Computer Vision News - September 2020

Open Source Software 18 On the left, there is a panel which allows to add landmarks and on the right a visualization tool. By clicking on Add you can increase the number of landmarks to fine-tune the transformation. The red button on the right of each landmark allows to go back to that point even after you already set many of them, to change it or to look at the views from a specific angle. On the right, the three layers show in order the fixed image , the moving one and the outcome of the registration ( transformed image). On this side, you are supposed to move the landmarks along the surface to match each chosen point in the two volumes and obtain a better result. Before scrolling down and applying the transformation, you need to select the registration type (affine) and mode (rigid). Once you are satisfied with the number of landmarks and their precision, you can proceed by clicking on the SAVE icon which should let you store landmarks, transform and transformed image or only the desired ones among these. Once this is completed, we can load the results of the registration between two different modalities (here MR and CT). On the software fsleyes, the result looks like below, with the MR image overlaid on the CT in a red- yellow scale and sagittal- coronal-axial slices showing that the two match correctly.

RkJQdWJsaXNoZXIy NTc3NzU=