Accurate knowledge of the disease volume is required for planning a course of radiotherapy. The gross disease volume is usually demarcated on anatomic images including radiographic, CT, and/or MRI studies. Functional imaging, using positron-emission tomography (PET) with fluorine-18 deoxyglucose (18F-FDG) administration is being investigated for identifying disease tissues with elevated metabolic activity. Elevated specific uptake values (SUV) of FDG within the tumor bed may provide additional information for defining the planning target volume (PTV) requiring radiation therapy. However, the measured SUV distribution is due both to physical resolution limitations of PET scanners and biological variations of FDG accumulation between tissues. The biological variation of FDG in PET images was obtained after processing PET images using a modified Wiener filter WF(f) to correct for the resolution limitations of PET. The frequency dependent WF is given by (1a)WF(f)=1/MTF(f) for f < fcut (1b)WF(f)=(1/MTF(fcut))∗exp-(kf2) for f > fcut where MTF(f) is the modulation transfer function determined from the measured point-spread function of the PET scanner, and k is a parameter that describes the Gaussian roll-off of the Weiner filter at spatial frequencies greater than fcut. The values of the parameter k and fcut were bracketed from an analysis of PET-FDG images of a phantom containing various size spheres with known activity. The observer has the option of viewing PET-FDG images reconstructed with low, medium, and high spatial frequency gains by selection of the parameters k and fcut in equations 1a and b. Image processing both sharpens boundaries and increases contrast of objects in the PET image. For example, as a result of image processing with a low-gain filter the maximum counts in the image of the 18 mm sphere and its percent contrast increased by a factor of 2. The measured FWHM of the image of the 18-mm diameter sphere containing FDG decreased by 1.19. Preliminary clinical data on processed FDG-PET brain images also produced images with sharper boundaries of the disease tissue and greater enhancement of the contrast between normal and the metabolically active disease volume (MADV). An analysis of the activity in the designated normal and disease tissue ROIs was used to determine the MADV. An algorithmic identification of the MADV within the tumor bed was developed and used to investigate the relationship between the determined MADV before and following image processing. The SUV distribution was analyzed in two regions containing; (1) normal tissue and (2) gross disease with at least a 1-cm margin in all directions. The SUV value that included 95% of the voxels in the normal region of interest SUVN95 was determined and used as a lower threshold for elevated SUVs in the ROI containing the gross disease. Preliminary data indicated that the determined MADV decreased between 10%–15% with image processing. These differences may be significant in defining the planning target volume used in radiotherapy.