Prediction of incident cardiovascular events using machine learning and CMR radiomics

Abstract

Objectives Evaluation of the feasibility of using cardiovascular magnetic resonance (CMR) radiomics in the prediction of incident atrial fibrillation (AF), heart failure (HF), myocardial infarction (MI), and stroke using machine learning techniques. Methods We identified participants from the UK Biobank who experienced incident AF, HF, MI, or stroke during the continuous longitudinal follow-up. The CMR indices and the vascular risk factors (VRFs) as well as the CMR images were obtained for each participant. Three-segmented regions of interest (ROIs) were computed: right ventricle cavity, left ventricle (LV) cavity, and LV myocardium in end-systole and end-diastole phases. Radiomics features were extracted from the 3D volumes of the ROIs. Seven integrative models were built for each incident cardiovascular disease (CVD) as an outcome. Each model was built with VRF, CMR indices, and radiomics features and a combination of them. Support vector machine was used for classification. To assess the model performance, the accuracy, sensitivity, specificity, and AUC were reported. Results AF prediction model using the VRF+CMR+Rad model (accuracy: 0.71, AUC 0.76) obtained the best result. However, the AUC was similar to the VRF+Rad model. HF showed the most significant improvement with the inclusion of CMR metrics (VRF+CMR+Rad: 0.79, AUC 0.84). Moreover, adding only the radiomics features to the VRF reached an almost similarly good performance (VRF+Rad: accuracy 0.77, AUC 0.83). Prediction models looking into incident MI and stroke reached slightly smaller improvement. Conclusions Radiomics features may provide incremental predictive value over VRF and CMR indices in the prediction of incident CVDs. Key Points • Prediction of incident atrial fibrillation, heart failure, stroke, and myocardial infarction using machine learning techniques. • CMR radiomics, vascular risk factors, and standard CMR indices will be considered in the machine learning models. • The experiments show that radiomics features can provide incremental predictive value over VRF and CMR indices in the prediction of incident cardiovascular diseases.

Artificial Intelligence in the Image-Guided Care of Atrial Fibrillation

“…The process of automated extraction of a large number of radiomic features is known as radiomics [ 62 ]. Radiomic features can be classified as first-order features, shape features, and texture features, as illustrated in Figure 7 a. First-order features are based on the image histogram, shape features are based on the geometry of the structures studied, and texture features are based on the spatial distribution of the pixels [ 61 ]. Standardized definitions and validated reference values have been provided for a set of radiomic features [ 63 ], which can be extracted using open-source platforms, such as PyRadiomics [ 64 ] and QMaZda [ 65 ].…”

“…Ensuring effective feature selection is critical to avoid the curse of dimensionality. It helps reduce computational complexity, minimize the generalization error, and enhance the clinical explainability of the model [ 61 ]. Reproducibility is a vital aspect to consider when selecting radiomic features extracted from CT [ 70 ] or MRI [ 71 ].…”

“…Prediction models for incident AF have been developed in patients at risk of AF [ 52 , 59 ], as well as in the general population [ 61 ]. Cardiotoxicity induced by cancer therapy [ 81 ] poses a risk for cancer survivors, who may develop AF [ 82 ].…”

“…This model demonstrated the ability to predict incident AF with time-dependent AUCs of 0.80, 0.79, and 0.78 at 1, 2, and 3 years after LGE-MRI acquisition, respectively. Pujadas et al [ 61 ] predicted incident AF in the participants of the UK Biobank imaging enhancement [ 83 ]. Using radiomic and non-imaging features with a support vector machine (SVM) model, an AUC of 0.76 ± 0.07 was achieved.…”

Cardiac Radiomics Are Associated With Dyspnea

“…Pujadas et al recently showed that a combination of clinical risk factors, CMR-derived radiomics can predict incident heart failure in the UK biobank (accuracy 0.77, area under curve 0.83). 5 Collectively, the data suggests that CMR-derived radiomics may provide additional information not available in routinely collected clinical or CMR-derived data. Given the limitations of the study including lack of external validation, additional validation in a larger external cohort and association with long-term outcomes is warranted.…”