Multimodal medical image registration is one of the important techniques in medical imaging, which can provide better treatment, diagnosis and planning in the area of radiation therapy, neurosurgery, cardio thoracic surgery and many others. To perform the registration of multimodal images, there are several components that need to be selected such as similarity metric, transformation, optimizer to decide amount of transformation, and mapper/interpolator. There are no certain predefined methods but based on the application and types of images a registration frame work will be designed. Proposed literature encompasses a multilevel registration technique to enhance registration of multimodal medical image sequences such as Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) sequence of the brain. Mutual information metric is measured to assess registration results, and it estimates the probability density using non-parametric and semi-parametric models to analyze the MI trends during the multilevel image registration technique for each dataset.Proposed methodology elucidates how multilevel image registration can enhance the results of multimodal medical image registration such as CT and MRI sequences. The average mutual information analysis recorded at multilevel, if performed to evaluate the registration performed on CT and MRI sequence of the brain, is 0.82. Density curve measured and plotted for MI measured for multilevel shows a narrow trend compared to the conventional single registration technique. In case of mis-registrations, the density curve shows a broader trend.By performing a multilevel registration for multimodal medical images such as CT and MRI, it gives promising results by enhancing the similarity between both images. Analysis of mutual information using density function shows an enhancement in the resultant of multilevel registration technique.
|Number of pages||10|
|Journal||International Journal of Advanced Science and Technology|
|Publication status||Published - 10-04-2020|
All Science Journal Classification (ASJC) codes
- Computer Science(all)