Chapter 40 - GPU Acceleration of Iterative Digital Breast Tomosynthesis

Purpose:Digital breast tomosynthesis (DBT) has been shown to somewhat alleviate the breast tissue overlapping issues of two-dimensional (2D) mammography. However, the improvement in current DBT systems over mammography is still limited. Statistical image reconstruction (SIR) methods have the potential to reduce through-plane artifacts in DBT, and thus may be used to further reduce anatomical clutter. The purpose of this work was to study the impact of SIR on anatomical clutter in the reconstructed DBT image volumes.Methods:An SIR with a slice-wise total variation (TV) regularizer was implemented to reconstruct DBT images which were compared with the clinical reconstruction method (filtered backprojection). The artifact spread function (ASF) was measured to quantify the reduction of the through-plane artifacts level in phantom studies and microcalcifications in clinical cases. The anatomical clutter was quantified by the anatomical noise power spectrum with a power law fitting model: NPSa(f) = αf–β. The β values were measured from the reconstructed image slices when the two reconstruction methods were applied to a cohort of clinical breast exams (N = 101) acquired using Hologic Selenia Dimensions DBT systems.Results:The full width half maximum (FWHM) of the measured ASF was reduced from 8.7 ± 0.1 mm for clinical reconstruction to 6.5 ± 0.1 mm for SIR which yields a 25% reduction in FWHM in phantom studies and the same amount of ASF reduction was also found in clinical measurements from microcalcifications. The measured β values for the two reconstruction methods were 3.17 ± 0.36 and 2.14 ± 0.39 for the clinical reconstruction method and the SIR method, respectively. This difference was statistically significant (P ≪ 0.001). The dependence of β on slice location using either method was negligible.Conclusions:Statistical image reconstruction enabled a significant reduction of both the through-plane artifacts level and anatomical clutter in the DBT reconstructions. The β value was found to be β≈2.14 with the SIR method. This value stays in the middle between the β≈1.8 for cone beam CT and β≈3.2 for mammography. In contrast, the measured β value in the clinical reconstructions (β≈3.17) remains close to that of mammography.

Noise is an intrinsic property of every imaging system. For imaging systems using ionizing radiation, such as digital breast tomosynthesis (DBT) or digital mammography (DM), we strive to ensure that x-ray quantum noise is the limiting noise source in images, while using the lowest radiation dose possible to achieve clinically satisfactory images. Therefore, new computer methods are being sought to help reduce the dose of these systems. In the case of DBT, this can be achieved when solving the inverse problem of tomographic reconstruction. In this work, we propose to use a Non-Local Gaussian Markov Random Field (NLGMRF) model to represent a priori knowledge in a Bayesian (Maximum a Posteriori - MAP) reconstruction approach for DBT. The main advantage of the Non-Local Markov Random Field models is that they explicitly consider two important constraints to regularize the solution of this inverse problem - smoothing and redundancy. To evaluate this new method in DBT, a number of experiments were performed to compare these methods to existing reconstruction techniques. Comparable or superior results were achieved when compared with methods in the DBT reconstruction literature in terms of structural similarity index (SSIM), artifact spread function (ASF) and visual analysis, demonstrating that the NLGMRF model is suitable to regularize the MAP solution in DBT reconstruction.

Time for Change in Digital Breast Tomosynthesis Research.

Background: In invasive breast cancer patients being treated with neoadjuvant chemotherapy (NACT), achieving pathological complete response (pCR) is a useful goal of treatment. Monitoring response to NACT and predicting pCR is helpful in planning further therapy and providing robust prognostic information. Digital mammography (DM) and additional digital breast tomosynthesis (DBT) features are important tell-tales of tumor characteristics and behaviour. Following NACT, the mammographic features- both DM and DBT- of responding tumors can vary considerably. In this prospective study, we correlated the DM and DBT features of pre-NACT and post-NACT mammograms to investigate if these can reliably predict pCR to NACT. Methods: Following approval by institutional ethics committee, starting January 2016, 200 consecutive invasive breast carcinoma patients (mean age 51.2 years, all palpable breast masses) undergoing diagnostic breast imaging had their DM and DBT reviewed by two radiologists independently, who were blinded of the cyto/histology and the original DM and DBT reporting. Of these, 47 patients who were treated with NACT and had pre- and post-NACT DM and DBT were recruited. After a core-biopsy, radio-opaque marker(s) were placed in tumor core/margin. The pre- and post-NACT DM and DBT findings were compared and correlated with the extent of response of the primary breast tumor to NACT. DM and DBT characteristics predictive of (in-breast) pCR of index breast lesion were identified. Results: Of the 47 patients who underwent NACT, 44 received both anthracycline and taxane, and 3 received only an anthracycline based combination chemotherapy. Twelve patients underwent breast conservative surgery and the remaining underwent mastectomy. pCR was seen in 17 (36.2%) patients based on the surgical specimen histology. On clinical examination, 19 (40.4%) patients had clinical complete response (cCR) of the breast tumor, 11 (64.7%) of whom had pCR as well. Five patients had radiological complete response (rCR, no breast lesion visualised on post-NACT imaging)- 2 patients on DM alone, 2 patients on DBT alone, and one patient on both DM and DBT. Radio-opaque clips had some obscuring effects in 3 of these 5 patients, especially on DBT, in form of reduced visibility of breast lesion on DBT, c.w. corresponding DM images. All 5 patients with rCR had pCR (sensitivity=29.4%, specificity=100%), in contrast to only 11 (57.9%) patients with cCR having pCR. Patients with pCR had benign appearing (forced Bi-RADS 2 and 3) lesions on mammography more commonly on DM (p<0.001) than on DBT (p=0.042) (41.2% vs 23.5%). Post NACT lesion morphology varied significantly between patients with and without pCR on DM (p=0.038) but not on DBT (p=0.182). Pre-NACT forced Bi-RADS score, lesion morphology or margin characteristics on DM and DBT did not vary significantly amongst patients with and without pCR. Conclusions: Post-NACT DM and DBT features can predict pCR with high specificity but with low sensitivity. Pre-NACT DM and DBT features did not reliably predict response to NACT, and pCR in this study. DM may be better than DBT for assessing response to NACT in the presence of radio-opaque markers/clips. Citation Format: Agarwal G, Sonthineni C, Mohindra N, Jain N, Neyaz Z, Agrawal V, Krishnani N, Maylivahnan S, Mishra A, Lal P. Predicting pathological complete response (pCR) to neoadjuvant chemotherapy (NACT) based on pre- and post-NACT digital mammography and digital breast tomosynthesis findings [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P1-15-11.

Background Although digital breast tomosynthesis (DBT) is an emerging technique yielding higher sensitivity and specificity compared to digital mammography (DM) alone, relative contribution of prior mammograms on the interpretation of DBT combined with DM has not been investigated. Purpose To retrospectively compare the diagnostic performances of DM, DM + DBT, and DM + DBT with prior mammograms. Material and Methods Three breast radiologists independently reviewed images of 116 patients with 24 cancers in the sequential order of DM, DM + DBT, and DM + DBT with prior mammograms using Breast Imaging Reporting and Data System (BI-RADS) assessment categories. Results The average areas under the receiver operating characteristic curve (AUC) of DM, DM + DBT, and DM + DBT with prior mammograms were 0.712, 0.777, and 0.816, respectively. Adding prior mammograms did not significantly affect the AUC of DM + DBT ( P = 0.108), whereas adding DBT significantly increased the AUC of DM ( P = 0.009). Sensitivity for DM, DM + DBT, and DM + DBT with prior mammograms was 58.3%, 69.4%, and 69.4%, and specificities were 84.1%, 85.9%, and 93.8%, respectively. Addition of DBT significantly increased the sensitivity ( P = 0.0090) of DM. Prior mammograms significantly improved the specificity of DM + DBT ( P = 0.0004), whereas adding prior mammogram did not affect sensitivity of DM + DBT ( P = 1.000). Conclusion DBT significantly increases the overall sensitivity and diagnostic performance of DM. Prior mammograms significantly increase the specificity of DM + DBT but have no significant effect on sensitivity and overall diagnostic performance.

Effect of Arrival of Prior Mammograms on Recall Negation for Screening Mammograms Performed With Digital Breast Tomosynthesis in a Clinical Setting

.PurposeBreast cancer screening is predominantly performed using digital mammography (DM), but digital breast tomosynthesis (DBT) has higher sensitivity. DBT demands more resources than DM, and it might be more feasible to reserve DBT for women with a clear benefit from the technique. We explore if artificial intelligence (AI) can select women who would benefit from DBT imaging.ApproachWe used data from Malmö Breast Tomosynthesis Screening Trial, where all women prospectively were examined with separately double read DM and DBT. We retrospectively analyzed DM examinations () with a breast cancer detection system and used the provided risk score (1 to 10) for risk stratification. We tested how different score thresholds for adding DBT to an initial DM affects the number of detected cancers, additional DBT examinations needed, detection rate, and false positives.ResultsIf using a threshold of 9.0, 25 (26%) more cancers would be detected compared to using DM alone. Of the 41 cancers only detected on DBT, 61% would be detected, with only 1797 (12%) of the women examined with both DM and DBT. The detection rate for the added DBT would be 14/1000 women, whereas the false-positive recalls would be increased with 58 (21%).ConclusionUsing DBT only for selected high gain cases could be an alternative to complete DBT screening. AI can analyze initial DM images to identify high gain cases where DBT can be added during the same visit. There might be logistical challenges, and further studies in a prospective setting are necessary.

To compare the diagnostic performance of two-dimensional (2D) mammography, 2D mammography plus digital breast tomosynthesis (DBT), and synthetic 2D mammography plus DBT in depicting malignant radiographic features. In this multicenter, multireader, retrospective reading study (the TOMMY trial), after written informed consent was obtained, 8869 women (age range, 29-85 years; mean, 56 years) were recruited from July 2011 to March 2013 in an ethically approved study. From these women, a reading dataset of 7060 cases was randomly allocated for independent blinded review of (a) 2D mammography images, (b) 2D mammography plus DBT images, and (c) synthetic 2D mammography plus DBT images. Reviewers had no access to results of previous examinations. Overall sensitivities and specificities were calculated for younger women and those with dense breasts. Overall sensitivity was 87% for 2D mammography, 89% for 2D mammography plus DBT, and 88% for synthetic 2D mammography plus DBT. The addition of DBT was associated with a 34% increase in the odds of depicting cancer (odds ratio [OR] = 1.34, P = .06); however, this level did not achieve significance. For patients aged 50-59 years old, sensitivity was significantly higher (P = .01) for 2D mammography plus DBT than it was for 2D mammography. For those with breast density of 50% or more, sensitivity was 86% for 2D mammography compared with 93% for 2D mammography plus DBT (P = .03). Specificity was 57% for 2D mammography, 70% for 2D mammography plus DBT, and 72% for synthetic 2D mammography plusmDBT. Specificity was significantly higher than 2D mammography (P < .001in both cases) and was observed for all subgroups (P < .001 for all cases). The addition of DBT increased the sensitivity of 2D mammography in patients with dense breasts and the specificity of 2D mammography for all subgroups. The use of synthetic 2D DBT demonstrated performance similar to that of standard 2D mammography with DBT. DBT is of potential benefit to screening programs, particularly in younger women with dense breasts. (©) RSNA, 2015.

Architectural distortion (AD) is one of the breast abnormal signs in digital breast tomosynthesis (DBT) and digital mammography (DM). It is hard to be detected because of its subtle appearance and similar intensity with surrounding tissue. Since DBT is a three-dimensional imaging, it can address the problem of tissue superimposition in DM, so as to reduce false positives and false negatives. Several clinical studies have confirmed that radiologists can detect more ADs in DBT than in DM. These conclusions are based on subjective experience. To explore whether the engineering model and the experience of radiologists are consistent in AD detection tasks, this study compared the AD detection performance of a deep-learning-based computer-aided detection (CADe) model in DBT and DM images of the same group of cases. 394 DBT volumes and their corresponding DM images were collected retrospectively from 99 breast cancer screening cases. Among them, 203 DBT volumes and DM images contained ADs and the remaining 191 ones were negative group without any AD. Ten-fold cross-validation was used to train and evaluate the models and mean true positive fraction (MTPF) was used as figure-of-merit. The results showed that the CADe model achieved significantly better detection performance in DBT than DM (MTPF: 0.7026&plusmn;0.0394 for DBT vs. 0.5870&plusmn;0.0407 for DM, p=0.002). Qualitative analysis illustrated that DBT indeed had the ability to overcome tissue superimposition and showed more details of breast tissue. It helped the CADe model detect more ADs, which was consistent with clinical experience.

Digital Breast Tomosynthesis (DBT) is rapidly replacing mammography as the standard of care in breast cancer screening and diagnosis. DBT is a form of computed tomography, in which a limited set of projection images are acquired over a small angular range and reconstructed into tomographic data. The angular range varies from 15° to 50° and the number of projections varies between 9 and 25 projections, as determined by the equipment manufacturer. It is equally valid to treat DBT as the digital analog of classical tomography – that is, linear tomography. In fact, the name “tomosynthesis” stands for “synthetic tomography.” DBT shares many common features with classical tomography, including the radiographic appearance, dose, and image quality considerations. As such, both the science and practical physics of DBT systems is a hybrid between computed tomography and classical tomographic methods. In this lecture, we will explore the continuum from radiography to computed tomography to illustrate the characteristics of DBT. This lecture will consist of four presentations that will provide a complete overview of DBT, including a review of the fundamentals of DBT acquisition, a discussion of DBT reconstruction methods, an overview of dosimetry for DBT systems, and summary of the underlying image theory of DBT thereby relating image quality and dose. Learning Objectives: 1. To understand the fundamental principles behind tomosynthesis image acquisition. 2. To understand the fundamentals of tomosynthesis image reconstruction. 3. To learn the determinants of image quality and dose in DBT, including measurement techniques. 4. To learn the image theory underlying tomosynthesis, and the relationship between dose and image quality. ADM is a consultant to, and holds stock in, Real Time Tomography, LLC. ADM receives research support from Hologic Inc., Analogic Inc., and Barco NV.; ADM is a member of the Scientific Advisory Board for Gamma Medica Inc.; A. Maidment, Research Support, Hologic, Inc.; Research Support, Barco, Inc.; Scientific Advisory Board, Gamma Medica, Inc.; Scientific Advisory Board, Real-Time Tomography, LLC.; Shareholder, Real-Time Tomography, LLC; J. Mainprize, Our lab has a research agreement with GE Healthcare on various topics in digital mammography and digital tomosynthesis; W. Zhao, Research grant from Siemens Health Care

BACKGROUND. Prior studies comparing cancer screening by digital breast tomosynthesis (DBT) and digital mammography (DM) have largely entailed prospective trials and investigations at tertiary academic centers, which may encounter high-risk patient populations and lack heterogeneity among interpreting radiologists. Thus, results may not generalize across real-world community settings in the United States. OBJECTIVE. The purpose of this study was to compare DBT and DM in terms of screening performance and tumor characteristics of screen-detected cancers in a community practice setting. METHODS. Data were retrospectively retrieved for all screening mammography examinations performed by DBT or DM at four outpatient private practice facilities from January 1, 2012, to July 10, 2019. Examinations were interpreted by one of 26 radiologists (21 breast radiologists, five general radiologists). Further detailed information was recorded from radiology and pathology reports for all screen-detected cancers. Statistical comparisons were performed between DBT and DM in terms of screening performance and tumor characteristics of screen-detected cancers. RESULTS. A total of 310 cancers were detected in 47,096 screening DBT examinations and 83,200 screening DM examinations. Cancer detection rate was higher (p < .001) for DBT (3.4 per 1000 women) than for DM (1.8 per 1000 women). PPV1 was higher (p < .001) for DBT (3.5% [161/4641]) than for DM (2.1% [149/7116]). Patients with DBT-detected cancer were younger than those with DM-detected cancer (mean age [range], 61 years [40-87 years] vs 64 years [37-88 years]; p = .02). A greater percentage of DBT-detected than DM-detected cancers were invasive (85.1% [137/161] vs 72.5% [108/149]; p = .006), grade 1 when invasive (27.9% [38/136] vs 17.8% [19/107]; p = .04), and node negative (92.2% [71/77] vs 78.4% [58/74]; p = .02). Cancers detected by DBT and DM were not significantly different in histologic subtype, molecular subtype, or mean size (all p > .05). CONCLUSION. DBT showed a higher cancer detection rate and PPV1 than DM, and patients were younger at cancer diagnosis with DBT. Cancers detected on DBT were more often invasive, grade 1, and node negative. CLINICAL IMPACT. The findings support the generalizability of insights into DBT-based screening, which previously have been investigated primarily in academic settings.

As a special case of tomography, digital breast tomosynthesis (DBT) can realize quasi-3D image reconstruction for breast lesion detection from few-view and limited-angle projection data. For DBT image reconstruction, iterative algorithms are needed to suppress artifacts due to undersampling, and adaptive regularizations are necessary for preserving the edges of masses and calcifications. This paper presents a novel reconstruction method by regularizing projection onto convex sets (POCS) with multiscale Tikhonov-total variation (MTTV). The regularization, known as adaptive multiscale anisotropic diffusion, is able to preserve edges to a considerable extent and selectively suppress noise without introducing artifacts. The proposed method is referred to as MTTV-POCS and is evaluated quantitatively using 3D numerical breast and Shepp-Logan phantoms as well as two clinical volume images acquired from an advanced DBT machine. Experimental results show that the proposed method has better performance in metrics of peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) over two existing methods: adaptive-steepest-descent-POCS (ASD-POCS) and selective-diffusion regularized simultaneous algebraic reconstruction technique (SD-SART). As indicated by the results, the proposed method is applicable to DBT for high-quality image reconstruction.

Purpose To determine whether the rates and tumor characteristics of screening-detected and interval cancers differ for two-dimensional digital mammography (DM) versus digital breast tomosynthesis (DBT) mammography. Materials and Methods Consecutive screening mammograms from January 2009 to February 2011 (DM group, before DBT integration) and from January 2013 to February 2015 (DBT group, after complete DBT integration) were reviewed. Cancers were considered screening detected if diagnosed within 365 days of a positive screening examination and interval if diagnosed within 365 days of a negative screening examination. Z tests were used to compare cancers on DM versus DBT examinations. Results A total of 948 breast cancers were diagnosed after 78 385 DM and 76 896 DBT examinations. Although the overall rate of screening-detected cancers was similar with DM and DBT (5.0 vs 5.0 per 1000 examinations, P = .98), a higher proportion of screening-detected cancers were invasive rather than in situ with DBT (74.2% [287 of 387] vs 66.0% [260 of 394], P = .01). There were no significant differences in tumor characteristics, including size at pathologic examination, grade, hormone receptor status, and nodal status, between the screening-detected invasive cancers on DM versus DBT (P = .09-.99). The rate of interval cancers was similar with DM and DBT (1.1 vs 1.1 per 1000 examinations, P = .84). Compared with symptomatic interval cancers, magnetic resonance imaging-detected interval cancers were more likely to be minimal cancers. Conclusion The overall rates of screening-detected and interval cancers are similar with DM and DBT, but a higher proportion of screening-detected cancers are invasive rather than in situ with DBT. © RSNA, 2017.

Digital breast tomosynthesis (DBT) is a breast cancer screening modality that has gained popularity in recent years. Although insurance coverage for DBT is not mandated under the Patient Protection and Affordable Care Act, several states have required coverage without cost sharing for private insurers. To evaluate the association between state-level insurance coverage mandates for DBT and changes in DBT use, price, and out-of-pocket payments. This cohort study used an event-study design with repeated cross-sectional observations of US states. Data were obtained from the Blue Cross Blue Shield Axis database for commercially insured women aged 40 to 64 years who underwent screening mammography between January 1, 2015, and June 30, 2019. Data were analyzed between January 14, 2021, and January 20, 2022. Passage of state-level legislation requiring insurance coverage of DBT. Change in DBT use among women screened for breast cancer, overall DBT price, and out-of-pocket payments for DBT in states with mandates for coverage of DBT compared with states that did not pass legislation. This study included 9 604 084 screening mammograms from 5 754 123 women (mean [SD] age, of 53 [6.7] years). During the study period, 15 states enacted DBT coverage mandates and 34 states did not. In states with coverage mandates, DBT use increased by 9.0 percentage points (95% CI, 1.8-16.3 percentage points; P = .02) 2 years after the mandate compared with states without coverage mandates. Coverage mandates were also associated with a net $38.7 (95% CI, $13.4-$63.9; P = .003) decrease in the mean price of DBT compared with no coverage mandates. There was no association between coverage mandates and out-of-pocket payments 2 years after mandate passage ($-2.1; 95% CI, $-5.3 to $1.0; P = .18). In this cohort study, DBT coverage mandates were associated with an increase in DBT use but not with any change in out-of-pocket payments. The findings suggest that coverage mandates for DBT may have been associated with broader use but were not associated with changes in direct costs to patients.

AimsTo assess the diagnostic performance of digital breast tomosynthesis (DBT) in older women across varying breast densities and to compare its effectiveness for cancer detection with 2D mammography and ultrasound (U/S) for different breast density categories. Furthermore, our study aimed to predict the potential reduction in unnecessary additional examinations among older women due to DBT.MethodsThis study encompassed a cohort of 224 older women. Each participant underwent both 2D mammography and digital breast tomosynthesis examinations. Supplementary views were conducted when necessary, including spot compression and magnification, ultrasound, and recommended biopsies. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC) were calculated for 2D mammography, DBT, and ultrasound. The impact of DBT on diminishing the need for supplementary imaging procedures was predicted through binary logistic regression.ResultsIn dense breast tissue, DBT exhibited notably heightened sensitivity and NPV for lesion detection compared to non-dense breasts (61.9% vs. 49.3%, p < 0.001) and (72.9% vs. 67.9%, p < 0.001), respectively. However, the AUC value of DBT in dense breasts was lower compared with non-dense breasts (0.425 vs. 0.670). Regarding the ability to detect calcifications, DBT demonstrated significantly improved sensitivity and NPV in dense breasts compared to non-dense breasts (100% vs. 99.2%, p < 0.001) and (100% vs. 94.7%, p < 0.001), respectively. On the other hand, the AUC value of DBT was slightly lower in dense breasts compared with non-dense (0.682 vs. 0.711). Regarding lesion detection for all cases between imaging examinations, the highest sensitivity was observed in 2D mammography (91.7%, p < 0.001), followed by DBT (83.7%, p < 0.001), and then ultrasound (60.6%, p < 0.001). In dense breasts, sensitivity for lesion detection was highest in 2D mammography (92.9%, p < 0.001), followed by ultrasound (76.2%, p < 0.001), and the last one was DBT. In non-dense breasts, sensitivities were 91% (p < 0.001) for 2D mammography, 50.7% (p < 0.001) for ultrasound, and 49.3% (p < 0.001) for DBT. In terms of calcification detection, DBT displayed significantly superior sensitivity compared to 2D mammography in both dense and non-dense breasts (100% vs. 91.4%, p < 0.001) and (99.2% vs. 78.5%, p < 0.001), respectively. However, the logistic regression model did not identify any statistically significant relationship (p > 0.05) between DBT and the four dependent variables.ConclusionOur findings indicate that among older women, DBT does not significantly decrease the requirement for further medical examinations.