Design of a 32'' color TFT LCD PACS monitor and its clinical evaluation through ROC analysis

Abstract

One of the most important stages in digital radiology process is the transfer of the image to the observer, as the variations in light and color from a physical display. The common practice in observing these visual results is to use a medical grade LCD monitor. The main problem with medical grade monitors is their high cost. First objective of the study was to design a 32” LCD PACS monitor to be as compatible as possible to a medical grade LCD monitor with a remarkable cost advantage. The second was to test the hypothesis for a significant difference between a medical grade LCD monitor and the designed one in terms of diagnostic image quality. After the design’s validation, 60 digital radiographs with definite findings were obtained in cooperation with the authors of a previous study. Three experienced radiologists from Acıbadem Hospital examined these radiographs both on a medical grade Reference branded LCD monitor and on the 32” design. To check observers’ performance, the receiver operating characteristic (ROC) curves for all monitor-reader cases were statistically compared by using the same content and observers. The area under curve (AUC) of each ROC curve was used as a metric for detecting lung modules in the radiographs. With 95% confidence interval, the hypothesis was tested for a significant statistical difference between the related monitors. AUC for Reference monitor for observer 1, 2 and 3 were calculated as 0.634, 0.703 and 0.755 respectively. AUC for 32” design were 0.811, 0.746 and 0.811 For observer 1, the 32”design showed superior performance. For observer 2 and 3, Though AUC was far better on behalf of the new design, no significant statistical difference could be proven. As a result, it is possible to implement the new 32” design as a PACS monitor for medical diagnosis purposes without sacrificing any diagnostic value.|Keywords: PACS Monitor, ROC curve, AUC