Table 1.

Previous CMR studies about non-contrast AI models in ischemic and non-ischemic cardiomyopathy.

Authors	Years	Number of Patients	Variables	Results
Baessler et al. [59]	2018	120	Radiomics features of cine-CMR images	The model demonstrated an AUC of 0.93 and 0.092 to diagnose large and small myocardial infarction on cine-CMR.
Avard et al. [60]	2022	72	Radiomics features of cine-CMR images	The authors reported optimal performance for the logistic regression model with an AUC of 0.93 ± 0.03, an accuracy of 0.86 ± 0.05, a recall of 0.87 ± 0.1, a precision of 0.93 ± 0.03, an F1 Score = 0.90 ± 0.04, and for the support vector machine model with an AUC of 0.92 ± 0.05, an accuracy of 0.85 ± 0.04, a recall of 0.92 ± 0.01, a precision of 0.88 ± 0.04, and an F1 Score of 0.90 ± 0.02, respectively.
Larroza et al. [61]	2017	44	Radiomics features of cine-CMR images	Radiomics analysis of cine-CMR images achieved an AUC of 0.82 ± 0.06 with a sensitivity of 0.79 ± 0.10, and specificity of 0.80 ± 0.10 for differentiation of acute myocardial infarction from chronic myocardial infarction.
Larroza et al. [62]	2018	50	Radiomics features of cine-CMR images	The model demonstrated an AUC of 0.849, and a sensitivity of 92% to detect nonviable segments, 72% to detect viable segments, and 85% to detect remote segments.
Zhang et al. [63]	2019	212	Cine-CMR	The DL model showed a sensitivity of 89.8% and a specificity of 99.1%, with an AUC of 0.94.
Zhang et al. [64]	2022	843	Cine-CMR images, T1 mapping images	Virtual native enhancement demonstrated a strong correlation with LGE in quantifying scar size (R, 0.89; intraclass correlation coefficient, 0.94) and transmurality (R, 0.84; intraclass correlation coefficient, 0.90), achieving an accuracy of 84% for detecting scars with a specificity of 100% and sensitivity of 77%, and excellent visuospatial agreement with the histopathological porcine model.
Chen et al. [65]	2022	150	Physiological, clinical, and paraclinical features	The proposed model demonstrated a mean error of 0.056 and 0.012 for the quantification, and 88.67 and 77.33% for the classification accuracy of the state of the myocardium.
Goldfarb et al. [66]	2019	90	CMR water–fat separation and parametric mapping	The DL model could visualize myocardial fat deposition in chronic myocardial infarction and intramyocardial hemorrhage in acute myocardial infarction.
Xu et al. [67]	2020	165	Cine-CMR images	The proposed AI-based model achieved a pixel classification accuracy of 96.98%, and the mean absolute error of the infarction size was 17.15 mm2.
Xu et al. [68]	2018	165	Cine-CMR images	The proposed framework for the pixel-wise delineation of the myocardial infarction area achieved an accuracy of 95.03% and optimal consistency (Kappa statistic: 0.91; Dice: 89.87%) in comparison to the ground truths manually segmented LGE images.
Abdulkareem et al. [69]	2022	272	Cine-CMR images	The SVM model achieved accuracy, F1, and precision scores of 0.68, 0.69, and 0.64, respectively. Conversely, the DT models achieved accuracy, F1, and precision scores of 0.62, 0.63, and 0.72, respectively.
Zhang et al. [70]	2021	1196	Cine-CMR images, T1 mapping images	The authors reported high agreement between virtual native enhancement and LGE in the visuospatial distribution and quantification of lesions.
Baeßler et al. [71]	2018	32	Radiomics features of T1 mapping images	The proposed ML model achieved an AUC of 0.95 with a diagnostic sensitivity of 91% and a specificity of 93%.
Fahmy et al. [72]	2022	759	Radiomics features of cine-CMR	The DL model using radiomics data of cine-CMR images correctly identified 43% and 28% of HCM patients without scars in the internal and external datasets.
Cau et al. [73]	2022	43	CMR parameters, demographics data	The model showed a sensitivity of 92% (95% CI 78–100), specificity of 86% (95% CI 80–92), and AUC of 0.94 (95% CI 0.90–0.99) in diagnosing Takotsubo cardiomyopathy.
Eckstein et al. [74]	2022	96	CMR strain and function parameters	The supervised ML model demonstrated an accuracy of 90.9% (0.996; precision = 94%; sensitivity = 100%; F1 Score = 97%) to identify cardiac amiloidosis patients.
Krebs et al. [75]	2021	350	Cine-CMR images	The proposed score remained significantly associated with ventricular arrhythmia after adjustment in multivariable regression analysis.