We present our work on developing anatomically realistic, quantitative multiparametric MRI (qMRI) phantoms designed to standardize imaging of both healthy and diseased breast tissue. The lack of standardized, morphology-accurate phantoms currently limits the effectiveness of qMRI in both clinical and research settings. Without realistic tissue-mimicking models, radiologists, technologists, and researchers face challenges in learning optimal acquisition protocols and interpreting quantitative data.

Conventional relaxometry and diffusion phantoms provide quantitative references but fail to capture the anatomical complexity and heterogeneity of the human breast. To address these limitations, we have developed a patient-derived, 3D-printed breast phantom that replicates realistic breast morphology and composition and incorporates quantitative relaxation and diffusion (ADC) parameters. It includes mimics for fibroglandular and adipose tissues, as well as benign and malignant tumor regions, enabling physiologically relevant representation of both normal and diseased states.

This work establishes a foundation for organ-specific qMRI quality control and assurance (QC/QA), improving reproducibility, training, and the development of robust AI models for quantitative breast imaging.

Stephen Russek (National Institute of Standards and Technology Boulder), Todor Karaulanov (CaliberMRI, Inc. (CMRI)), Willaim Hollander, Chamni Jayarathna, Callie Weiant, John Wenzel, Shannon John