Abstract:
One of the most powerful strategies for improving skin cancer patient's chances of survival is finding the disease at its earliest possible stage for diagnosis. It is an ...Show MoreMetadata
Abstract:
One of the most powerful strategies for improving skin cancer patient's chances of survival is finding the disease at its earliest possible stage for diagnosis. It is an essential step to find out the type of cancer in order to ensure efficient treatment machine learning and deep learning have seen explosive growth in the past few years, particularly for the detection and categorization of even the deadliest diseases such as cancer. The use of deep convolutional neural networks (DCNN) has also increased significantly due to their revolutionary impact on computer vision and medical image processing. In this work, we are using three different Deep learning models for skin lesions classification. The main goal of this work is to classify skin lesions in RGB images using deep learning methods such as Densely connected Convolutional Neural Networks, Residual Neural Networks, and Convolutional Neural Networks. The existing methods are costly and time-consuming. This has been a problem for many patients who are in need of critical medical attention. But, due to the emergence of deep learning techniques, this problem could be overcome at a better cost than naive methods. This project aims to Classify an image using Neural networks. Image pre-processing is done on an image. After implementing three different deep learning models to classify the skin lesions from the images, we select the most accurate model and use it moving forward. The model classifies the skin lesion type from the test case images and will come to know which type of skin cancer the patient is suffering from. These kinds of use cases are closely related to the training data. Generalizability can be achieved by carefully picking the right data, collecting more of such data, and using it for training.
Published in: 2023 IEEE International Conference on Recent Advances in Systems Science and Engineering (RASSE)
Date of Conference: 08-11 November 2023
Date Added to IEEE Xplore: 25 December 2023
ISBN Information: