Purpose: To propose the use of a psychophysical staircase method for determination of absolute detection thresholds in task-based evaluation of fluoroscopy systems.
Methods: A neonate radiograph was iodine printed to serve as an anatomically realistic background for the clinical imaging task of vesicoureteral reflux (VUR) detection. Ten pages for each depiction of ureter-kidney anatomies of two VUR severity grades (3 and 5) were printed with single passes of iodine ink and imaged on a SIEMENS Agile Max fluoroscopy unit. A staircase method was used to determine the absolute detection threshold. In this method, stimuli corresponding to the presence of VUR were presented to the reader in a graduated descending or ascending contrast scale removing or adding a single page from the stack. When tasks were presented in descending order, the number of stacked pages when the task became undetectable was reported. Tasks were then presented in ascending order of contrast and the number of stacked pages when the task became visible was recorded. This process was repeated for a total of five stairsteps per reader for high (61 kVp, 24.4 mA) and low (60 kVp, 13.5 mA) dose, and the average detectability thresholds were calculated. Viewing conditions were kept constant throughout the study.
Results: Absolute detectability thresholds decreased with decreasing dose with values of 4.3±0.49 versus 5.9±0.42 for grade 3 high and lose dose, respectively. Grade 5, being larger and with less detail, had overall higher detection thresholds with values of 4.7±0.58 and 6.1±0.20 for high and low dose, respectively.
Conclusion: The proposed methodology was used to determine the absolute intensive detection threshold of VUR as a surrogate task for quantitative, task-based assessment of fluoroscopic image quality inside the fluoroscopy suite. Future work will extend this methodology to compare different fluoroscopy systems.
Funding Support, Disclosures, and Conflict of Interest: This project was supported by the Department of Radiology Hodges Society, the Women's Board of the University of Chicago, the NIH T-32 Training Grant T32 EB002103, and the 2020-2022 AAPM Graduate Fellowship.
Image Visualization, Lesion Detectability