Ensemble Learning for Skin Lesion Classification: A Robust Approach for Improved Diagnostic Accuracy (ELSLC)

• Published in: 2023 3rd International Conference on Innovative Mechanisms for Industry Applications (ICIMIA) pp. 390 - 395
• Main Authors: Bhargavi, Maridu, Renugadevi, R., Sivabalan, S., Phani, Pamulapati, Ganesh, Janga, Bhanu, Konda
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21-12-2023
• Subjects: CNN; Convolutional neural networks; Deep learning; Ham10000; Lesions; Medical diagnosis; Medical diagnostic imaging; Skin; Skin cancer; Skin lesion; Training
• DOI: 10.1109/ICIMIA60377.2023.10425888

In the field of medicine, effective detection of skin cancer through image analysis is difficult. Skin lesion identification is a time-consuming process in medical procedures that may eventually result in patient death. Effective detection of skin cancer is difficult. Consequently, there are not enough skilled dermatologists worldwide to meet the needs of modern medicine. Data imbalance issues arise from stark differences between data from different classifications in the healthcare industry. These learning techniques are usually spent more time training in one class than others because of problems with data imbalance. In the proposed method, an unbalanced dataset was used. The dataset used, Skin Cancer MNIST: HAM10000, contains seven different categories of skin lesions and a total of 10015 images. Our examination is how Deep Learning(DL) algorithms and Convolutional Neural Network (CNN) models were classified. Deep learning models are commonly employed in image-based disease diagnosis. Skin cancer was classified using a deep-learning-based model.