Volumetric Quantification of Breast Tissue Composition Using Transmission Ultrasound: A Validation Study Against MRI

Abstract

Rationale and Objectives:

Accurate quantification of fibroglandular tissue volume (FGV) and the fibroglandular ratio which is the ratio of FGV to the total breast volume (TBV) (FGR = FGV/TBV) is critical for assessment of a patient’s lifetime breast cancer risk and personalized screening strategies. Magnetic resonance imaging (MRI) is considered the research standard for volumetric breast composition analysis but is limited by cost, availability, and contraindications. This study evaluates the capability of quantitative transmission (QT) imaging technology to measure FGV and FGR and compares these results with MRI-derived metrics, establishing quantitative equivalence between the two modalities.

Materials and Methods:

A retrospective analysis included 53 breasts from 29 women imaged using QT scan and MRI within 30 days as part of the ACCRUE Study (ClinicalTrials.gov NCT03052166). QT data were acquired with the QT Scanner 2000 Model A and reconstructed into 3D speed-of-sound maps and respective reflection maps via a full-wave inverse scattering algorithm. FGV and FGR were computed using a fuzzy c-means segmentation as applied to the quantitative tissue maps. MRI datasets were segmented using a validated template-based method. Statistical comparisons included Pearson correlation (r), intraclass correlation coefficient (ICC), and Bland-Alt-man analysis.

Results:

QT and MRI measurements demonstrated strong correlations for both FGV (r = 0.893) and FGR (r = 0.915, p < 0.0001). Mean biases were −13.2 cm³ for FGV and +12.7% for FGR. Regression slopes were 0.63 and 1.35, respectively, with significant proportional bias for both (p < 0.0001). Systematic differences were consistent with modality-specific contrast mechanisms and field-of-view definitions; the strong correlation and stable, predictable bias support quantitative concordance between the modalities.

Conclusion:

QT technology provides accurate, reproducible, and volumetric quantification of breast fibroglandular tissue, demonstrating strong concordance with MRI. These results supported FDA clearance of the QT FGR feature, establishing it as a validated quantitative imaging biomarker for clinical breast density assessment.

Introduction

Accurate quantification of breast fibroglandular tissue volume (FGV) and the proportion of fibroglandular tissue relative to total breast volume (TBV) plays a critical role in breast cancer risk assessment, diagnostic interpretation, and individualized patient management [1]. Numerous studies have established the relationship between dense breast tissue, characterized by a higher proportion of fibroglandular tissue, and an elevated risk of breast cancer [2]. Traditional mammographic density measurements, though widely used, provide only two-dimensional projections and are known to be limited by variability in tissue compression, projection angles, and the effects of overlapping structures [1]. These limitations have prompted the medical imaging community to seek more accurate and reproducible three-dimensional (3D) volumetric approaches [3].

Magnetic resonance imaging (MRI) has been considered as a relatively more established modality for quantifying breast density and related volumetric measures of breast parenchyma. MRI-based volumetric breast density measurements have demonstrated high accuracy and reproducibility, facilitating improved risk stratification and potentially guiding preventive strategies and screening protocols. Notably, MRI can produce detailed 3D anatomical representations, capturing subtle variations in tissue composition and distribution [4]. This capability enables the generation of reliable volumetric maps of fibroglandular tissue within the breast and has set a benchmark for other imaging modalities seeking to offer quantitative assessments [1,4,5].

However, despite the recognized utility of MRI, its adoption as a routine screening or diagnostic tool for breast density assessment faces practical challenges. High costs, limited availability, long examination times, and patient contraindications to MRI contrast agents or the MRI environment itself can limit its application [6,7]. In this context, there is a growing need for alternative imaging methods that combine safety, accuracy, practicality, and accessibility. Ultrasound-based imaging, particularly in the form of transmission ultrasound, offers a promising avenue. Compared to conventional ultrasound reflection imaging, transmission ultrasound does not rely on backscatter echoes; instead, it measures how sound waves pass through the breast, thereby potentially providing more homogeneous and quantitative maps of tissue composition [8,9]. Such an approach could maintain the strengths of MRI-based volumetric quantification, namely, the generation of accurate and reproducible 3D breast density metrics, while offering a more accessible, cost-effective, and patient-friendly platform.

QT Imaging’s dedicated breast imaging system, utilizing transmission ultrasound technology has shown potential in generating consistent 3D breast imaging volumes suitable for quantitative tissue characterization [8,10,11]. Previous feasibility studies suggest that the system can provide volumetric data that is correlated with breast tissue composition [12]. By quantifying FGV and TBV, and subsequently calculating their ratio, this emerging modality offers a robust surrogate and practical tool for clinical breast density assessment.

The purpose of this manuscript is to evaluate the capability of QT Imaging’s ultrasound technology to accurately measure breast FGV and the ratio of FGV to TBV. In doing so, we aim to compare these ultrasound-derived metrics with equivalent measures obtained through established MRI protocols. Demonstrating concordance between ultrasound and MRI measurements would signify a critical step toward offering a non-ionizing, more accessible imaging solution for routine clinical breast density assessment and risk evaluation. By rigorously validating ultrasound-based measurements with MRI, a recognized reference standard in volumetric breast density quantification, we seek to advance the field toward more widespread adoption of quantitative breast imaging techniques that are both clinically feasible and economically sustainable.

Materials and methods

Study overview and design

This study was designed to quantitatively compare breast FGV and volumetric breast density, defined as fibroglandular ratio (FGR) and is the ratio of FGV to TBV (i.e., FGR = FGV/TBV), as derived from QT images and from MRI. MRI-based measurements served as the reference standard, representing the accepted volumetric ground truth for assessing breast density. The analysis was conducted retrospectively on previously acquired paired QT and MRI datasets obtained as part of the prospective ACCRUE Study (ClinicalTrials.gov identifier NCT03052166), which was performed at MemorialCare Long Beach Medical Center, Long Beach, CA. The study protocol was approved by the Western Institutional Review Board (WIRB), and written informed consent was obtained from all participants prior to enrollment.

Of the 40 ACCRUE participants with available paired QT and MRI data for analyses, a total of 53 breast datasets from 29 female participants were included. These represented study participants for whom paired QT and MRI examinations of the same breast were available within 30 days; MRI was performed in only a subset of the overall ACCRUE cohort, and the only exclusions were cases in which QT or MRI image quality or algorithmic failure precluded reliable calculation of fibroglandular volume or breast density. None of the images analyzed were used in FGR algorithm development or training. Prior analyses of the QT case-collection cohort have been reported previously [13,14].

Imaging systems and acquisition protocols

All QT scans were performed using the QT Scanner 2000 Model A (QT Imaging, Inc., Novato, CA). The system is a multimodal platform that reconstructs three-dimensional, quantitative tomographic maps of the breast’s acoustic properties, specifically speed of sound (SOS), attenuation, and reflection. For each QT scan, the breast was scanned while suspended in a temperature-controlled water bath to provide acoustic coupling. Figure 1 shows a patient positioned prone on the QT scanner. A transmitter-receiver array rotated 360° about the breast, acquiring projection data at 2° increments per rotation level. The complete dataset was reconstructed using a nonlinear, full-wave 3D inverse scattering algorithm previously described [15]. The resulting SOS maps provide voxel-wise quantitative measurements that directly reflect tissue composition, with fibroglandular tissue exhibiting higher SOS values relative to adipose tissue. In addition to transmission imaging, the system simultaneously acquires reflection (pulse-echo) data using three transceiver arrays of varying focal lengths, producing refraction-corrected B-mode-like reflection images that are spatially compounded to co-register with and complement the SOS maps.

Figure 1.

(Left) Patient positioned prone on the breast acoustic CT system during image acquisitions; (right) acquisition geometry depicting transmitter–receiver configuration.

Breast MR imaging was performed using a Siemens Magnetom Aera 1.5T scanner equipped with a dedicated bilateral breast coil. Axial, non-fat-suppressed T1-weighted sequences were acquired to maintain intensity contrast between fibroglandular and fatty tissues. Acquisition parameters included slice thickness of 2 mm, field of view 320 × 320 mm², matrix 384 × 384, and TR/TE of approximately 650/9 ms. MRI datasets were processed using a fully automated template-based segmentation algorithm which excludes the chest wall and muscular structures by registering a chest template to each subject’s anatomy and then applying clustering and morphological post-processing to delineate the breast volume and fibroglandular content[16].

Image segmentation and quantitative analysis

The QT-based segmentation was performed using a fuzzy c-means (FCM) clustering algorithm [11]. Briefly, attenuation images were used to isolate the breast from the water bath, after which the breast volume was segmented into fibroglandular and adipose compartments based on SOS distributions. The FCM algorithm partitions the voxel space X = {x₁, …, x_n} into two clusters c₁, c₂ by minimizing the objective function,

$$J=\sum _{i=1}^n\sum _{j=1}^2{w}_{ij}^m{\Vert x_i-c_j\Vert }^2$$

where x_i is the speed-of-sound value at voxel i and X = {x₁,…,x_n} is the set of breast voxels; c_j is the centroid (characteristic speed of sound) of cluster j, with c₁ and c₂ corresponding to adipose and fibroglandular tissue, respectively; w_ij is the fuzzy membership of voxel i in cluster j, subject to 0 ≤ w_ij ≤ 1 and Σ_j w_ij = 1; and m is the fuzzifier exponent (set to m = 2). Fuzzy c-means was selected because the speed-of-sound values of adipose and fibroglandular tissue vary between patients; estimating per-case cluster centroids is therefore more robust than applying fixed literature thresholds." The resulting membership map was binarized at an empirically optimized threshold to yield a fibroglandular tissue mask. Skin voxels were excluded using a boundary-detection heuristic based on the attenuation profile. The QT-TBV was computed as the sum of all non-skin breast voxels, and the QT-FGV as the sum of voxels classified as fibroglandular. The volumetric density metric (QT-FGR = QT-FGV/QT-TBV) was then calculated for each breast. This segmentation workflow resulting in calculation of QT-FGV and QT-FGR was fully automated and required no manual operator input or intervention. The algorithm workflow and a representative segmented image is shown in Figure 2.

Figure 2.

QT segmentation workflow for fibroglandular ratio (FGR) computation. Left: processing pipeline. The breast is first segmented from the surrounding water bath using attenuation images; boundary ("border") pixels are identified by proximity; high-speed (fibroglandular) tissue is separated from the total breast volume by fuzzy c-means (FCM) clustering into two categories, with membership-map generation and thresholding; skin is then removed; and FGR is calculated as the ratio of fibroglandular volume to total breast volume. Right: representative results for three breasts shown in axial, coronal, and sagittal views (columns). Top row, speed-of-sound (SoS) images; middle row, segmented total breast volume with the fibroglandular compartment delineated; bottom row, the final fibroglandular tissue mask after skin removal.

For the reference MRI data, the template-based chest-exclusion segmentation was applied to define the breast volume. Within this region, fibroglandular tissue was segmented using intensity-based clustering combined with morphological filtering. This method has been validated in prior studies, demonstrating a mean absolute error of <3% compared to radiologist-corrected ground truth [16]. The MR-derived volume (MR-FGV) and breast density metrics (MR-FGR) served as the ground-truth comparator for statistical evaluation.

Statistical analysis

The primary analyses quantified both relative correlation and absolute agreement between QT- and MR-derived measures of FGV and FGV/TBV. The Pearson correlation coefficient (r) was used to assess linear association, while the intraclass correlation coefficient (ICC, two-way mixed-effects, absolute-agreement model) quantified the degree of absolute concordance. Statistical significance was defined at p < 0.05 with 95% confidence intervals (CI) computed for all correlation metrics. Agreement bias and variability were visualized using Bland–Altman plots, with limits of agreement defined as mean difference ± 1.96 standard deviations[17]. The direction of all differences was defined as QT – MRI. Proportional bias was assessed by linear regression of the Bland–Altman differences against their means; a slope significantly different from zero was taken as evidence of proportional bias.

All analyses were performed using GraphPad Prism v10.6.1 (GraphPad Software, San Diego, CA). For consistency with previous breast density validation studies, results are reported both in absolute units (cm³ for volume) and relative units (% for FGV/TBV).

Results

A total of 53 breast image sets from 29 participants were included for quantitative comparison between QT- and MR-derived metrics. Each case had complete paired volumetric data, including FGV and the derived fibroglandular ratio FGR. Representative sets of QT and MR images from the same patients are shown in Figures 3 and 4.

Figure 3.

(a) Non–fat-saturated T1-weighted MRI demonstrating predominantly fatty parenchyma; (b–d) Corresponding coronal, sagittal, and axial QT scan generated speed-of-sound images of the right breast, with minimal areas of higher sound speed corresponding to sparse fibroglandular tissue. Speed-of-sound is displayed on a grayscale, indicated by a colorbar, spanning 1360–1640 m/s, with brighter pixels indicating higher speed of sound.

Figure 4.

(a) Non–fat-saturated T1-weighted MRI demonstrating predominantly fibroglandular representative of a relatively dense breast; (b–d) Corresponding coronal, sagittal, and axial quantitative transmission (QT) speed-of-sound images of the right breast show diffusely elevated sound speed values consistent with dense fibroglandular tissue Speed-of-sound is displayed on a grayscale, indicated by a colorbar, spanning 1360–1640 m/s, with brighter pixels indicating higher speed of sound.

The cohort exhibited a wide distribution of breast compositions, with MRI-derived FGV ranging from 20.2 to 453.3 cm³ (mean ± SD = 132.2 ± 88.6 cm³) and QT-derived FGV ranging from 36.2 to 374.2 cm³ (mean = 119.0 ± 62.6 cm³). MRI-based FGR values spanned 2.0%–48.4% (mean = 15.6 ± 11.2%), while QT-based FGR ranged from 8.7%–72.4% (mean = 28.3 ± 16.6%). MRI-derived total breast volume ranged 379.3–2001.0 cm³ (mean ± SD = 956.4 ± 417.2 cm³) and QT-derived total breast volume ranged 79.4–1492.0 cm³ (mean = 502.8 ± 267.1 cm³). QT thus retained, on average, 90% of MRI fibroglandular volume but only 53% of MRI total breast volume; of the mean 454 cm³ difference in total breast volume, approximately 97% was non-fibroglandular tissue.

A strong linear association was observed between QT- and MRI-derived FGV values (Pearson r = 0.893, p < 0.0001; 95% CI = 0.84–0.94). The regression of QT-FGV on MR-FGV yielded a slope = 0.63 ± 0.06 and intercept = 35.7 cm³, indicating that QT systematically underestimates fibroglandular volume at higher glandular levels while maintaining strong proportionality (Figure 5).

Figure 5.

Scatterplot of QT- vs. MRI-derived FGV. The solid line shows the fitted linear regression (r = 0.893); dotted lines indicate the 95% confidence interval of the fit.

The mean bias (QT–MRI) was −13.2 cm³, showing that QT FGVs were on average 10%–15% lower than MRI-based values. Error analysis demonstrated RMSE = 44.7 cm³ and MAE = 35.2 cm³, corresponding to approximately 9% of the full volumetric dynamic range. Despite this slight offset, the strong correlation and low dispersion confirm that the QT reconstruction accurately quantifies glandular volume across the full physiologic range. A significant negative proportional bias was present for FGV (regression of the QT – MRI difference on the mean: slope = −0.36, 95% CI −0.49 to −0.23, p < 0.0001), consistent with the regression slope below unity and indicating progressively greater underestimation at larger fibroglandular volumes.

Cross-modality comparison of the fibroglandular ratio showed an even stronger relationship (r = 0.915, p < 0.0001; 95% CI = 0.87–0.95). Regression of QT-derived FGR against MRI-derived values produced a slope = 1.35 ± 0.10 and intercept = 7.25%, reflecting a consistent upward bias in QT-based estimates (Figure 6). The mean bias (QT–MRI) was +12.7% ± 7.8%, confirming that QT tends to yield higher FGR values than MRI. Because FGR is a ratio, a uniform field-of-view difference would scale numerator and denominator together and largely cancel; such cancellation holds only when the excluded tissue is of average breast composition [1]. That condition is not met here: the volume excluded by the QT field of view relative to MRI was almost entirely non-fibroglandular (≈97%), so the predominantly adipose peripheral tissue reduces the total-breast-volume denominator far more than the fibroglandular numerator, elevating the ratio. Differences in segmentation definition between SoS-based fuzzy c-means classification and MRI intensity-based templating may also contribute. The present analysis does not isolate the relative contribution of these factors. A significant positive proportional bias was present for FGR (regression of the QT—MRI difference on the mean: slope = +0.40, 95% CI: 0.29 to 0.51, p < 0.0001), indicating a larger offset in denser breasts. Error metrics for FGR were RMSE = 14.8% and MAE = 12.7%, indicating tight variance around the regression fit.

Figure 6.

Scatterplot of QT- vs. MRI-derived fibroglandular ratio (FGR). The solid line shows the fitted linear regression (r = 0.915); dotted lines indicate the 95% confidence interval of the fit.

Bland–Altman analysis (Figure 7) quantified inter-modality agreement. For FGV, the mean difference = −13.2 cm³, with 95% limits of agreement (LoA) approximately −99 cm³ to +73 cm³, consistent with mild systematic underestimation and moderate volumetric dispersion. For FGR, the mean difference = +12.7%, with LoA spanning roughly −3% to +29%, confirming a consistent positive offset of QT-based ratios across the density spectrum. Consistent with the regression analyses, the FGR offset increased significantly with breast density and FGV showed a significant negative proportional trend; inter-modality differences should therefore be interpreted as proportional across the measurement range rather than constant. Absolute agreement between QT- and MR-derived fibroglandular ratio values was also high (ICC, two-way mixed-effects, absolute agreement = 0.85; 95% CI: 0.75–0.91; p < 0.0001), consistent with the observed strong correlation and stable inter-modality bias.

Figure 7.

Bland–Altman plots for (a) FGV and (b) FGR differences between QT and MRI.

Observed effect sizes were large for both FGV and FGR (Cohen’s q = 1.26 and 1.48, respectively). Post-hoc power analysis confirmed > 0.999 achieved power (α = 0.05, two-tailed) for detecting correlations exceeding 0.8 with N = 53. Thus, the study was sufficiently powered to characterize the quantitative correspondence between QT and MRI.

These results demonstrate that QT technology provides quantitative fibroglandular ratio measurements that are highly correlated with MR-based ground truth. The systematic offsets are consistent with differences in field-of-view definition and modality-specific segmentation, and their stability across the cohort supports QT’s use as a non-ionizing, volumetric modality for assessing breast composition and density objectively.

Discussion

This study demonstrates that QT technology provides reliable and reproducible volumetric estimates of FGV and FGR, showing strong concordance with MRI, the current research standard for volumetric breast density assessment. The high linear correlations observed for both FGV (r = 0.893) and FGR (r = 0.915) confirm that QT scan captures volumetric tissue composition with quantitative fidelity comparable to MRI. At the same time, the systematic offsets observed between the two modalities: QT slightly underestimating FGV and overestimating FGR, must be interpreted within the context of fundamental differences in imaging physics, field-of-view, and segmentation definitions, rather than as simple measurement errors.

MRI and QTI technologies are fundamentally distinct in their contrast mechanisms. MRI signal intensity arises from proton relaxation phenomena modulated by water and fat content, with fibroglandular regions appearing hyperintense on non–fat-suppressed T1-weighted sequences. QTI technology, by contrast, reconstructs three-dimensional speed-of-sound (SOS) maps where contrast derives from differences in acoustic propagation properties rather than proton density or relaxation effects. Consequently, tissue classification boundaries depend on different intrinsic physical properties: dielectric and magnetic in MRI, and acoustic in QTI, and therefore are not expected to yield numerically identical volume fractions. Thus, the modest differences in absolute FGV or FGR are better interpreted as modality-dependent quantitative signatures reflecting complementary biological contrasts than as simple over- or underestimation, although the present study cannot fully separate these contributions.

Differences in field-of-view (FOV) geometry also contribute to observed offsets. MRI’s definition of “total breast volume” varies with chest-wall exclusion, axillary tail inclusion, and superior-inferior coverage, leading to study-to-study variability influenced by coil geometry, positioning, and body habitus. In contrast, QTI employs a fixed, mechanically defined FOV in which the entire region suspended in the water bath is consistently reconstructed, reducing operator dependence. Although extreme breast sizes may cause peripheral truncation, QTI’s FOV is mechanically deterministic whereas MRI’s is anatomically interpretive. Quantitatively, QTI retained 90% of MRI fibroglandular volume but only 53% of MRI total breast volume, indicating that the truncated tissue is predominantly peripheral fat; because this excluded tissue is far from average breast composition, the ratio is not protected from the field-of-view difference in the way a uniform truncation would allow [1]. Consistency of correlation, together with a stable and explainable offset, rather than absolute numerical equivalence, therefore serves as the appropriate validation criterion.

The FGR metric, representing the fraction of fibroglandular to total breast volume, is widely recognized as a biophysical correlate of mammographic density and thus as an independent risk factor for breast cancer. Historically, breast density has been quantified using two-dimensional projection mammography, where the apparent area of radiopaque tissue is visually categorized by a radiologist or computed through automated volumetric models. Recent advances in artificial intelligence (AI) have enabled automated estimation of fibroglandular tissue volume and breast density from mammographic images, improving reproducibility compared with visual assessment while remaining inherently projection-based and model-dependent [18]. Nevertheless, because mammography is inherently two-dimensional, its density assessment conflates tissue overlap with actual composition. Volumetric imaging methods such as MRI and now QT scan enable true three-dimensional quantification, free from superimposition effects.

It is worth emphasizing that even Volpara, the most widely used and FDA-cleared automated mammographic density software, derived its validation framework using MRI-based fibroglandular volume as ground truth during its original development and clearance process. Although Volpara estimates volumetric density from 2D x-ray projections, its calibration curves and internal modeling were built upon MRI comparisons to establish proportional accuracy. Thus, using MRI as a reference for QTI’s validation aligns with the precedent set by prior density technologies. QTI’s performance, showing correlations comparable to those reported in mammography, MRI validation studies, supports its credibility as a modality capable of providing objective, quantitative assessments of breast composition.

Importantly, the findings presented in this study formed the clinical validation dataset used in the regulatory submission and FDA clearance of the QT-derived FGV and FGR features. The analysis provided quantitative evidence of equivalence between QT- and MR-derived fibroglandular ratio values, demonstrating both statistical robustness and physiological plausibility of QT-derived density metrics. These results, together with earlier phantom data, supported the feature’s recognition by the FDA as a quantitative imaging biomarker suitable for clinical use. This regulatory milestone underscores that QTI’s FGR measurement satisfies not only scientific validity but also the rigorous performance and reliability standards required for medical device clearance.

Although the present study addresses the quantification of breast composition rather than cancer detection, the clinical motivation for accurate volumetric density warrants brief context. Elevated fibroglandular content is an established risk factor and also reduces mammographic sensitivity, which declines from approximately 93% in fatty breasts to as low as 30% in extremely dense breasts; reproducible volumetric density is therefore valuable for identifying women who may benefit from supplemental screening. Current density-based pathways employ supplemental whole-breast ultrasound, contrast-enhanced mammography (CEM), and MRI [19]. Beyond morphologic detection, quantitative ultrasound techniques such as shearwave and strain elastography characterize tissue stiffness for lesion work-up [20,21,22], and opportunistic assessment of breast density on routine chest CT has also been described [23]. QTI’s transmission tomography is complementary to these methods: whereas elastography and contrast techniques target lesion characterization and detection, QTI provides operator-independent volumetric quantification of tissue composition, and future work may combine compositional and mechanical biomarkers. The accurate quantification of fibroglandular tissue provided by QTI is also relevant to evolving multimodal density-classification frameworks: ultrasound-specific tissue composition criteria, formalized in the ACR BI-RADS 5th edition ultrasound lexicon [24] and reinforced in the recent DEGUM Best Practice Guidelines for breast ultrasound [25], now provide a structured basis for cross-modality density reporting, while AI-based tools have further emerged as a means of automating BI-RADS density classification directly from ultrasound images [26,27]. QT-derived FGR, as a fully automated, operator-independent volumetric output, is well-positioned to contribute to such multimodal frameworks; its concordance with ultrasound-specific BI-RADS tissue composition categories and integration with AI-driven density pipelines are subjects for future investigation. Lesion detection, axillary nodal assessment, and the relationship of QT-derived FGR to clinical risk categories were outside the scope of this validation study and are the subject of ongoing work.

Several limitations of the present study should be acknowledged. First, while MRI serves as the accepted research standard for volumetric breast density, it is not a true “ground truth” in the physical sense, its segmentation accuracy is itself dependent on intensity thresholds, coil sensitivity profiles, and operator-defined chest-wall boundaries. Thus, some of the inter-modality differences may reflect segmentation philosophy rather than measurement error. Second, the field-of-view mismatch between MRI and QT remains a key geometric limitation. QT imaging defines the total breast as the entire imaged volume within the coupling bath, whereas MRI acquisitions can variably include portions of the axillary tail or chest wall depending on coil setup and breath-hold reproducibility. Third, this analysis used data obtained retrospectively with matched QTI and MRI scans but without direct co-registration; small orientation or deformation differences may influence volumetric comparison. Although the confidence intervals were narrow (FGV:0.82–0.94; FGR: 0.86–0.95; ICC: 0.75–0.91), indicating adequate precision for the agreement objective, this single-institution cohort warrants confirmation in larger multi-center, multi-platform studies. Finally, this study did not explore the relationship of QT-derived FGR to clinical risk or mammographic categories, which will be critical for translating quantitative metrics into prognostic and screening frameworks.

Conclusions

The present study provides strong evidence that QTI technology can reproducibly quantify breast composition in a manner consistent with MRI-based metrics. Although differences in absolute values exist, these are consistent with expected physical and geometric distinctions between the modalities. By demonstrating quantitative correspondence with MRI while retaining advantages of cost, safety, and accessibility, QTI technology establishes itself as a promising tool for objective, volumetric breast density assessment and risk stratification in clinical practice.