Radiomic Cardiac MRI Signatures for Predicting Ventricular Arrhythmias in Patients With Nonischemic...

Area of Science:

• Cardiology
• Medical Imaging
• Radiomics

Background:

• Risk stratification in nonischemic dilated cardiomyopathy (DCM) is challenging.
• Late gadolinium enhancement (LGE) cardiovascular magnetic resonance identifies major arrhythmia risk factors.
• The prognostic utility of LGE radiomics in DCM is not well-established.

Purpose of the Study:

• To determine if LGE radiomics can enhance arrhythmia risk stratification in DCM patients.
• To compare the prognostic value of LGE radiomics against existing clinical and imaging markers.

Main Methods:

• Retrospective analysis of DCM patients who received primary prevention implantable cardioverter-defibrillators (ICDs).
• Extraction of left ventricular myocardial radiomic features from LGE cardiovascular magnetic resonance images.
• Development of logistic regression models with and without LGE radiomics to predict appropriate ICD intervention (VT/VF).

Main Results:

• LGE radiomics significantly improved risk prediction models in both development (C-statistic 0.71 vs 0.61) and external validation cohorts (C-statistic 0.70 vs 0.61).
• Radiomic analysis demonstrated superior risk stratification capabilities, evidenced by improved net reclassification indices.
• Gray level co-occurrence matrix autocorrelation emerged as an independent predictor of VT/VF in both cohorts.

Conclusions:

• LGE radiomics analysis offers significant added prognostic value for predicting ventricular arrhythmias in DCM patients.
• This approach surpasses the predictive capability of LGE presence alone.
• Radiomic insights from LGE imaging represent a promising advancement in DCM risk stratification.