Intravoxel incoherent motion diffusion-weighted MR imaging: Advancing Medical Image Processing with Machine Learning

<p dir="ltr">This research study aims to enhance the clinical utility of intravoxel incoherent motion diffusion-weighted magnetic resonance images (IVIM DW-MRI) through deep learning techniques. The objectives were to evaluate the effectiveness of a conditional GAN-based autosegmentation model and an ESRGAN-based image enhancement model in improving segmentation accuracy, image quality, and reliability of quantitative perfusion and diffusion metrics. The findings demonstrated that the autosegmentation model achieved high accuracy (up to 80% Dice coefficient), and the image enhancement model significantly improved diagnostic image quality while maintaining consistency in derived metrics. These results support the potential of machine learning to streamline radiological workflows and advance personalized medicine.</p>