Deep learning reconstruction of digital breast tomosynthesis images for accurate breast density and patient-specific radiation dose estimation
Radboud University Medical Center（拉德堡德大学医学中心）；Netherlands Cancer Institute（荷兰癌症研究所）；The University of Chicago（芝加哥大学）；University of Pennsylvania（宾夕法尼亚大学）；Lund University（隆德大学）；Vall d’Hebron Institute of Oncology（巴尔德希伯伦肿瘤研究所）；University of Barcelona（巴塞罗那大学）；Dutch Expert Centre for Screening (LRCB)（荷兰筛查专家中心）
Jonas Teuwen, Nikita Moriakov, Christian Fedon, Marco Caballo, Ingrid Reiser, Pedrag Bakic, Eloy García, Oliver Diaz, Koen Michielsen, Ioannis Sechopoulos
The two-dimensional nature of mammography makes estimation of the overall breast density challenging, and estimation of the true patient-specific radiation dose impossible. Digital breast tomosynthesis (DBT), a pseudo-3D technique, is now commonly used in breast cancer screening and diagnostics. Still, the severely limited 3rd dimension information in DBT has not been used, until now, to estimate the true breast density or the patient-specific dose. This study proposes a reconstruction algorithm for DBT based on deep learning specifically optimized for these tasks. The algorithm, which we name DBToR, is based on unrolling a proximal-dual optimization method. The proximal operators are replaced with convolutional neural networks and prior knowledge is included in the model. This extends previous work on a deep learning-based reconstruction model by providing both the primal and the dual blocks with breast thickness information, which is available in DBT. Training and testing of the model were performed using virtual patient phantoms from two different sources. Reconstruction performance, and accuracy in estimation of breast density and radiation dose, were estimated, showing high accuracy (density <+/-3%; dose <+/-20%) without bias, significantly improving on the current state-of-the-art. This work also lays the groundwork for developing a deep learning-based reconstruction algorithm for the task of image interpretation by radiologists.
In this paper, we would like to quantitatively measure the tumor volume contained in the breast imaged by the Digital Breast Tomosynthesis (DBT), a reconstructed 3D image. The estimated volume will add to the prognostic value of risk classification of breast cancer. We develop an algorithm that offers an alternative way of estimating the volume of the tumor in a breast. We segment the region of interest by expressing the ratio of tumor region to normal region as a function of the threshold value in the image. Next, we determine the volume of the tumor region as a function of the threshold. We …
Measurements of breast density have the potential to improve the efficiency and reduce the cost of screening mammography through personalized screening. Breast density has traditionally been evaluated from the dense area in a mammogram, but volumetric assessment methods, which measure the volumetric fraction of fibro-glandular tissue in the breast, are potentially more consistent and physically sound. The purpose of the present study is to evaluate a method for measuring the volumetric breast density using photon-counting spectral tomosynthesis. The performance of the method was evaluated usin…
There are a number of clinically relevant tasks in digital breast tomosynthesis (DBT) involving the detection and visual assessment of fiber-like structures such as Cooper's ligaments, blood vessels, and spiculated lesions. Such structures can exhibit orientation dependent variations in conspicuity. This study demonstrates the presence of in-plane orientation-dependent signal conspicuity for fiber-like signals in DBT and shows how reconstruction algorithm design can mitigate this phenomenon. We uncover a tradeoff between minimizing orientation-dependence and preserving depth resolution that is…
We present the first evaluation of a recently developed silicon-strip detector for photon-counting dual-energy breast tomosynthesis. The detector is well suited for tomosynthesis with high dose efficiency and intrinsic scatter rejection. A method was developed for measuring the spatial resolution of a system based on the detector in terms of the three-dimensional modulation transfer function (MTF). The measurements agreed well with theoretical expectations, and it was seen that depth resolution was won at the cost of a slightly decreased lateral resolution. This may be a justifiable trade-off …
Purpose: The goal of this study is to develop a novel deep learning (DL) based reconstruction framework to improve the digital breast tomosynthesis (DBT) imaging performance. Methods: In this work, the DIR-DBTnet is developed for DBT image reconstruction by unrolling the standard iterative reconstruction algorithm within the deep learning framework. In particular, such network learns the regularizer and the iteration parameters automatically through network training with a large amount of simulated DBT data. Afterwards, both numerical and experimental data are used to evaluate its performance.…
Reconstruction of digital breast tomosynthesis is a challenging problem due to the limited angle data available in such systems. Due to memory limitations, deep learning-based methods can help improve these reconstructions, but can not (yet) attain sufficiently high resolution. In addition to this practical issue, questions remain on the possibility of such models introducing 'ghost' information from the training data that is not compatible with the projection data. To take advantage of some of the benefits of deep learning-based reconstructions while avoiding these limitations, we propose to …
Digital breast tomosynthesis (DBT) is an emerging modality for breast imaging. A typical tomosynthesis image is reconstructed from projection data acquired at a limited number of views over a limited angular range. In general, the quantitative accuracy of the image can be significantly compromised by severe artifacts and non-isotropic resolution resulting from the incomplete data. Nevertheless, it has been demonstrated that DBT may yield useful information for detection/classification tasks and thus is considered a promising breast imaging modality currently undergoing pre-clinical evaluation …
Radboud University Medical Center（拉德堡德大学医学中心）；Delft University of Technology（代尔夫特理工大学）；KTH Royal Institute of Technology（瑞典皇家理工学院）；Elekta（医科达）；Dutch Expert Centre for Screening（荷兰筛查专家中心）
Digital breast tomosynthesis is rapidly replacing digital mammography as the basic x-ray technique for evaluation of the breasts. However, the sparse sampling and limited angular range gives rise to different artifacts, which manufacturers try to solve in several ways. In this study we propose an extension of the Learned Primal-Dual algorithm for digital breast tomosynthesis. The Learned Primal-Dual algorithm is a deep neural network consisting of several `reconstruction blocks', which take in raw sinogram data as the initial input, perform a forward and a backward pass by taking projections a…
Digital Breast Tomosynthesis (DBT) provides an insight into the fine details of normal fibroglandular tissues and abnormal lesions by reconstructing a pseudo-3D image of the breast. In this respect, DBT overcomes a major limitation of conventional X-ray mammography by reducing the confounding effects caused by the superposition of breast tissue. In a breast cancer screening or diagnostic context, a radiologist is interested in detecting change, which might be indicative of malignant disease. To help automate this task image registration is required to establish spatial correspondence between t…
PURPOSE: We develop a practical, iterative algorithm for image-reconstruction in under-sampled tomographic systems, such as digital breast tomosynthesis (DBT). METHOD: The algorithm controls image regularity by minimizing the image total $p$-variation (TpV), a function that reduces to the total variation when $p=1.0$ or the image roughness when $p=2.0$. Constraints on the image, such as image positivity and estimated projection-data tolerance, are enforced by projection onto convex sets (POCS). The fact that the tomographic system is under-sampled translates to the mathematical property that m…
Fiber-like features are an important aspect of breast imaging. Vessels and ducts are present in all breast images, and spiculations radiating from a mass can indicate malignancy. Accordingly, fiber objects are one of the three types of signals used in the American College of Radiology digital mammography (ACR-DM) accreditation phantom. This work focuses on the image properties of fiber-like structures in digital breast tomosynthesis (DBT) and how image reconstruction can affect their appearance. The impact of DBT image reconstruction algorithm and regularization strength on the conspicuity of …
Mammography is a specific type of breast imaging that uses low-dose X-rays to detect cancer in early stage. During the exam, the women breast is compressed between two plates until a nearly uniform breast thickness is obtained. This technique improves image quality and reduces dose but can also be the source of discomfort and sometimes pain for the patient. Therefore, alternative techniques allowing reduced breast compression is of potential interest. The aim of this work is to develop a 3D biomechanical Finite Element (FE) breast model in order to analyze various breast compression strategies…
Simulation-based image quality metrics are adapted and investigated for characterizing the parameter dependences of linear iterative image reconstruction for DBT. Three metrics based on 2D DBT simulation are investigated: (1) a root-mean-square-error (RMSE) between the test phantom and reconstructed image, (2) a gradient RMSE where the comparison is made after taking a spatial gradient of both image and phantom, and (3) a region-of-interest (ROI) Hotelling observer (HO) for signal-known-exactly/background-known-exactly (SKE/BKE) and signal-known-exactly/background-known-statistically (SKE/BKS)…
Cone-beam breast computed tomography (CT) provides true 3D breast images with isotropic resolution and high-contrast information, detecting calcifications as small as a few hundred microns and revealing subtle tissue differences. However, breast is highly sensitive to x-ray radiation. It is critically important for healthcare to reduce radiation dose. Few-view cone-beam CT only uses a fraction of x-ray projection data acquired by standard cone-beam breast CT, enabling significant reduction of the radiation dose. However, insufficient sampling data would cause severe streak artifacts in CT imag…
Radiation dose is an important performance indicator of a dedicated breast CT (DBCT). In this paper, the method of putting thermoluminescent dosimeters (TLD) into a breast shaped PMMA phantom to study the dose distribution in breasts was improved by using smaller TLDs and a new half-ellipsoid PMMA phantom. Then the weighted CT dose index (CTDIw) was introduced to average glandular assessment in DBCT for the first time and two measurement modes were proposed for different sizes of breasts. The dose deviations caused by using cylindrical phantoms were simulated using the Monte Carlo method and a…
Purpose: The purpose of the work was experimental investigations of the breast dose distributions with adaptive filtration. Adaptive filtration reduces detector dynamic range and improves image quality. The adaptive filter with predetermined shape is placed at the x-ray beam such that the x-ray intensity at the detector surface is flat. However, adaptive filter alters the mean dose to the breast, as well as volume distribution of the dose. Methods: The dose was measured using a 14 cm diameter cylindrical acrylic breast phantom. An acrylic adaptive filter was fabricated to match the 14 cm diame…
Purpose: CBCT-based adaptive radiotherapy requires daily images for accurate dose calculations. This study investigates the feasibility of applying a single convolutional network to facilitate CBCT-to-CT synthesis for head-and-neck, lung, and breast cancer patients. Methods: Ninety-nine patients diagnosed with head-and-neck, lung or breast cancer undergoing radiotherapy with CBCT-based position verification were included in this study. CBCTs were registered to planning CTs according to clinical procedures. Three cycle-consistent generative adversarial networks (cycle-GANs) were trained in an u…
A new strategy for radiation therapy dosimetry planning (RTDP) used to reduce dose estimation errors due to respiratory motion in breast treatment was illustrated and evaluated in this study. On CT data set acquired for breast treatment, six different RTDP tangential techniques were performed: (i) three-dimensional conformal radiotherapy (3DCRT) with physical wedge, (ii) 3DCRT with virtual wedge, (iii) motion management technique (MMT) with physical wedge, (iv) MMT with virtual wedge, (v) 3DCRT with field-in-field, and (vi) intensity-modulated radiation therapy (IMRT) with direct aperture opti…
Towards order of magnitude X-ray dose reduction in breast cancer imaging using phase contrast and deep denoising
The University of Melbourne（墨尔本大学）；The University of Sydney（悉尼大学）；Eastern Health（东部健康）；Monash Health（莫纳什健康）；Monash University（莫纳什大学）；Australian Synchrotron, ANSTO（澳大利亚同步辐射中心，ANSTO）；Commonwealth Scientific and Industrial Research Organisation（澳大利亚联邦科学与工业研究组织）
PURPOSE: Permanent breast seed implant (PBSI) using 103Pd is emerging as an effective adjuvant radiation technique for early-stage breast cancer. However, clinical dose evaluations follow the water-based TG-43 approach with its considerable approximations. Towards clinical adoption of advanced TG-186 model-based dose evaluations, this study presents a comprehensive investigation for PBSI considering both target and normal tissue doses. METHODS: Dose calculations are performed with the free open-source Monte Carlo (MC) code, egs_brachy, using 2 types of virtual patient models: TG43sim (simulate…
For a century, clinical X-ray imaging has visualised only the attenuation properties of tissue, which fundamentally limits the contrast, particularly in soft tissues like the breast. Imaging based on refraction can overcome this limitation, but so far has been constrained to high-dose ex-vivo applications or required highly coherent X-ray sources, like synchrotrons. It has been predicted that grating interferometry (GI) could eventually allow computed tomography (CT) to be more dose-efficient. However, the benefit of refraction in clinical CT has not been demonstrated so far. Here we show that…
Mammography is currently the primary imaging modality for breast cancer screening and plays an important role in cancer diagnostics. A standard mammographic image acquisition always includes the compression of the breast prior x-ray exposure. The breast is compressed between two plates (the image receptor and the compression paddle) until a nearly uniform breast thickness is obtained. The breast flattening improves diagnostic image quality 1 and reduces the absorbed dose 2. However, this technique can also be a source of discomfort and might deter some women from attending breast screening by …
Background: The deep inspiration breath hold technique (DIBH) is widely used for left-sided breast radiotherapy (RT), in order to reduce the dose to the heart and the risk of cardiotoxicity. The volumetric arc therapy (VMAT) technique increases the dose conformity to the planning target volume (PTV). We hereby describe the procedure of combining VMAT and DIBH without a dedicated equipment, and report a dosimetric study related on our implemented technique. Methods: The DIBH technique is based on voluntary breath hold of the patient which is controlled by a laser on tattoo marks. Patients were …
Monitoring Neoadjuvant chemotherapy (NAC) effects is necessary to capture resistant patients and stop or change treatment. The aim of this study was to assess the tumor response at an early stage, after the first doses of the NAC, based on the variability of the backscattered ultrasound energy, and backscatter statistics. The backscatter statistics has not previously been used to monitor NAC effects. The B-mode ultrasound images and raw radio frequency data from breast tumors were obtained using an ultrasound scanner before chemotherapy and 1 week after each NAC cycle. Twenty-four malignant br…
Automating the radiotherapy treatment planning process is a technically challenging problem. The majority of automated approaches have focused on customizing and inferring dose volume objectives to used in plan optimization. In this work we outline a multi-patient atlas-based dose prediction approach that learns to predict the dose-per-voxel for a novel patient directly from the computed tomography (CT) planning scan without the requirement of specifying any objectives. Our method learns to automatically select the most effective atlases for a novel patient, and then map the dose from those at…
Impact of loss functions on the performance of a deep neural network designed to restore low-dose digital mammography
Rensselaer Polytechnic Institute（伦斯勒理工学院）；Fudan University（复旦大学）；Shanghai Center for Brain Science and Brain-inspired Technology（上海脑科学与类脑研究中心）；São Carlos School of Engineering, University of São Paulo（圣保罗大学圣卡洛斯工程学院）；Ribeirão Preto School of Medicine, University of São Paulo（圣保罗大学里贝朗普雷图医学院）
Changes in Visual Attention Patterns for Detection Tasks due to Dependencies on Signal and Background Spatial Frequencies
Department of Biomedical Engineering, University of Houston（生物医学工程系，休斯顿大学）；Department of Physics, University of Houston（物理系，休斯顿大学）；Department of Electrical and Computer Engineering, University of Houston（电气与计算机工程系，休斯顿大学）
Treatment planning system calculations in inhomogeneous regions may present significant inaccuracies due to loss of electronic equilibrium. In this study, three different dose calculation algorithms, pencil beam, collapsed cone, and Monte-Carlo, provided by our planning system were compared to assess their impact on the three-dimensional planning of lung and breast cases. A total of five breast and five lung cases were calculated using the PB, CC, and MC algorithms. Planning treatment volume and organs at risk delineation was performed according to our institutions protocols on the Oncentra Ma…
Background: Mammography is a specific type of breast imaging that uses low-dose X-rays to detect cancer in early stage. During the exam, the women breast is compressed between two plates in order to even out the breast thickness and to spread out the soft tissues. This technique improves exam quality but can be uncomfortable for the patient. The perceived discomfort can be assessed by the means of a breast biomechanical model. Alternative breast compression techniques may be computationally investigated trough finite elements simulations.Methods: The aim of this work is to develop and evaluate…
Cone beam computed tomography (CBCT) for breast imaging has potential to replace conventional mammograms. However, concerns over dose and image quality prevent CBBCT systems from the clinical trial phase to next stage. The time of flight (TOF) method was recently shown to reduce the x-ray scattering effects by 95% and improve the image CNR by 110% for large volume objects. The advancements in x-ray sources like in compact Free Electron Lasers (FEL) and advancements in detector technology show potential for the TOF method to be feasible in CBCT when imaging large objects. In this study, we inve…
Breast X-ray CT imaging is being considered in screening as an extension to mammography. As a large fraction of the population will be exposed to radiation, low-dose imaging is essential. Iterative image reconstruction based on solving an optimization problem, such as Total-Variation minimization, shows potential for reconstruction from sparse-view data. For iterative methods it is important to ensure convergence to an accurate solution, since important image features, such as presence of microcalcifications indicating breast cancer, may not be visible in a non-converged reconstruction, and th…
In this study we performed a feasibility investigation on implementing a fast and accurate dose calculation based on a deep learning technique. A two dimensional (2D) fluence map was first converted into a three dimensional (3D) volume using ray traversal algorithm. A 3D U-Net like deep residual network was then established to learn a mapping between this converted 3D volume, CT and 3D dose distribution. Therefore an indirect relationship was built between a fluence map and its corresponding 3D dose distribution without using significantly complex neural networks. 200 patients, including nasop…
Cherenkov Imaged Bio-morphological Features Verify Patient Positioning with Deformable Tissue Translocation in Breast Radiotherapy
Dartmouth College（达特茅斯学院）；Dartmouth Cancer Center（达特茅斯癌症中心）；Dartmouth Health（达特茅斯健康）；University of Wisconsin-Madison（威斯康星大学麦迪逊分校）；University of Missouri-Columbia（密苏里大学哥伦比亚分校）
Purpose: To develop a multi-criteria optimization framework for image guided radiotherapy. Methods: An algorithm is proposed for a multi-criteria framework for the purpose of patient setup verification decision processes. Optimal patient setup shifts and rotations are not always straightforward, particularly for deformable or moving targets of the spine, abdomen, thorax, breast, head and neck and limbs. The algorithm relies upon dosimetric constraints and objectives to aid in the patient setup such that the patient is setup to maximize tumor dose coverage and minimize dose to organs at risk wh…
Purpose: Many useful image quality metrics for evaluating linear image reconstruction techniques do not apply to or are difficult to interpret for non-linear image reconstruction. The vast majority of metrics employed for evaluating non-linear image reconstruction are based on some form of global image fidelity, such as image root mean square error (RMSE). Use of such metrics can lead to over-regularization in the sense that they can favor removal of subtle details in the image. To address this shortcoming, we develop an image quality metric based on signal detection that serves as a surrogate…
