To evaluate uptake, patterns of use, and perception of digital breast tomosynthesis (DBT) among practicing breast radiologists.
Institutional Review Board exemption was obtained for this Health ...Insurance Portability and Accountability Act-compliant electronic survey, sent to 7023 breast radiologists identified via the Radiological Society of North America database. Respondents were asked of their geographic location and practice type. DBT users reported length of use, selection criteria, interpretive sequences, recall rate, and reading time. Radiologist satisfaction with DBT as a diagnostic tool was assessed (1-5 scale).
There were 1156 (16.5%) responders, 65.8% from the United States and 34.2% from abroad. Of these, 749 (68.6%) use DBT; 22.6% in academia, 56.5% private, and 21% other. Participants are equally likely to report use of DBT if they worked in academics versus in private practice (78.2% 169 of 216 vs 71% 423 of 596) (odds ratio, 1.10; 95% confidence interval: 0.87-1.40; P = 1.000). Of nonusers, 43% (147 of 343) plan to adopt DBT. No US regional differences in uptake were observed (P = 1.000). Although 59.3% (416 of 702) of DBT users include synthetic 2D (s2D) for interpretation, only 24.2% (170 of 702) use s2D alone. Majority (66%; 441 of 672) do not perform DBT-guided procedures. Radiologist (76.6%) (544 of 710) satisfaction with DBT as a diagnostic tool is high (score ≥ 4/5).
DBT is being adopted worldwide across all practice types, yet variations in examination indication, patient selection, utilization of s2D images, and access to DBT-guided procedures persist, highlighting the need for consensus and standardization.
Purpose:
Develop a computer-aided detection (CAD) system for masses in digital breast tomosynthesis (DBT) volume using a deep convolutional neural network (DCNN) with transfer learning from ...mammograms.
Methods:
A data set containing 2282 digitized film and digital mammograms and 324 DBT volumes were collected with IRB approval. The mass of interest on the images was marked by an experienced breast radiologist as reference standard. The data set was partitioned into a training set (2282 mammograms with 2461 masses and 230 DBT views with 228 masses) and an independent test set (94 DBT views with 89 masses). For DCNN training, the region of interest (ROI) containing the mass (true positive) was extracted from each image. False positive (FP) ROIs were identified at prescreening by their previously developed CAD systems. After data augmentation, a total of 45 072 mammographic ROIs and 37 450 DBT ROIs were obtained. Data normalization and reduction of non-uniformity in the ROIs across heterogeneous data was achieved using a background correction method applied to each ROI. A DCNN with four convolutional layers and three fully connected (FC) layers was first trained on the mammography data. Jittering and dropout techniques were used to reduce overfitting. After training with the mammographic ROIs, all weights in the first three convolutional layers were frozen, and only the last convolution layer and the FC layers were randomly initialized again and trained using the DBT training ROIs. The authors compared the performances of two CAD systems for mass detection in DBT: one used the DCNN-based approach and the other used their previously developed feature-based approach for FP reduction. The prescreening stage was identical in both systems, passing the same set of mass candidates to the FP reduction stage. For the feature-based CAD system, 3D clustering and active contour method was used for segmentation; morphological, gray level, and texture features were extracted and merged with a linear discriminant classifier to score the detected masses. For the DCNN-based CAD system, ROIs from five consecutive slices centered at each candidate were passed through the trained DCNN and a mass likelihood score was generated. The performances of the CAD systems were evaluated using free-response ROC curves and the performance difference was analyzed using a non-parametric method.
Results:
Before transfer learning, the DCNN trained only on mammograms with an AUC of 0.99 classified DBT masses with an AUC of 0.81 in the DBT training set. After transfer learning with DBT, the AUC improved to 0.90. For breast-based CAD detection in the test set, the sensitivity for the feature-based and the DCNN-based CAD systems was 83% and 91%, respectively, at 1 FP/DBT volume. The difference between the performances for the two systems was statistically significant (p-value < 0.05).
Conclusions:
The image patterns learned from the mammograms were transferred to the mass detection on DBT slices through the DCNN. This study demonstrated that large data sets collected from mammography are useful for developing new CAD systems for DBT, alleviating the problem and effort of collecting entirely new large data sets for the new modality.
Purpose
High dose rate (HDR) brachytherapy rapidly delivers dose to targets with steep dose gradients. This treatment method must adhere to prescribed treatment plans with high spatiotemporal ...accuracy and precision, as failure to do so may degrade clinical outcomes. One approach to achieving this goal is to develop imaging techniques to track HDR sources in vivo in reference to surrounding anatomy. This work investigates the feasibility of using an isocentric C‐arm x‐ray imager and tomosynthesis methods to track Ir‐192 HDR brachytherapy sources in vivo over time (4D).
Methods
A tomosynthesis imaging workflow was proposed and its achievable source detectability, localization accuracy, and spatiotemporal resolution were investigated in silico. An anthropomorphic female XCAT phantom was modified to include a vaginal cylinder applicator and Ir‐192 HDR source (0.5 × 0.5 × 5.0 mm3), and the workflow was carried out using the MC‐GPU Monte Carlo image simulation platform. Source detectability was characterized using the reconstructed source signal‐difference‐to‐noise‐ratio (SDNR), localization accuracy by the absolute 3D error in its measured centroid location, and spatiotemporal resolution by the full‐width‐at‐half‐maximum (FWHM) of line profiles through the source in each spatial dimension considering a maximum C‐arm angular velocity of 30° per second. The dependence of these parameters on acquisition angular range (θtot = 0°–90°), number of views, angular increment between views (Δθ = 0°–15°), and volumetric constraints imposed in reconstruction was evaluated. Organ voxel doses were tallied to derive the workflow's attributable effective dose.
Results
The HDR source was readily detected and its centroid was accurately localized with the proposed workflow and method (SDNR: 10–40, 3D error: 0–0.144 mm). Tradeoffs were demonstrated for various combinations of image acquisition parameters; namely, increasing the tomosynthesis acquisition angular range improved resolution in the depth‐encoded direction, for example from 2.5 mm to 1.2 mm between θtot = 30o and θtot = 90o, at the cost of increasing acquisition time from 1 to 3 s. The best‐performing acquisition parameters (θtot = 90o, Δθ = 1°) yielded no centroid localization error, and achieved submillimeter source resolution (0.57 × 1.21 × 5.04 mm3 apparent source dimensions, FWHM). The total effective dose for the workflow was 263 µSv for its required pre‐treatment imaging component and 7.59 µSv per mid‐treatment acquisition thereafter, which is comparable to common diagnostic radiology exams.
Conclusions
A system and method for tracking HDR brachytherapy sources in vivo using C‐arm tomosynthesis was proposed and its performance investigated in silico. Tradeoffs in source conspicuity, localization accuracy, spatiotemporal resolution, and dose were determined. The results suggest this approach is feasible for localizing an Ir‐192 HDR source in vivo with submillimeter spatial resolution, 1–3 second temporal resolution and minimal additional dose burden.
•Novel study to apply AI to breast tomosynthesis for treatment response prediction.•AI model shows high discriminative performance (AUC 0.83) for response prediction.•Model decision analysis, ...explainable AI, was used in the study.
Different imaging tools, including digital breast tomosynthesis (DBT), are frequently used for evaluating tumor response during neoadjuvant chemotherapy (NACT). This study aimed to explore whether using artificial intelligence (AI) for serial DBT acquisitions during NACT for breast cancer can predict pathological complete response (pCR) after completion of NACT.
A total of 149 women (mean age 53 years, pCR rate 22 %) with breast cancer treated with NACT at Skane University Hospital, Sweden, between 2014 and 2019, were prospectively included in this observational cohort study (ClinicalTrials.gov: NCT02306096). DBT images from both the cancer and contralateral healthy breasts acquired at three time points: pre-NACT, after two cycles of NACT, and after the completion of six cycles of NACT (pre-surgery) were analyzed. The deep learning AI system used to predict pCR consisted of a backbone 3D ResNet and an attention and prediction module. The GradCAM method was used to obtain insights into the model decision basis through a quantitative analysis of the importance maps on the validation set. Moreover, specific model choices were motivated by ablation studies.
The AI model reached an AUC of 0.83 (95% CI: 0.63–1.00) (test set). The spatial correlation of importance maps for input volumes from the same patient but at different time points was high, possibly indicating that the model focuses on the same areas during decision-making.
We demonstrate a high discriminative performance of our algorithm for predicting pCR/non-pCR. Availability of larger datasets would permit more comprehensive training of the models and more rigorous evaluation of their prediction performance for future patients.
Saving Women's Lives Petitti, Diana B; Penhoet, Edward E; Joy, Janet E
02/2005
eBook
Open access
The outlook for women with breast cancer has improved in recent years. Due to the combination of improved treatments and the benefits of mammography screening, breast cancer mortality has decreased ...steadily since 1989. Yet breast cancer remains a major problem, second only to lung cancer as a leading cause of death from cancer for women. To date, no means to prevent breast cancer has been discovered and experience has shown that treatments are most effective when a cancer is detected early, before it has spread to other tissues. These two facts suggest that the most effective way to continue reducing the death toll from breast cancer is improved early detection and diagnosis.
Building on the 2001 report Mammography and Beyond , this new book not only examines ways to improve implementation and use of new and current breast cancer detection technologies but also evaluates the need to develop tools that identify women who would benefit most from early detection screening. Saving Women's Lives: Strategies for Improving Breast Cancer Detection and Diagnosis encourages more research that integrates the development, validation, and analysis of the types of technologies in clinical practice that promote improved risk identification techniques. In this way, methods and technologies that improve detection and diagnosis can be more effectively developed and implemented.
•Optimal angular range for DBT image acquisition varies based on breast and lesion characteristics.•We used a modular phantom and a prototype DBT system to test different angular ranges with 11 ...projections.•Findings suggest that the ±12.5° angular range could provide the highest overall lesion visibility.
To determine the optimal angular range (AR) for digital breast tomosynthesis (DBT) systems that provides highest lesion visibility across various breast densities and thicknesses.
A modular DBT phantom, consisting of tissue-equivalent adipose and glandular modules, along with a module embedded with test objects (speckles, masses, fibers), was used to create combinations simulating different breast thicknesses, densities, and lesion locations. A prototype DBT system operated at four ARs (AR±7.5°, AR±12.5°, AR±19°, and AR±25°) to acquire 11 projection images for each combination, with separate fixed doses for thin and thick combinations. Three blinded radiologists independently assessed lesion visibility in reconstructed images; assessments were averaged and compared using linear mixed models.
Speckle visibility was highest with AR±7.5° or AR±12.5°, decreasing with wider ARs in all density and thickness combinations. The difference between AR±7.5° and AR±12.5° was not statistically significant, except for the tube-side speckles in thin-fatty combinations (5.83 AR±7.5° vs. 5.39 AR±12.5°, P = 0.019). Mass visibility was not affected by AR in thick combinations, while AR±12.5° exhibited the highest mass visibility for both thin-fatty and thin-dense combinations (P = 0.032 and 0.007, respectively). Different ARs provided highest fiber visibility for different combinations; however, AR±12.5° consistently provided highest or comparable visibility. AR±12.5° showed highest overall lesion visibility for all density and thickness combinations.
AR±12.5° exhibited the highest overall lesion visibility across various phantom thicknesses and densities using a projection number of 11.
This session along with the others on Digital Breast Tomosynthesis (DBT) offered at this years meeting as part of the Educational, Scientific and Partners in Solutions is intended to meet the FDA ...requirements for physicists to survey a DBT system. Specifically, this vendor offered session held as part of the Partners in Solutions is also meant to focus on those requirements. This session is being done along with the other two vendors of approved DBT systems in order to cover the machine specific features of those systems.
This session along with the others on Digital Breast Tomosynthesis (DBT) offered at this years meeting as part of the Educational, Scientific and Partners in Solutions is intended to meet the FDA ...requirements for physicists to survey a DBT system. Specifically, this vendor offered session held as part of the Partners in Solutions is also meant to focus on those requirements. This session is being done along with the other two vendors of approved DBT systems in order to cover the machine specific features of those systems.