Quality assurance in PET: evaluation of the clinical relevance of detector defects.

UNLABELLED Defective detector blocks in PET may cause serious image artifacts. To estimate the influence of malfunctioning detectors on image quality, a method is described for transferring the actual detector defect onto previously acquired scans. METHODS Consequences of detector defects of varying types and extensions were simulated in phantom studies as well as in clinical 18F-fluorodeoxyglucose investigations. First, a condition frame was obtained by dividing the sinogram of a blank measurement, obtained with rod sources on the defective PET camera, by the sinogram of a reference blank acquired before the appearance of the defect. Second, the sinogram of a previously acquired typical patient study was multiplied by the condition frame and reconstructed. Thereafter, images from corrupted sinograms were compared visually with their originals. For repairing defective sinograms, linear interpolation and the constrained Fourier space method were tested. RESULTS The effects of detector defects can be simulated accurately in patient studies. The correction methods applied are especially helpful in cases of (a) several neighboring defective detectors and small study objects, (b) small hot artifacts and (c) several nonadjacent defective detectors. Linear interpolation is faster than the constrained Fourier space method; it is more widely applicable and provides similar results. CONCLUSION The proposed approach allows specific evaluation of clinical consequences of detector defects. This technique simplifies the decision as to whether a planned patient study can be performed or must be postponed. Even in cases of serious detector problems, sinogram repair may help eliminate image artifacts and minimize the loss of image quality.