Medical Image Colorization Using Deep Learning

Medical Image Colorization Using Deep Learning

Pranav L K , Raghavendra Rathod, Rakesh L, Saquib Pasha S, Dr. Krishna Kumar P.R,

Computer Science and Engineering & S.E.A College of Engineering And technology

ABSTRACT-

Medical imaging techniques like X-rays, CT scans, and MRI produce grayscale images that can lack visual detail and make it challenging to differentiate between various anatomical structures. This project proposes a deep learning-based approach to automatically colorize medical grayscale images, enhancing their visual quality and aiding in improved diagnosis and analysis. By leveraging convolutional neural networks (CNNs) and the Inception-ResNetV2 architecture, the developed model learns complex patterns and features from a large dataset of colored medical images. The colorization process maps the single-channel grayscale input to a three- channel RGB output, producing a vibrant, colorized version of the original image. The proposed methodology aims to achieve accurate color mapping, real-time performance, and strong generalization capabilities, addressing limitations of existing colorization techniques. Quantitative evaluation using metrics like Mean Squared Error (MSE) and Peak Signal-to- Noise Ratio (PSNR) demonstrates the model's effectiveness in producing high-quality colorized outputs. This technology has the potential to enhance medical image interpretation, improve historical preservation of medical records, and create unique artistic representations, contributing to the advancement of healthcare and related fields