engleski

Multimodality Image Fusion - Techniques and Benefits

Fusion of images from the various diagnostic imaging modalities brings the whole new dimension to the diagnostics. This presentation outlines the fusion or registration problem in general, and covers some of the example applications and their benefits. Fusion/registration can be divided into two broad categories: intra-modality (registering two same modality patient exams for therapy effectiveness evaluation purposes ; intra-operative deformation corrections) and inter-modality (delivering simultaneous correlated display of fine anatomical details and particular physiological function: PET/CT, SPECT/CT, PET/MR). Image fusion is a 3D registration process involving rigid-body transformations: voxel size scaling, translation and rotation (in some situations external fiducial markers are used to help image repositioning). In addition to these, in some situations/application non-rigid-body transformations coping with object deformation and intra-body organ movement are necessary. Algorithms can use image segmentation or be image-intensity-based ; or can use advanced algorithms such as mutual information / entropy maximization approach. Non-rigid algorithms are very computing intensive tasks that require the use of supercomputers or distributed computing environment. Image fusion/registration is also used for correction purposes such as attenuation and scatter correction in SPECT. For such purposes the obstacles are still internal organ motions such as heart beating, respiratory motions, etc. These motions can produce artifacts near object boundaries. However, despite some problems waiting to be solved, image fusion techniques and applications are rapidly evolving and bring some new and exciting insights into medical diagnosis.

image fusion; image registration

nije evidentirano