What does this research mean for the field?

High-resolution restriction spectrum imaging (RSI) is a promising non-contrast MRI method that improves the detection of small breast cancer lesions compared to standard diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) MRI. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To evaluate the effectiveness of high-resolution restriction spectrum imaging for non-contrast detection of breast cancer lesions.

February 19, 2026

Abstract PS1-06-07: Non-contrast detection of breast cancer using high-resolution restriction spectrum imaging

Key Points

To evaluate the effectiveness of high-resolution restriction spectrum imaging for non-contrast detection of breast cancer lesions.
Developed a novel RSI model using multi-shell diffusion-weighted imaging (DWI).
Collected high-resolution multi-shell DWI on breast cancer patients using multi-band acquisition.
Calculated tri- and tetra-exponential models to analyze diffusion components and assess lesion conspicuity.
Measured contrast-to-noise ratios (CNR) and established receiver operating characteristic (ROC) curves for model performance.
3C-RSI and 4C-RSI models demonstrated high contrast-to-noise ratios compared to DCE and ADC.
The area under the curve (AUC) for 3C-RSI was 0.87 and for 4C-RSI was 0.96, indicating superior diagnostic performance.
RSI models achieved higher specificity compared to ADC, particularly in differentiating between benign and malignant lesions.

Abstract

Abstract Problem: Breast MRI is recommended annually for high-risk individuals, typically using dynamic contrast-enhanced (DCE) MRI, which requires intravenous gadolinium. However, the added cost and invasiveness limit its use in average-risk populations. Developing a non-contrast MRI method would allow for expansion to broader screening. Diffusion-weighted imaging (DWI), a non-contrast technique, holds promise but suffers from low spatial resolution, limiting detection of small lesions. Restriction Spectrum Imaging (RSI), which uses multi-shell DWI, separates signal into restricted, hindered, free, and flow components without exogenous contrast, and has shown improved lesion characterization. This abstract presents a novel RSI model using multi-b-value input and high-resolution multi-band (MB) acquisition, achieving 2 mm isotropic resolution and improving small lesion detectability. Methods: Along with standard clinical DWI and DCE, high-resolution multi-shell (b-values (# of directions) 0 (1), 100 (6), 800 (6), 1500 (6), 3000 (12) s/mm2) DWI was collected using MB on 122 breast cancer and screening patients. RSI models follow a multi-exponential formula: S(b) = S0Σexp(-b*ADCi) = ΣCiexp(-b*ADCi). Tri- and tetra-exponential models (3C- and 4C-RSI) ADCi (fixed compartment apparent diffusion coefficients) were first calculated on large cancers (largest dimension 2cm - 61 cancers). Ci signal contribution maps (C-maps) were estimated and provided spatial representation of each diffusion component. The products C1C2 and C2C3 were calculated as well. ADC maps from clinical DWI and DCE subtraction images (peak enhancement - pre-phase) were computed. To assess lesion conspicuity, contrast-to-noise ratios (CNR) were measured in C-maps on 28 lesions from a subset of 22 patients and compared to CNR from ADC and DCE. Finally, receiver operating characteristic (ROC) curves of each model and ADC were measured on the subset, separating malignant from benign lesions, by fitting a generalized linear model. Results: RSI models’ separation of diffusion contents are summarized on Table 1 . Regarding CNR, 3C-RSI showed highest values in C1C2 (median: 3.42, IQR: 2.66) and C2C3 (2.72, 1.72). For 4C-RSI, C1C2 (3.15, 3.42) and C2C3 (3.66, 3.79) were also highest. DCE had the highest (3.86, 1.45) and ADC the lowest (1.29, 0.49). Of the 28 test lesions (median lesion size: 2 cm, min: 5 mm, max: 7 cm - largest dimension), 7 were benign. ADC had an AUC of 0.78 while 3C- and 4C-RSI had an AUC of 0.87 and 0.96 respectively. DCE is considered 100% sensitive as selected lesions are based on DCE findings and had a specificity of 75% (7 lesions are benign), while at 80% sensitivity ADC, 3C- and 4C-RSI had a specificity of 50, 83 and 100%, respectively. Conclusion: RSI models achieved comparable CNR to DCE, outperforming ADC. They also had the highest AUCs and specificities, highlighting RSI as a potential non-contrast method for early detection. Citation Format: S. Loubrie, J. Lim, S. Batasin, H. J. Yu, J. Zou, S. Ebrahimi, C. C. Conlin, A. Guidon, T. M. Seibert, A. M. Dale, A. Wallace, H. Ojeda-Fournier, A. E. Rodriguez-Soto, R. Rakow-Penner. Non-contrast detection of breast cancer using high-resolution restriction spectrum imaging abstract. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS1-06-07.

KI fragen

Bookmark