An Efficient Approach for classification of skin cancer lesions using Inception V3

One of the most common types of cancer, skin cancer, has an increasing number of victims worldwide. Early skin cancer detection is crucial. Medical professionals used computer-based diagnosis systems to help them identify diseases, which allowed for more effective treatment and improved patient survival rates. The classification issue with skin diseases is dealt with in the suggested approach. Malignant and benign tumours are categorized using an automatic, reliable system that is created. A modified pre-trained Deep Convolutional Neural Network (DCNN) is used in this system. The proposed system was developed using a pre-trained model called Inception V3. The last layers of the pre-trained Inception V3 model are changed in accordance with the suggested system requirements. According to binary classification detection, the softmax layer is adjusted and flattened. The suggested system model has been thoroughly trained on a 3297 -image dataset of benign and malignant skin cancer tumours, with 960 photos in each class. After proper training, the suggested system model is tested on 300 images, with 180 images for Benign and 150 for Malignant. The suggested system model is examined using the following metrics: precision, sensitivity, and specificity. The proposed system achieved an accuracy of 99% in training and 97%, 95% in testing (for Benign and Malignant) is higher than that of conventional methods of categorization