Medical Image Registration Toolkit, And A Novel Metric Combining Spatial And Intensity Information

Medical image registration is vital in many bio-medical applications and tech-nologies. With the fast developing imaging technologies, medical images with variesmodalities have turned up. These imaging technologies help to reveal physical infor-mation from di?erent point of view. The information helps physicians to diagnoseand treat the patients. Thus there is requirement that the physicians can see the dif-ferent information as a whole. Take CT and MRI for example, the bone tissues canbe very clear in CT images while MRI is very suitable for imaging soft tissues. Com-bining the CT and MRI images, physicians can gain information not only from bonetissues but also soft tissues. It makes the diagnosis more precise. Many researchersare devoted into fusing the medical images from di?erent modalities or imaging tech-nologies. Medical image registration is a process to search the best alignment criteriaiteratively. With huge development in imaging technologies, many alignment criteria,i.e. medical image registration metrics are available. But the mainstream ones likemutual information are based on the intensity-scale information without incorporat-ing spatial information, e.g. contours. Many researchers point out that these metricsfail in some applications. A possible solution is to incorporate spatial information withintensity-scaleinformation. Inthispaper, weproposedanovelregistrationmetriccom-bining spatial and intensity-scale information. All the experiments are carried out ona C++-based medical image registration platform which is the other main work of thispaper. The experiments reveal that our metric combining spatial and intensity-scaleinformation receives better accuracy.