The additional value of image subtraction analysis in nuclear medicine

1031 Learning Objectives Nuclear medicine (NM) physicians often perform visual side-by-side comparison of planar or tomographic images of the same patient. These images may be part of a longitudinal analysis (e.g. staging in oncology) or the NM investigations may be performed to compare two conditions (e.g. ictal-interictal SPECT, stress-rest myocardial perfusion). Using visual side-by-side comparison, NM physicians then try to improve the diagnosis. However, this is very time consuming, rather subjective, and the more subtle but significant differences may not always be appreciated by means of this method alone. Therefore, we studied the use of an image subtraction analysis technique which could be of additional help during the inspection and comparison of NM images. Image subtraction analysis requires the following steps: (1) image registration (rigid, non-rigid), (2) image normalization, (3) image subtraction, and (4) post-processing (e.g. smoothing, conversion to a z-score, thresholding). Over recent years, some sophisticated rigid and non-rigid image registration algorithms have been developed, providing the necessary solutions for the first step. The remaining steps, however, need much more attention. Currently, there are only a few commercial software packages that allow for image subtraction analysis. Since most of them are used for neurological applications, we have developed a software package that is more flexible and can be used for further investigation and exploration of this technique. Hopefully, this could lead to a more general approach of subtraction analysis. We present our experience with the subtraction technique and discuss the need for further refinements of the method. We show some clinical examples and indicate some other NM examinations that could benefit from this subtraction technique, e.g.: FDG-PET, lung ventilation-perfusion, dual isotope parathyroid, and bone imaging. In conclusion, we used a subtraction analysis technique as a usefull tool for the inspection and comparison of NM images of the same patient