Dynamic Contrast Enhanced MRI Mapping of Vascular Permeability for Evaluation of Breast Cancer Neoadjuvant...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Oncology

Background:

• Dynamic contrast-enhanced (DCE) MRI is crucial for quantifying tumor microvascular permeability using pharmacokinetic (PK) models.
• Conventional PK modeling is computationally intensive, limiting its clinical application for DCE-MRI analysis.
• Image-to-image conditional generative adversarial networks (cGANs) show promise for complex image translation tasks.

Purpose of the Study:

• To develop and validate a novel cGAN-based approach for rapid generation of PK vascular permeability parameter maps from DCE-MRI data.
• To assess the accuracy and efficiency of the developed cGAN model in mapping vascular permeability.
• To evaluate the utility of the cGAN-derived permeability parameters for predicting neoadjuvant chemotherapy response in breast cancer patients.

Main Methods:

• Developed a novel image-to-image cGAN (DCE-to-PK cGAN) for direct mapping of DCE-MRI data to PK vascular permeability parameter maps.
• Trained and validated the cGAN model using open-source breast cancer patient DCE-MRI data from The Cancer Imaging Archive (TCIA).
• Analyzed the percentage change in K_trans derived from cGAN-generated maps to predict treatment response.

Main Results:

• The DCE-to-PK cGAN successfully generated high-quality vascular permeability parameter maps comparable to ground truth.
• The cGAN approach achieved a computational speed-up exceeding 1000-fold compared to conventional PK modeling.
• Early prediction of neoadjuvant chemotherapy responders was achieved by analyzing the percentage change in K_trans, consistent with previous studies.

Conclusions:

• The developed DCE-to-PK cGAN offers a highly efficient and accurate method for generating vascular permeability maps from DCE-MRI.
• This AI-driven approach significantly reduces analysis time, facilitating clinical translation.
• The cGAN-derived K_trans changes provide a reliable biomarker for early prediction of treatment response in breast cancer.