Key result
Quantitative LGE radiomics discriminates cardiac sarcoidosis from HCM with an AUC of ~1.0.
Why the study?
Do quantitative LGE radiomics patterns on CMRI discriminate cardiac sarcoidosis from hypertrophic cardiomyopathy?
Observational (n=1,285)
Do quantitative LGE radiomics patterns on CMRI discriminate cardiac sarcoidosis from hypertrophic cardiomyopathy?
Answers are grounded in this paper and cite their sources.
Quantitative LGE radiomics patterns on CMRI can accurately discriminate cardiac sarcoidosis from hypertrophic cardiomyopathy and correlate with left ventricular ejection fraction.
Introduction: Cardiac sarcoidosis (CS) is characterized by patchy granulomatous inflammation of the myocardium. Late gadolinium enhancement (LGE) on cardiac magnetic resonance imaging (CMRI) highlights areas of myocardial fibrosis and supports a diagnosis of CS. However, while LGE offers prognostic value, findings can be nonspecific and overlap with other nonischemic cardiomyopathies, including hypertrophic cardiomyopathy (HCM). We assessed quantitative 'radiomics' patterns of LGE on CMRI to investigate specificity for CS and relationship with left ventricular ejection fraction (LVEF), a measure of cardiac dysfunction. Methods: We segmented LV myocardium on CMRI scans from 56 patients with CS to extract 953 LGE radiomic features. We performed principal component analysis (PCA) on an independent cohort of 1,229 patients with HCM from Beth Israel Deaconess Medical Center, using similarly computed LGE radiomics methods. CS features were projected into the PCA space of the HCM cohort. Using the resulting principal components (PCs), we characterized the relationship of CS LGE radiomics features to cardiomyopathy etiology and LVEF. Specifically, we performed cross-validated logistic regression to assess whether PCs could discriminate CS from HCM. We then performed linear regression to assess the association between these PCs and LVEF. Results: Logistic regression with PCs distinguished CS from HCM, with increasing performance with the number of PCs, from area under the receiver operating curve (AUC)=0.607 ± 0.023 (3 PCs) to AUC=0.995 ± 0.005 (10 PCs). Linear regression with CS PCs showed that variance in LVEF is explained from R2= 0.187 (3 PCs) to R2 = 0.381 (10 PCs). Conclusions: Quantitative LGE patterns hold promise for discriminating CS from HCM, potentially improving diagnostic accuracy compared to nonspecific qualitative findings. Furthermore, radiomics computed on LGE exhibits a relationship to LVEF, indicating potential as a prognostic biomarker.
No takes yet. Share an insight, caveat, or question.
Le et al. (2025) conducted an observational in Cardiac sarcoidosis and hypertrophic cardiomyopathy (n=1,285). Quantitative late gadolinium enhancement (LGE) radiomics patterns vs. Hypertrophic cardiomyopathy cohort was evaluated on Discrimination of cardiac sarcoidosis from hypertrophic cardiomyopathy (AUC 0.995 ± 0.005). Quantitative LGE radiomics patterns successfully discriminated cardiac sarcoidosis from hypertrophic cardiomyopathy, achieving an AUC of 0.995 ± 0.005 using 10 principal components.
Synapse has enriched 10 related papers. Consider them for comparative context: