accurate early skin cancer diagnoses in their early stages.The future
extension of this study includes increasing forecast accuracy through parameter tuning.
