Improvement of observer performance during fluoroscopy by local adaptive contrast enhancement

The ability of a video processor (FluoroVision FV-1), which performs two-dimensional locally adaptive contrast enhancement, to improve the detection of a low-contrast object was evaluated by means of receiver operating characteristic (ROC) curves. Three independent observers viewed a videotape of 50 fluoroscopic images of a varied background, in which a test object was present in 25. Each observer viewed the tape under four conditions: (1) no processing, (2) temporal noise reduction (gaussian weighted time averaging, (3) contrast enhancement, and (4) both noise reduction and contrast enhancement. The results were that detection was significantly improved by the locally adaptive contrast enhancement. Noise reduction did not significantly improve performance, probably because washer detection was limited by background contrast variations as well as noise and because only a small amount of noise reduction was used. The authors conclude that the processing device is potentially valuable in improving the quality of clinical fluoroscopic images