# MRI‐based radiomics models for early predicting pathological response to neoadjuvant chemotherapy in triple‐negative breast cancer: A systematic review and meta‐analysis

## TL;DR

This study reviews and analyzes how MRI-based radiomics can predict treatment response in triple-negative breast cancer patients before surgery.

## Contribution

The study provides a meta-analysis of MRI-based radiomics models for early prediction of treatment response in triple-negative breast cancer.

## Key findings

- MRI-based radiomics showed a pooled AUC of 0.83 for predicting pathological complete response.
- SVM and LightGBM classifiers achieved the highest diagnostic efficacy with an AUC of 0.86.
- Low heterogeneity supports the consistency of MRI-based radiomics across studies.

## Abstract

This meta‐analysis evaluates the accuracy of MRI‐based radiomics in predicting pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) in triple‐negative breast cancer (TNBC) patients.

A systematic search of PubMed, Cochrane Library, Embase, Scopus, and Web of Science was conducted up to September 2024. Ten studies meeting inclusion criteria were assessed for methodological quality using the Radiomics Quality Score (RQS) and QUADAS‐2 tools. Pooled diagnostic performance metrics, including AUC, sensitivity, and specificity, were calculated using a fixed‐effects model.

The fixed‐effects model yielded a pooled AUC of 0.83 (95% CI: 0.79–0.86), with sensitivity of 0.80 (95% CI: 0.68–0.88) and specificity of 0.85 (95% CI: 0.76–0.91). Support Vector Machine (SVM) and Light Gradient Boosting Machine (LightGBM) classifiers demonstrated the highest diagnostic efficacy (AUC = 0.86). Heterogeneity was low (I
2 = 32%), supporting the use of a fixed‐effects approach.

MRI‐based radiomics exhibits strong and consistent predictive performance for pCR in TNBC patients undergoing NAC, supporting its potential as a non‐invasive tool for early treatment response assessment. Further standardization and prospective validation are needed for clinical implementation.