Delta Radiomics and Tumor Size: A New Predictive Radiomics Model for Chemotherapy Response in Liver Metastases from...

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Radiomic features correlate with tumor size on pretreatment images.
• This association changes post-treatment, influenced by treatment efficacy and patient response.
• Predicting chemotherapy response in liver metastases is crucial for effective treatment.

Purpose of the Study:

• To introduce and evaluate a novel baseline-referenced Delta radiomics model.
• To predict chemotherapy response in liver metastases from breast cancer (BC) and colorectal cancer (CRC).
• To integrate the relationship between radiomic features and tumor size into Delta radiomics.

Main Methods:

• Retrospective analysis of contrast-enhanced CT scans from 83 BC and 84 CRC patients.
• Extraction of radiomic features from segmented liver lesions (up to three per patient).
• Development and evaluation of classification models using pretreatment data, Delta radiomics, and baseline-referenced Delta radiomics.

Main Results:

• Baseline-referenced Delta radiomics performed comparably or better than existing radiomics models.
• Sensitivity, specificity, and balanced accuracy for predicting response ranged from 0.66–0.97, 0.81–0.97, and 80%–90%, respectively.
• The model demonstrated effectiveness in predicting tumor response in liver metastases.

Conclusions:

• Baseline-referenced Delta radiomics is a promising approach for predicting chemotherapy response.
• Integrating radiomic features and tumor size relationship enhances predictive capabilities.
• This method offers improved prediction for liver metastases in BC and CRC patients.