Novel Radiomics Features for Automated Detection of Cardiac Abnormality in Patients with Pacemaker.

Area of Science:

• Cardiology and Medical Imaging
• Artificial Intelligence in Healthcare
• Biomedical Engineering

Background:

• Cardiac pacemakers are crucial for treating symptomatic bradycardia but can lead to pacing-induced cardiomyopathy (PICM) due to altered conduction pathways.
• PICM, characterized by reduced left ventricular ejection fraction (LVEF) from chronic right ventricle pacing, lacks effective early detection and treatment standards.
• Current diagnostic methods for cardiac desynchrony are often subjective and may not detect subtle changes indicative of early-stage PICM.

Purpose of the Study:

• To develop and evaluate a novel computer-aided diagnosis (CAD) tool for early and accurate detection of pacemaker variations and PICM.
• To utilize machine learning models and radiomics features from heart ultrasound images for classifying pacemaker function and identifying PICM indicators.

Main Methods:

• A computer-aided diagnosis (CAD) system was developed using machine learning models: decision tree, Support Vector Machine (SVM), random forest, and AdaBoost.
• Radiomics features, specifically textures, were extracted from heart ultrasound images and screened using Relief-F scores.
• The top-ranked features were used to train and test the machine learning models for classifying pacemaker variations and detecting PICM.

Main Results:

• The random forest model achieved a maximum test accuracy of 97.73% using four radiomics features from the R-wave dataset.
• The SVM model achieved a maximum test accuracy of 96.59% using three radiomics features from the T-wave dataset.
• The experimental results demonstrate the robustness of the proposed CAD system in accurately detecting pacemaker variations.

Conclusions:

• The developed CAD tool shows significant potential for the early and accurate detection of pacing-induced cardiomyopathy.
• Machine learning analysis of radiomics features from cardiac ultrasound images offers a promising noninvasive approach for monitoring pacemaker function.
• This system could aid clinicians in timely intervention and management of patients at risk of developing PICM.