Variability of Breast Density Classification Between US and UK Radiologists

Enhancing Accuracy in Breast Density Assessment Using Deep Learning: A Multicentric, Multi-Reader Study

The evaluation of mammographic breast density, a critical indicator of breast cancer risk, is traditionally performed by radiologists via visual inspection of mammography images, utilizing the Breast Imaging-Reporting and Data System (BI-RADS) breast density categories. However, this method is subject to substantial interobserver variability, leading to inconsistencies and potential inaccuracies in density assessment and subsequent risk estimations. To address this, we present a deep learning-based automatic detection algorithm (DLAD) designed for the automated evaluation of breast density. Our multicentric, multi-reader study leverages a diverse dataset of 122 full-field digital mammography studies (488 images in CC and MLO projections) sourced from three institutions. We invited two experienced radiologists to conduct a retrospective analysis, establishing a ground truth for 72 mammography studies (BI-RADS class A: 18, BI-RADS class B: 43, BI-RADS class C: 7, BI-RADS class D: 4). The efficacy of the DLAD was then compared to the performance of five independent radiologists with varying levels of experience. The DLAD showed robust performance, achieving an accuracy of 0.819 (95% CI: 0.736-0.903), along with an F1 score of 0.798 (0.594-0.905), precision of 0.806 (0.596-0.896), recall of 0.830 (0.650-0.946), and a Cohen's Kappa (κ) of 0.708 (0.562-0.841). The algorithm achieved robust performance that matches and in four cases exceeds that of individual radiologists. The statistical analysis did not reveal a significant difference in accuracy between DLAD and the radiologists, underscoring the model's competitive diagnostic alignment with professional radiologist assessments. These results demonstrate that the deep learning-based automatic detection algorithm can enhance the accuracy and consistency of breast density assessments, offering a reliable tool for improving breast cancer screening outcomes.

Intraobserver and Interobserver Reproducibility of Breast Diffusion-Weighted Imaging Quantitative Parameters: Readout-Segmented vs. Single-Shot Echo-Planar Imaging

BACKGROUND

The recommended technique for breast diffusion-weighted imaging (DWI) acquisitions is not sufficiently standardized in clinical practice.

PURPOSE

To investigate the intraobserver and interobserver reproducibility of DWI measurements, diffusion-kurtosis imaging (DKI) parameters, and image quality evaluation in breast lesions between single-shot echo-planar imaging (ss-EPI) and readout-segmented echo-planar imaging (rs-EPI).

STUDY TYPE

Prospective.

POPULATION

A total of 295 women with 209 malignant and 86 benign breast lesions.

FIELD STRENGTH/SEQUENCE

A 3-T; fat-saturated T2-weighted MR imaging (T2WI); multi-b-value DWI with both ss-EPI and rs-EPI readouts; T1-weighted dynamic contrast-enhanced MRI (DCE-MRI).

ASSESSMENT

Mean kurtosis (MK), mean diffusion (MD), and apparent diffusion coefficient (ADC) values were measured for each lesion on ss-EPI and rs-EPI, respectively. Image quality was visually evaluated regarding image sharpness, geometric distortion, lesion conspicuity, visualization of anatomic structures, and overall quality. Quantitative and qualitative analyses were performed twice with a time interval of 2 weeks.

STATISTICAL TESTS

Intraobserver and interobserver reproducibility were evaluated using intra-class correlation coefficients (ICC), within-subject coefficient of variation (wCV), and Bland-Altman plots.

RESULTS

MK, MD, and ADC quantitative parameters for breast lesions showed excellent intraobserver and interobserver reproducibility, with ICCs >0.75 and wCV values ranging from 2.51% to 7.08% for both sequences. The wCV values in both intraobserver and interobserver measurements were higher in the ss-EPI sequence (3.63%-7.08%) than that of the rs-EPI sequence (2.51%-3.62%). The wCV values differed in subgroups with different histopathological types of lesions, breast density, lesion morphology, and lesion sizes, respectively. Furthermore, rs-EPI (ICCs, 0.76-0.97; wCV values, 2.41%-6.04%) had better intraobserver and interobserver reproducibility than ss-EPI (ICCs, 0.54-0.90; wCV values, 6.18%-13.69%) with regard to image quality.

DATA CONCLUSION

Compared to the ss-EPI, the rs-EPI sequence showed higher intraobserver and interobserver reproducibility for quantitative diffusion-related parameters and image quality assessments measured in breast DWI and DKI.

EVIDENCE LEVEL

2.

TECHNICAL EFFICACY

Stage 2.

[Breast density assessment and organised breast cancer screening]

INTRODUCTION

The objective of this study was to evaluate the intra- and inter-rater agreement of radiologists regarding the evaluation of breast density.

METHODOLOGY

Breast density assessments of 120 cases were performed by four radiologists in the city of Ouagadougou according to the fifth edition of the American College of Radiology BI-RADS atlas. Cohen's weighted kappa coefficients and Fleiss kappa coefficients were used to estimate agreement between observers and with a panel of three experts radiologists. A new evaluation of the 120 cases was performed by all raters one month after the initial evaluation.

RESULTS

Inter-rater kappa coefficients ranged from 0.55 to 0.74. The Fleiss kappa coefficient was 0.58, 0.43, 0.41, and 0.43 for categories A, B, C, and D respectively. In terms of classification into "sparse breasts" and "dense breasts", the kappa coefficients ranged from 0.47 to 0.67. Taking the results of the expert panel as a reference, the proportion of false positives in the diagnosis "dense breasts" ranged from 18.6% to 26.8%. Intraobserver agreement was good.

CONCLUSION

Our study showed moderate to good intra- and inter-raters agreements. Upgrading and harmonisation of practices will be used to empower radiologists to participate in organised breast cancer screening in Burkina Faso.

Mammography biomarkers of cardiovascular and musculoskeletal health: A review

Breast density (BD) and breast arterial calcifications (BAC) can expand the role of mammography. In premenopause, BD is related to body fat composition: breast adipose tissue and total volume are potential indicators of fat storage in visceral depots, associated with higher risk of cardiovascular disease (CVD). Women with fatty breast have an increased likelihood of hypercholesterolemia. Women without cardiometabolic diseases with higher BD have a lower risk of diabetes mellitus, hypertension, chest pain, and peripheral vascular disease, while those with lower BD are at increased risk of cardiometabolic diseases. BAC, the expression of Monckeberg sclerosis, are associated with CVD risk. Their prevalence, 13 % overall, rises after menopause and is reduced in women aged over 65 receiving hormonal replacement therapy. Due to their distinct pathogenesis, BAC are associated with hypertension but not with other cardiovascular risk factors. Women with BAC have an increased risk of acute myocardial infarction, ischemic stroke, and CVD death; furthermore, moderate to severe BAC load is associated with coronary artery disease. The clinical use of BAC assessment is limited by their time-consuming manual/visual quantification, an issue possibly solved by artificial intelligence-based approaches addressing BAC complex topology as well as their large spectrum of extent and x-ray attenuations. A link between BD, BAC, and osteoporosis has been reported, but data are still inconclusive. Systematic, standardised reporting of BD and BAC should be encouraged.

Assessing breast density using the chemical-shift encoding-based proton density fat fraction in 3-T MRI

Objectives

There is a clinical need for a non-ionizing, quantitative assessment of breast density, as one of the strongest independent risk factors for breast cancer. This study aims to establish proton density fat fraction (PDFF) as a quantitative biomarker for fat tissue concentration in breast MRI and correlate mean breast PDFF to mammography.

Methods

In this retrospective study, 193 women were routinely subjected to 3-T MRI using a six-echo chemical shift encoding-based water-fat sequence. Water-fat separation was based on a signal model accounting for a single T2* decay and a pre-calibrated 7-peak fat spectrum resulting in volumetric fat-only, water-only images, PDFF- and T2*-values. After semi-automated breast segmentation, PDFF and T2* values were determined for the entire breast and fibroglandular tissue. The mammographic and MRI-based breast density was classified by visual estimation using the American College of Radiology Breast Imaging Reporting and Data System categories (ACR A-D).

Results

The PDFF negatively correlated with mammographic and MRI breast density measurements (Spearman rho: −0.74, p < .001) and revealed a significant distinction between all four ACR categories. Mean T2* of the fibroglandular tissue correlated with increasing ACR categories (Spearman rho: 0.34, p < .001). The PDFF of the fibroglandular tissue showed a correlation with age (Pearson rho: 0.56, p = .03).

Conclusion

The proposed breast PDFF as an automated tissue fat concentration measurement is comparable with mammographic breast density estimations. Therefore, it is a promising approach to an accurate, user-independent, and non-ionizing breast density assessment that could be easily incorporated into clinical routine breast MRI exams.

Key Points

• The proposed PDFF strongly negatively correlates with visually determined mammographic and MRI-based breast density estimations and therefore allows for an accurate, non-ionizing, and user-independent breast density measurement.

• In combination with T2*, the PDFF can be used to track structural alterations in the composition of breast tissue for an individualized risk assessment for breast cancer.

A deep learning framework to classify breast density with noisy labels regularization

BACKGROUND AND OBJECTIVE

Breast density assessed from digital mammograms is a biomarker for higher risk of developing breast cancer. Experienced radiologists assess breast density using the Breast Image and Data System (BI-RADS) categories. Supervised learning algorithms have been developed with this objective in mind, however, the performance of these algorithms depends on the quality of the ground-truth information which is usually labeled by expert readers. These labels are noisy approximations of the ground truth, as there is often intra- and inter-reader variability among labels. Thus, it is crucial to provide a reliable method to obtain digital mammograms matching BI-RADS categories. This paper presents RegL (Labels Regularizer), a methodology that includes different image pre-processes to allow both a correct breast segmentation and the enhancement of image quality through an intensity adjustment, thus allowing the use of deep learning to classify the mammograms into BI-RADS categories. The Confusion Matrix (CM) - CNN network used implements an architecture that models each radiologist's noisy label. The final methodology pipeline was determined after comparing the performance of image pre-processes combined with different DL architectures.

METHODS

A multi-center study composed of 1395 women whose mammograms were classified into the four BI-RADS categories by three experienced radiologists is presented. A total of 892 mammograms were used as the training corpus, 224 formed the validation corpus, and 279 the test corpus.

RESULTS

The combination of five networks implementing the RegL methodology achieved the best results among all the models in the test set. The ensemble model obtained an accuracy of (0.85) and a kappa index of 0.71.

CONCLUSIONS

The proposed methodology has a similar performance to the experienced radiologists in the classification of digital mammograms into BI-RADS categories. This suggests that the pre-processing steps and modelling of each radiologist's label allows for a better estimation of the unknown ground truth labels.

Development and Validation of an AI-driven Mammographic Breast Density Classification Tool Based on Radiologist Consensus

Mammographic breast density (BD) is commonly visually assessed using the Breast Imaging Reporting and Data System (BI-RADS) four-category scale. To overcome inter- and intraobserver variability of visual assessment, the authors retrospectively developed and externally validated a software for BD classification based on convolutional neural networks from mammograms obtained between 2017 and 2020. The tool was trained using the majority BD category determined by seven board-certified radiologists who independently visually assessed 760 mediolateral oblique (MLO) images in 380 women (mean age, 57 years ± 6 [SD]) from center 1; this process mimicked training from a consensus of several human readers. External validation of the model was performed by the three radiologists whose BD assessment was closest to the majority (consensus) of the initial seven on a dataset of 384 MLO images in 197 women (mean age, 56 years ± 13) obtained from center 2. The model achieved an accuracy of 89.3% in distinguishing BI-RADS a or b (nondense breasts) versus c or d (dense breasts) categories, with an agreement of 90.4% (178 of 197 mammograms) and a reliability of 0.807 (Cohen κ) compared with the mode of the three readers. This study demonstrates accuracy and reliability of a fully automated software for BD classification. Keywords: Mammography, Breast, Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms Supplemental material is available for this article. © RSNA, 2022.

Comparison of breast density assessment between human eye and automated software on digital and synthetic mammography: Impact on breast cancer risk

PURPOSE

To evaluate the agreement between automatic assessment software of breast density based on artificial intelligence (AI) and visual assessment by a senior and a junior radiologist, as well as the impact on the assessment of breast cancer risk (BCR) at 5 years.

MATERIALS AND METHODS

We retrospectively included 311 consecutive women (mean age, 55.6±8.5 [SD]; range: 40-74 years) without a personal history of breast cancer who underwent routine mammography between January 1, 2019 and February 28, 2019. Mammographic breast density (MBD) was independently evaluated by a junior and a senior reader on digital mammography (DM) and synthetic mammography (SM) using BI-RADS (5th edition) and by an AI software. For each MBD, BCR at 5 years was estimated per woman by the AI software. Interobserver agreement for MBD between the two readers and the AI software were evaluated by quadratic κ coefficients. Reproducibility of BCR was assessed by intraclass correlation coefficient (ICC).

RESULTS

Agreement for MBD assessment on DM and SM was almost perfect between senior and junior radiologists (κ=0.88 [95% CI: 0.84-0.92] and κ=0.86 [95% CI: 0.82-0.90], respectively) and substantial between the senior radiologist and AI (κ=0.79; 95% CI: 0.73-0.84). There was substantial agreement between DM and SM for the senior radiologist (κ=0.79; 95% CI: 0.74-0.84). BCR evaluation at 5 years was highly reproducible between the two radiologists on DM and SM (ICC=0.98 [95% CI: 0.97-0.98] for both), between BCR evaluation based on DM and SM evaluated by the senior (ICC=0.96; 95% CI: 0.95-0.97) or junior radiologist (ICC=0.97; 95% CI: 0.96-0.98) and between the senior radiologist and AI (ICC=0.96; 95% CI: 0.95-0.97).

CONCLUSION

This preliminary study demonstrates a very good agreement for BCR evaluation based on the evaluation of MBD by a senior radiologist, junior radiologist and AI software.

Staying abreast of imaging - Current status of breast cancer detection in high density breast

OBJECTIVES

The aim of this paper is to illustrate the current status of imaging in high breast density as we enter a new decade of advancing medicine and technology to diagnose breast lesions.

KEY FINDINGS

Early detection of breast cancer has become the chief focus of research from governments to individuals. However, with varying breast densities across the globe, the explosion of breast density information related to imaging, phenotypes, diet, computer aided diagnosis and artificial intelligence has witnessed a dramatic shift in new screening recommendations in mammography, physical examination, screening younger women and women with comorbid conditions, screening women at high risk, and new screening technologies. Breast density is well known to be a risk factor in patients with suspected/known breast neoplasia. Extensive research in the field of qualitative and quantitative analysis on different tissue characteristics of the breast has rapidly become the chief focus of breast imaging. A summary of the available guidelines and modalities of breast imaging, as well as new emerging techniques under study that can potentially provide an augmentation or even a replacement of those currently available.

CONCLUSION

Despite all the advances in technology and all the research directed towards breast cancer, detection of breast cancer in dense breasts remains a dilemma.

IMPLICATIONS FOR PRACTICE

It is of utmost importance to develop highly sensitive screening modalities for early detection of breast cancer.

Subjective Versus Quantitative Methods of Assessing Breast Density

In order to find a consistent, simple and time-efficient method of assessing mammographic breast density (MBD), different methods of assessing density comparing subjective, quantitative, semi-subjective and semi-quantitative methods were investigated. Subjective MBD of anonymized mammographic cases (n = 250) from a national breast-screening programme was rated by 49 radiologists from two countries (UK and USA) who were voluntarily recruited. Quantitatively, three measurement methods, namely VOLPARA, Hand Delineation (HD) and ImageJ (IJ) were used to calculate breast density using the same set of cases, however, for VOLPARA only mammographic cases (n = 122) with full raw digital data were included. The agreement level between methods was analysed using weighted kappa test. Agreement between UK and USA radiologists and VOLPARA varied from moderate (κw = 0.589) to substantial (κw = 0.639), respectively. The levels of agreement between USA, UK radiologists, VOLPARA with IJ were substantial (κw = 0.752, 0.768, 0.603), and with HD the levels of agreement varied from moderate to substantial (κw = 0.632, 0.680, 0.597), respectively. This study found that there is variability between subjective and objective MBD assessment methods, internationally. These results will add to the evidence base, emphasising the need for consistent, simple and time-efficient MBD assessment methods. Additionally, the quickest method to assess density is the subjective assessment, followed by VOLPARA, which is compatible with a busy clinical setting. Moreover, the use of a more limited two-scale system improves agreement levels and could help minimise any potential country bias.