This site contains the visualization of the result of our intra-procedural prostate motion study for each dataset used in the evaluation. Verification of deformable registration results remains difficult, since metrics such as landmark registration error (LRE) and Dice overlap coefficient (DSC) are often not sufficient to characterize the results of the registration. To provide a more complete characterization, we include these results before and after registration for the reader's reference.
This work is supported in part by NIH grant P41 EB015898 National Center for Image Guided Therapy (NCIGT), see http://ncigt.org for details.
Panel on the right shows "flicker" interchanging views of the consecutive slices from the needle confirmation image volume and the moving image (when registration is "OFF"), and needle confirmation image volume and the registered image (when registration is "ON"). Patient and needle image selector allow to select a specific dataset used in the evaluation. Shown DSC values illustrate a comparison of the contoured gland of the first and the last needle confirmation image with and without registration.