9:15—9:30

Malignant tumors exhibit increased glucose metabolism which can be quantitated by SUV. SUV is criticized for its variability resulting from many factors including the method of drawing region of interest (ROI) over the tumor. The most common method manually draws or places ROI on various slices displaying highest FDG activity. This time consuming method is associated with significant individual variation. We describe a more reproducible, efficient spherical 3-D blob analysis method of SUV and tumor volume (TV) determination to evaluate CTA response. The spherical 3-D Blob analysis program is a completely automated method with data processing performed using IDL (RS Inc., Colorado). A threshold value is set for the ROI; all voxels above threshold are grouped by connectivity. Grouped voxels, called “blobs”, are displayed and statistics are calculated for each group. The threshold set for blob extraction and SUV determination is usually three times above background and is constant on repeat scans. Eight patients with metastatic carcinoma underwent PET/CT/MRI prior to and one week after chemotherapy. Three patients also had scans at one month. Maximum and average SUV's and TV were determined and appeared to be very reproducible when there was no clinical response or change on CT/MR. In these patients, repeat SUV's and volumes of the lesions (n = 38) varied by less than 20% of baseline. An increase/decrease in maximum SUV or TV correlated with a similar change in lesion size on CT/MR. The average SUV did not change. The spherical 3-D blob analysis program appears to be a reliable, efficient method of determining maximum SUV and volumetric measurements for following tumor response to CTA.