MR Image Evaluation Using Perception Models

Abstract

Magnetic Resonance Imaging (MRI) is a fast developing image modality. It is advantageous to other imaging methods from many sides such as radiation-free, excellent soft tissue contrast, and etc. Human perception model, which includes human visual system properties, can provide quantitative values for the assessment of image quality. In this study, we developed a more advanced and specialized computational human perception model to assess magnetic resonance images. This model is based on the idea to associate the perceptual difference model (PDM) and detection models. A modified PDM channel structure was implemented in the detection model. A detectability value was determined by the detection model. The model can quantitatively evaluate image quality with the improved accuracy. We anticipate that the proposed model can not only use detection of a low contrast lesion as a measure of image quality, but apply to other circumstances where detection is an inadequate measure such as interventional MRI and fast MR imaging.