COVID-19 AND SKIN CANCER DETECTION USING A STACKED ENSEMBLE APPROACH

Abstract

In recent years, Covid-19 and skin cancer have become two prevalent ill nesses with severe consequences if untreated. This research represents a significant step toward leveraging machine learning and ensemble techniques to improve the accuracy and efficiency of medical image diagnosis for criti cal diseases like Covid-19 (grayscale images) and skin cancer (RGB images). To enhance the precision and effectiveness of diagnosing both Covid-19 and skin cancer, a stacked ensemble learning approach has been used here. This proposed method combines deep neural network (CNN) pre-trained models for feature extraction, utilizing ResNet101, DenseNet121, and VGG16 for grayscale (COVID-19) and RGB (skin cancer) images. The performance of the model is evaluated using both individual CNNs and a combination of feature vectors generated from ResNet101, DenseNet121, and VGG16 neural network architectures. The feature vectors obtained through transfer learn ing are then fed into base-learner models consisting of five machine-learning algorithms. In the final step, the predictions from the base-learner models, ensemble validation dataset, and the feature vectors extracted from neural networks are ensembled and applied as input for the meta-learner model to obtain final predictions. The performance metrics of the stacked ensemble model reveal high accuracy for Covid-19 diagnosis and intermediate accuracy for skin cancer