Advanced Digital Image Enhancement Techniques for Medical and Environmental Imaging

Abstract

Medical image enhancement is important for the improvement of the quality of the image for the accurate diagnosis and treatment. Typical enhancement techniques such as histogram equalization, filtering, and frequency based such as these often suffer from noise, low contrast and degradation in resolution. In the recent years, there have been substantial improvements of medical image enhancement due to the recent progress in the field on Machine Learning (ML), Deep Learning (DL) and Artificial Intelligence (AI). Noise reduction, contrast enhancement and feature extraction are enabled by ML based approach, such as supervised and unsupervised learning methods. Computing process has been denoised, super resolution, structural detail preservation with the help of deep learning models: Convolutional Neural Networks (CNNs), Autoencoders, and Generative Adversarial Networks (GANs). AI based enhancement has been used on MRI, CT, X-ray, ultrasound, and retinal imaging. The research on image processing faced issues including data scarcity, generalizability problems, computational complexity, and interpretable model. Future work that can be developed includes supervised learning, federated learning, hybrid AI models, and AI aided radiology workflows. This work surveyed what state of the art uses of ML, DL, and AI medical image enhancement and analyses the solutions, datasets, challenges, and future directions of medical image enhancement, particularly those that are AI driven.