Deep Learning-Based Melanoma Detection: A CNN and RNN Hybrid Approach

Deep learning networks have recently performed excellently in medical image analysis tasks, including skin cancer detection [5]. Many works have proposed using a hybrid deep learning approach that combines Convolutional Neural Networks and Recurrent Neural Networks for accurate classification to detect lethal forms of skin cancer melanoma [17]. Our proposed method combines the advantages of CNN in image feature extraction and RNN in dealing with sequential data. To establish a comprehensive CNN-RNN hybrid model, we train on more skin images, and the model can separate [separate] between benign or malignant lesions. CNN captures the low-level image features, and RNN helps to analyze the sequential pattern of these features, thereby enhancing detection accuracy. Experimental results demonstrate that the proposed hybrid approach can achieve better accuracy, sensitivity, and specificity than traditional CNN-only models. Furthermore, the model proposed may also use patient-specific information such as demographics and medical history for better detection performance. The deep learning strategy can allow earlier melanoma diagnosis and help save patient lives.