GAN-Based Metaphase Image Enhancement

Abstract

Medical images like metaphase imaging can be effectively visualized thanks to machine learning's success in medicine. First and foremost, before training the data (pictures), any redundant or irrelevant information, such as noise, artefacts, or incorrect data, must be removed. The staining noises, uniformity, and blurring frequently occurring in chromosomal images meaningfully impact the karyotyping procedure. In this study, chromosomal image processing was done using transfer learning on a Generative Adversarial Network (GAN). The automated karyotyping approach produces promising results that are effectively employed for segmentation. Natural Image Quality Evaluator (NIQE) and Perception-based Image Quality Evaluator (PIQE) scores have been used to quantify the performance of the suggested method. On average, 1000 images used for testing and perfect in removing noise, blurriness and contrast have been obtained.