Detection of Melanoma using Deep learning
Pratham Satishkumar Saraiya, Ravi Rathod, Pranjal Prakash Sawant, Mehul Arun Phatangare, Priyanka Ghule · 2024
This study presents a comparative analysis between the two approaches for image classification in medical diagnostics, distinguishing between the image of skin lesions into benign or malignant conditions which help to identify if it’s melanoma or not. The first approach involves the development of a Proposed Convolutional Neural Network(CNN) and the second approach uses transfer learning on pre-trained MobileNet V2 augmented with handcrafted features. Through experimentation and evaluation, we observed that the transfer learning approach outperformed the Proposed CNN model in terms of accuracy. This finding shows the efficiency of transfer learning leveraged in the pre-trained model along with handcrafted features improves the performance on medical image classification tasks.