In PET imaging, anatomical expertise is essential because images are so fuzzy. I'm involved in a project where we want to automatically detect cancer lesion (for instance, lymphoma nodes) using PET/CT imaging. Lesions typically don't show up in the CT but the CT could be used for registering an anatomical model, for instance using the skeleton information, which would then be used to weed out false positives based on the location. For instance, the heart and liver show up prominently in the PET result, but there isn't any lymph nodes in either. Is there any work where an anatomical model is used in such a context? Thanks for any pointers.
I am not familiar with whole body PET being a brain person, however thinking from first principles you need to find a good 'anatomical' image which has enough common boundaries with your radioligand image and then use it to coregister them using some least squares type algorithm. You may get better results by separately dealing with separate structure for example as current SPM methods do for brain structures. You then need to be able to deal with
a) interscan movement - easy
b) intrascan movement- more difficult but methods exist to identify scans in which movement has occurred
c) partial volume corrections- essential if you want to increase signal sensitivity from small structures
I am sure that there will be multiple papers on this and you can then contact the relevant physicists
We have found layering the CT with the PET in a volume rendering mode has worked well. Requires good image acquisition parameters
As many centers use PET-CT but the reader is Nuclear Medicine physician if any endeavor is made to annotate the scan (based on automatic method where moment the curser is pointed to an image automatically anatomical configuration also highlights helping to correlate with anatomic CT in PET scan will be of great help.
the best is sit together with the radiologist and together you will perform the best.
CT scans used in PET/CT scanners are not usually acquired for good anatomical rendering, more for attenuation correction based on gross tissue density. I guess the CT parameters can be tweaked but that might require a double dose. Is there no way to rescan the subjects with an MRI that can be later registered to the PET/CT for much better anatomical definition? Even in the brain, fine anatomy (eg hippocampus, hypothalamus etc) differs enough to make the use of templates and models dubious. In the abdomen, I'm sure more so. Without decent anatomical boundaries in an MRI, I'm not sure how you would define something like a lymph node.
Thanks for all the good answers. We cannot rescan the patients in an MRI, we only have an anonymized database of PET+CT exams. This is true that the CT is mostly used for the attenuation but clinicians do use it for anatomical reference as well.
- Badges
- Science topic
- Similar topics
- Gels