Presentation Differentiation of cardiac sarcoidosis lesions and physiological FDG uptakes using texture analysis in patients with cardiac sarcoidosis

Osamu, Manabe  ,  Hirata, Kenji  ,  Ohira, Hiroshi  ,  Yoshinaga, Keiichiro  ,  Noriko, Oyama-Manabe  ,  Tsujino, Ichizo  ,  Nishimura, Masaharu  ,  Tamaki, Nagara

2017-05-09
Description
Background: Physiological 18F-fluorodeoxyglucose (FDG) uptake on PET/CT makes it difficult to distinguish between cardiac involvement of sarcoidosis (CS) and non-specific FDG uptake in the normal myocardium. Texture analysis is a group of computational methods extracting information about the relationship between adjacent pixels or voxels. Texture analysis evaluates the inhomogeneity of regional interest (ROI) of FDG accumulation and aims to improve the diagnostic accuracy of FDG PET. However, there has been no previous data to evaluate the efficacy of this approach for detecting CS. The aim of this study was to distinguish the CS lesions from physiological uptakes using the texture analysis in patients with CS and non-CS.Methods: Twenty-one CS patients (58.2 ± 15.0 y) (CS group) and 53 patients (62.5 ± 13.0 y) who underwent FDG PET/CT for detecting malignant tumors with any FDG cardiac uptake (non-CS group) were studied. All CS patients were fasted over 18 hours with low carbohydrate (LCD) preparation prior to the scan. ROIs were put on the left ventricle to make a polar map (256 × 256 matrices). Orlhac’s definition for the texture analysis was applied. Histogram analysis computed 5 parameters, and 4 matrices computed 31 parameters. Thus, a total of 36 texture parameters were evaluated for each patient.Results: Maximum of the standardized uptake value (SUVmax) showed no difference in cardiac FDG uptake between the two group (7.36 ± 2.77 vs. 8.78 ± 4.65, P = 0.45). In contrast, 29 of 36 texture parameters showed significant difference between the two groups. Among texture parameters, CS group showed higher long-run emphasis (LRE) parameter, a kind of gray-level run length matrix, than that of non-CS (32.9 ± 31.8 vs. 11.1 ± 4.42, p < 0.0001). Using this parameter, the sensitivity, specificity and diagnostic accuracy of CS were 100%, 72%, and 80%, respectively. Conclusion: FDG SUVmax did not differentiate cardiac FDG uptake between the CS and non-CS group. In contrast, the texture analysis showed higher inhomogeneity of cardiac FDG uptake in CS than physiological FDG uptake. The texture parameters may be useful for characterizing LV myocardial FDG uptake in CS.
ICNIC2017

Number of accesses :  

Other information