State-of-the-Art Skin Disease Classification Using Ensemble Learning and Advanced Image Processing

The rise in the prevalence of skin diseases increases the demand for accurate and efficient diagnostic tools. Various traditional diagnostic tools face a few major struggles such as time-consuming and prone to error. Therefore, this paper proposes a novel and reliable skin disease classification framework by integrating advanced preprocessing, feature extraction, and ensemble learning approaches to enhance diagnostic accuracy. The proposed method involves extensive data collection from the Skin disease image dataset, Skin Disease Dataset, and 33k skin disease dataset. Then the collected data are preprocessed by applying key approaches such as normalization, resizing, noise reduction, and grayscale conversion to improve the image quality. The Crayfish Optimization Algorithm with a Reverse Wheel Strategy is applied for an effective feature segmentation, where the most relevant features are segmented. Then the feature extraction is performed using the Gray Level Co-occurrence Matrix. For classification, the Meta Ensemble-based Random Cat Gradient Boost model is introduced by combining the merits of multiple classifiers to enhance prediction performance. The experimental findings demonstrate that the proposed model achieves excellent accuracy and precision of 99.78% and 98.51%, respectively, on the Skin Disease Image Dataset.