Submission Deadline: 22 June 2024 (closed) View: 321
Deep learning has shown a huge interest in computer vision in the last few years, especially for the application of medical imaging. In medical imaging, deep learning is employed for detecting and classifying cancers such as skin cancer, stomach cancer, brain tumor, and a few more. Dermoscopy, wireless capsule endoscopy, and MRI imaging technologies are adopted for these cancers. The manual inspection of these cancers is not an easy process and always requires an expert person; therefore, a computerized method is widely required. A computerized technique employs preprocessing, segmentation, feature extraction, selection, and classification steps. The contrast enhancement using deep learning techniques is useful when a fully automated system is designed. In addition, lesion segmentation is performed using custom CNN models, U-NET, and deep saliency maps. For feature extraction, deep learning based features are extracted. Several pre-trained models, GAN, residual networks, and LSTM models are adopted for this. A deep learning model requires a large amount of data for the training process; therefore, GANs are mostly employed for data augmentation in recent years. Recently, federated learning has shown much success in the classification process in medical imaging. Furthermore, the features fusion and selection process improved the performance of the proposed framework and reduced the testing time.