Medical Image Quality

About

Summary Stanford researchers have developed a method that produces a high resolution image from low resolution scans using multi-contrast images, multi-lateral filters, and deep learning neural networks. With two or more (low resolution) differing contrast input images, the deep learning framework denoises the image resulting in a high resolution image - a 3 x improvement. It can also generate an image with contrast similar to that of another modality, such as the creation of CT-like images from MRI data. The method is applicable to diagnostic and functional medical imaging when patients cannot tolerate standard imaging protocols, information needs to be acquired and acted upon quickly, and high quality quantitative data is required.   Applications Medical imaging (MRI, CT, PET, etc.) – especially for patients who cannot tolerate long exams or where high quality quantitative data is required.   Advantages Faster scan – high resolution image with low resolution scan time (3 fold improvement) Reduces radiation exposure More accurate – high quality image from low resolution scans Ability to create images with contrast similar to that of another modality – e.g. CT-like images from MRI data