Recently, imaging techniques such as high magnetic field imaging, multi-detector row CT provide remarkable improvement in diagnosis. However, the latest imaging systems easily produce more than a thousand diagnostic images per patient. Image reading of a huge number of multi-modality images by use of a multi-formatted display and/or stack display method would be a time consuming task for radiologists. Therefore, we have developed a pixel-based comprehensive image processing that can detect liver diseases by use of multi modality images. Radiologist’s criterion for differential diagnosis of liver disease was employed. In this scheme, accurate registration between multi-modality images would be very important because misregistration cause false positive pixels. Therefore, we performed a mutual information technique to match different types of images. Total of 40 cases which included 10 cyst, 10 hemangioma, 10 hepatocelluar carcinoma, and 10 metastatic cancer are used for the development of the comprehensive image processing. We could detect liver diseases in all cases correctly. Image registration technique based on the mutual information would be useful to perform the comprehensive image processing accurately.