Detection ability of Model Observer for diagnosis of active cardiac sarcoidosis on 18F-FDG-PET

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Objective: Sarcoidosis is an inflammatory granulomatous disease of unknown cause, and cardiac sarcoidosis (CS) is the main cause of death in patients with sarcoidosis. Cardiac 18 F-Fluorodeoxyglucose (FDG)-positron emission tomography (PET) is used to detect active CS, based on the patterns of 18 F-FDG uptake in the myocardium. In this study, we investigated the ability of a model observer, Channelized Hoteling Observer (CHO), in differentiating the active CS pattern, in comparison with the diagnosis by the expert cardia nuclear medicine (NM) physician as a reference standard. Methods: This retrospective study included 223 patients who underwent 18 F-FDG PET scans because they were clinically suspected of having CS. One expert cardiac NM physician and one oncology NM physician classified the cardiac 18 F-FDG uptake into four patterns (focal, focal on diffuse, diffuse, none) for all PET images. The oncology NM physician additionally determined the confidence scale of CS expressed as a percentage (0-100). The focal and focal on diffuse patterns were defined as CS patterns. In CHO, 80 subjects (CS: 40, non-CS: 40) were used for training data sets, and 143 subjects (CS: 66, non-CS: 77) were used for evaluation data sets of objective CS discrimination ability, respectively. The confidence scale of CS by CHO was also outputted as a percentage (0-100). Finally, ROC analysis was used to compare discrimination ability between the oncology NM physician and CHO. Results: The match rate of classified subject numbers between the two physicians was moderate (Kappa coefficient of 0.57). In comparison with the judgment of the expert cardiac NM physician, the Area under the ROC curve of oncology NM medicine and that of CHO were 0.73 and 0.78, respectively. Conclusions: The detection ability of CHO in CS pattern of myocardial 18 F-FDG uptake was almost equivalent to that of the oncology NM physician when the diagnosis by the cardiac expert NM physician was defined as a reference standard. The model observer has a potential alternative for assessing the discrimination ability of 18 F-FDG uptake patterns of CS.

Article activity feed