AI-Driven Early Detection of Skin Cancer in the USA: A Hybrid Image Processing and Neural Network Approach

Abstract

Skin cancer remains the most prevalent form of cancer in the United States, with melanoma posing a serious public health concern due to its potential to become life-threatening if not detected early. However, early-stage skin cancer is highly treatable and timely accurate diagnosis is crucial for improving patient survival rates. In response to the growing demand for accessible and rapid diagnostic tools, this study presents an AI-powered skin cancer detection model tailored for application within the U.S. healthcare landscape. The proposed approach integrates Artificial Neural Networks (ANN) with advanced image processing techniques to facilitate early and non-invasive identification of skin cancer. Dermoscopic images are first collected and pre-processed using noise reduction, contrast enhancement, and normalization to improve diagnostic quality. Segmentation is then performed using thresholding methods to highlight regions with abnormal pigmentation or texture. For feature extraction, a 2D Wavelet Transform is employed to capture critical visual cues at multiple resolutions, reflecting the lesion’s structural complexity. These extracted features are input into a Backpropagation Neural Network (BPNN), trained to accurately distinguish between malignant and benign skin lesions. This AI-based diagnostic framework demonstrates high reliability and efficiency, offering valuable clinical support for dermatologists and contributing to timely intervention ultimately supporting the broader objective of improving skin cancer outcomes across the United States.