Novel Functional Radiomics for Prediction of Cardiac Positron Emission Tomography Avidity in Lung Cancer Radiotherapy

Area of Science:

• Nuclear medicine imaging
• Radiomics analysis
• Cardiovascular oncology

Background:

• Traditional cardiotoxicity assessment relies on radiation dose to the heart.
• Functional imaging offers potential for improved cardiotoxicity prediction in lung cancer patients.
• 18F-FDG PET/CT is a standard tool for cancer staging.

Purpose of the Study:

• To develop a radiomics model for predicting clinical cardiac assessment.
• Utilize 18F-FDG PET/CT scans obtained before thoracic radiation therapy.
• Identify patients at risk for cardiotoxicity.

Main Methods:

• Utilized pretreatment 18F-FDG PET/CT scans from three patient cohorts (N=209).
• Clinician classified cardiac FDG uptake (no, diffuse, focal).
• Developed a radiomics model using 210 features, reduced to nine, with 10-fold cross-validation.

Main Results:

• Cardiac FDG uptake patterns were classified: no uptake (39.6%), diffuse (25.3%), focal (35.1%).
• The best radiomics model achieved 93% accuracy on training data.
• External validation showed 80% and 92% accuracy on two datasets.

Conclusions:

• A functional cardiac radiomic model was developed from standard 18F-FDG PET/CT scans with good predictive accuracy.
• The model can automate the prediction of cardiac conditions.
• It serves as an early functional biomarker for identifying patients at risk of radiotherapy-induced cardiac complications.