What type of study is this?

This is a Quantitative Study study.

September 17, 2025

Ensemble Machine Learning Model to Classify Pathologic Complete Response from Pre-NAC Breast Cancer DCE-MRI Pharmacokinetic Maps

Key Points

The model successfully predicted pathologic complete response with AUROCs of 0.78 and 0.95 in internal and external datasets.
Using machine learning and DCE-MRI data, the model leverages tumor permeability and radiomic features.
This approach allows early identification of treatment response, which can help adjust therapies to enhance patient care.
The findings suggest that integrating pharmacokinetic maps into routine assessments can facilitate personalized treatment strategies.

Abstract

Motivation: Tumor permeability and interstitial fluid pressure can be linked with resistance in transporting chemotherapeutic agent and affect the neoadjuvant chemotherapy (NAC) treatment response. However, predictive models based on these parameters are currently lacking. Goal(s): To develop a model leveraging permeability and flow parameters from pre-NAC DCE breast MRI to predict pathologic complete response (pCR). Approach: DCE-MRI data was analyzed to generate maps of permeability and flow parameters and to extract tumor radiomic features from the maps, which were used to develop machine learning models. Results: The best model attained AUROCs of 0.78 and 0.95 in the internal test set and external dataset, respectively. Impact: A machine learning model to classify pCR from pre-NAC breast cancer DCE-MRI enables early prediction of treatment response. This would allow clinicians to make timely treatment adjustments and to alter treatment plans to lower toxicity and improve patient outcomes.

Ensemble Machine Learning Model to Classify Pathologic Complete Response from Pre-NAC Breast Cancer DCE-MRI Pharmacokinetic Maps

Key Points

Abstract

Cite This Study

Also Consider

Also Consider