Fully iterative scatter corrected digital breast tomosynthesis using GPU-based fast Monte Carlo simulation and composition ratio update

A deep learning approach to estimate x-ray scatter in digital breast tomosynthesis: From phantom models to clinical applications

BACKGROUND

Digital breast tomosynthesis (DBT) has gained popularity as breast imaging modality due to its pseudo-3D reconstruction and improved accuracy compared to digital mammography. However, DBT faces challenges in image quality and quantitative accuracy due to scatter radiation. Recent advancements in deep learning (DL) have shown promise in using fast convolutional neural networks for scatter correction, achieving comparable results to Monte Carlo (MC) simulations.

PURPOSE

To predict the scatter radiation signal in DBT projections within clinically-acceptable times and using only clinically-available data, such as compressed breast thickness and acquisition angle.

METHODS

MC simulations to obtain scatter estimates were generated from two types of digital breast phantoms. One set consisted of 600 realistically-shaped homogeneous breast phantoms for initial DL training. The other set was composed of 80 anthropomorphic phantoms, containing realistic internal tissue texture, aimed at fine tuning the DL model for clinical applications. The MC simulations generated scatter and primary maps per projection angle for a wide-angle DBT system. Both datasets were used to train (using 7680 projections from homogeneous phantoms), validate (using 960 and 192 projections from the homogeneous and anthropomorphic phantoms, respectively), and test (using 960 and 48 projections from the homogeneous and anthropomorphic phantoms, respectively) the DL model. The DL output was compared to the corresponding MC ground truth using both quantitative and qualitative metrics, such as mean relative and mean absolute relative differences (MRD and MARD), and to previously-published scatter-to-primary (SPR) ratios for similar breast phantoms. The scatter corrected DBT reconstructions were evaluated by analyzing the obtained linear attenuation values and by visual assessment of corrected projections in a clinical dataset. The time required for training and prediction per projection, as well as the time it takes to produce scatter-corrected projection images, were also tracked.

RESULTS

The quantitative comparison between DL scatter predictions and MC simulations showed a median MRD of 0.05% (interquartile range (IQR), -0.04% to 0.13%) and a median MARD of 1.32% (IQR, 0.98% to 1.85%) for homogeneous phantom projections and a median MRD of -0.21% (IQR, -0.35% to -0.07%) and a median MARD of 1.43% (IQR, 1.32% to 1.66%) for the anthropomorphic phantoms. The SPRs for different breast thicknesses and at different projection angles were within ± 15% of the previously-published ranges. The visual assessment showed good prediction capabilities of the DL model with a close match between MC and DL scatter estimates, as well as between DL-based scatter corrected and anti-scatter grid corrected cases. The scatter correction improved the accuracy of the reconstructed linear attenuation of adipose tissue, reducing the error from -16% and -11% to -2.3% and 4.4% for an anthropomorphic digital phantom and clinical case with similar breast thickness, respectively. The DL model training took 40 min and prediction of a single projection took less than 0.01 s. Generating scatter corrected images took 0.03 s per projection for clinical exams and 0.16 s for one entire projection set.

CONCLUSIONS

This DL-based method for estimating the scatter signal in DBT projections is fast and accurate, paving the way for future quantitative applications.

A feasibility study on deep-neural-network-based dose-neutral dual-energy digital breast tomosynthesis

BACKGROUND

Diagnostic performance based on x-ray breast imaging is subject to breast density. Although digital breast tomosynthesis (DBT) is reported to outperform conventional mammography in denser breasts, mass detection and malignancy characterization are often considered challenging yet.

PURPOSE

As an improved diagnostic solution to the dense breast cases, we propose a dual-energy DBT imaging technique that enables breast compositional imaging at comparable scanning time and patient dose compared to the conventional single-energy DBT.

METHODS

The proposed dual-energy DBT acquires projection data by alternating two different energy spectra. Then, we synthesize unmeasured projection data using a deep neural network that exploits the measured projection data and adjacent projection data obtained under the other x-ray energy spectrum. For material decomposition, we estimate partial path lengths of an x-ray through water, lipid, and protein from the measured and the synthesized projection data with the object thickness information. After material decomposition in the projection domain, we reconstruct material-selective DBT images. The deep neural network is trained with the numerical breast phantoms. A pork meat phantom is scanned with a prototype dual-energy DBT system to demonstrate the feasibility of the proposed imaging method.

RESULTS

The developed deep neural network successfully synthesized missing projections. Material-selective images reconstructed from the synthesized data present comparable compositional contrast of the cancerous masses compared with those from the fully measured data.

CONCLUSIONS

The proposed dual-energy DBT scheme is expected to substantially contribute to enhancing mass malignancy detection accuracy particularly in dense breasts.

Deep Convolutional Neural Network With Adversarial Training for Denoising Digital Breast Tomosynthesis Images

Digital breast tomosynthesis (DBT) is a quasi-three-dimensional imaging modality that can reduce false negatives and false positives in mass lesion detection caused by overlapping breast tissue in conventional two-dimensional (2D) mammography. The patient dose of a DBT scan is similar to that of a single 2D mammogram, while acquisition of each projection view adds detector readout noise. The noise is propagated to the reconstructed DBT volume, possibly obscuring subtle signs of breast cancer such as microcalcifications (MCs). This study developed a deep convolutional neural network (DCNN) framework for denoising DBT images with a focus on improving the conspicuity of MCs as well as preserving the ill-defined margins of spiculated masses and normal tissue textures. We trained the DCNN using a weighted combination of mean squared error (MSE) loss and adversarial loss. We configured a dedicated x-ray imaging simulator in combination with digital breast phantoms to generate realistic in silico DBT data for training. We compared the DCNN training between using digital phantoms and using real physical phantoms. The proposed denoising method improved the contrast-to-noise ratio (CNR) and detectability index (d') of the simulated MCs in the validation phantom DBTs. These performance measures improved with increasing training target dose and training sample size. Promising denoising results were observed on the transferability of the digital-phantom-trained denoiser to DBT reconstructed with different techniques and on a small independent test set of human subject DBT images.

A semi-empirical model for scatter field reduction in digital mammography

X-ray mammography is the gold standard technique in breast cancer screening programmes. One of the main challenges that mammography is still facing is scattered radiation, which degrades the quality of the image and complicates the diagnosis process. Anti-scatter grids, the main standard physical scattering reduction technique, have some unresolved challenges as they increase the dose delivered to the patient, do not remove all the scattered radiation and increase the cost of the equipment. Alternative scattering reduction methods based on post-processing algorithms, have lately been under investigation. This study is concerned with the use of image post-processing to reduce the scatter contribution in the image, by convolving the primary plus scatter image with kernels obtained from simplified Monte Carlo (MC) simulations. The proposed semi-empirical approach uses up to five thickness-dependant symmetric kernels to accurately estimate the scatter contribution of different areas of the image. Single breast thickness-dependant kernels can over-estimate the scatter signal up to 60%, while kernels adapting to local variations have to be modified for each specific case adding high computational costs. The proposed method reduces the uncertainty to a 4%-10% range for a 35-70 mm breast thickness range, making it a very efficient, case-independent scatter modelling technique. To test the robustness of the method, the scattered corrected image has been successfully compared against full MC simulations for a range of breast thicknesses. In addition, clinical images of the 010A CIRS phantom were acquired with a mammography system with and without the presence of the anti-scatter grid. The grid-less images were post-processed and their quality was compared against the grid images, by evaluating the contrast-to-noise ratio and variance ratio using several test objects, which simulate calcifications and tumour masses. The results obtained show that the method reduces the scatter to similar levels than the anti-scatter grids.

Analytical Low-Dose CBCT Reconstruction Using Non-local Total Variation Regularization for Image Guided Radiation Therapy

Purpose: Conventional iterative low-dose CBCT reconstruction techniques are slow and tend to over-smooth edges through uniform weighting of the image penalty gradient. In this study, we present a non-iterative analytical low-dose CBCT reconstruction technique by restoring the noisy low-dose CBCT projection with the non-local total variation (NLTV) method. Methods: We modeled the low-dose CBCT reconstruction as recovering high quality, high-dose CBCT x-ray projections (100 kVp, 1.6 mAs) from low-dose, noisy CBCT x-ray projections (100 kVp, 0.1 mAs). The restoration of CBCT projections was performed using the NLTV regularization method. In NLTV, the x-ray image is optimized by minimizing an energy function that penalizes gray-level difference between pair of pixels between noisy x-ray projection and denoising x-ray projection. After the noisy projection is restored by NLTV regularization, the standard FDK method was applied to generate the final reconstruction output. Results: Significant noise reduction was achieved comparing to original, noisy inputs while maintaining the image quality comparable to the high-dose CBCT projections. The experimental validations show the proposed NLTV algorithm can robustly restore the noise level of x-ray projection images while significantly improving the overall image quality. The improvement in normalized mean square error (NMSE) and peak signal-to-noise ratio (PSNR) measured from the non-local total variation-gradient projection (NLTV-GPSR) algorithm is noticeable compared to that of uncorrected low-dose CBCT images. Moreover, the difference of CNRs from the gains from the proposed algorithm is noticeable and comparable to high-dose CBCT. Conclusion: The proposed method successfully restores noise degraded, low-dose CBCT projections to high-dose projection quality. Such an outcome is a considerable improvement to the reconstruction result compared to the FDK-based method. In addition, a significant reduction in reconstruction time makes the proposed algorithm more attractive. This demonstrates the potential use of the proposed algorithm for clinical practice in radiotherapy.

Simple method for computing scattered radiation in breast tomosynthesis

PURPOSE

Virtual clinical trials (VCT) are a powerful imaging tool that can be used to investigate digital breast tomosynthesis (DBT) technology. In this work, a fast and simple method is proposed to estimate the two-dimensional distribution of scattered radiation which is needed when simulating DBT geometries in VCTs.

METHODS

Monte Carlo simulations are used to precalculate scatter-to-primary ratio (SPR) for a range of low-resolution homogeneous phantoms. The resulting values can be used to estimate the two-dimensional (2D) distribution of scattered radiation arising from inhomogeneous anthropomorphic phantoms used in VCTs. The method has been validated by comparing the values of the scatter thus obtained against the results of direct Monte Carlo simulation for three different types of inhomogeneous anthropomorphic phantoms.

RESULTS

Differences between the proposed scatter field estimation method and the ground truth data for the OPTIMAM phantom had an average modulus and standard deviation of over the projected breast area of 2.4 ± 0.9% (minimum -17.0%, maximum 27.7%). The corresponding values for the University of Pennsylvania and Duke University breast phantoms were 1.8 ± 0.1% (minimum -8.7%, maximum 8.0%) and 5.1 ± 0.1% (minimum -16.2%, maximum 7.4%), respectively.

CONCLUSIONS

The proposed method, which has been validated using three of the most common breast models, is a useful tool for accurately estimating scattered radiation in VCT schemes used to study current designs of DBT system.

Detector Blur and Correlated Noise Modeling for Digital Breast Tomosynthesis Reconstruction

This paper describes a new image reconstruction method for digital breast tomosynthesis (DBT). The new method incorporates detector blur into the forward model. The detector blur in DBT causes correlation in the measurement noise. By making a few approximations that are reasonable for breast imaging, we formulated a regularized quadratic optimization problem with a data-fit term that incorporates models for detector blur and correlated noise (DBCN). We derived a computationally efficient separable quadratic surrogate (SQS) algorithm to solve the optimization problem that has a non-diagonal noise covariance matrix. We evaluated the SQS-DBCN method by reconstructing DBT scans of breast phantoms and human subjects. The contrast-to-noise ratio and sharpness of microcalcifications were analyzed and compared with those by the simultaneous algebraic reconstruction technique. The quality of soft tissue lesions and parenchymal patterns was examined. The results demonstrate the potential to improve the image quality of reconstructed DBT images by incorporating the system physics model. This paper is a first step toward model-based iterative reconstruction for DBT.

The compressed breast during mammography and breast tomosynthesis: in vivo shape characterization and modeling

To characterize and develop a patient-based 3D model of the compressed breast undergoing mammography and breast tomosynthesis. During this IRB-approved, HIPAA-compliant study, 50 women were recruited to undergo 3D breast surface imaging with structured light (SL) during breast compression, along with simultaneous acquisition of a tomosynthesis image. A pair of SL systems were used to acquire 3D surface images by projecting 24 different patterns onto the compressed breast and capturing their reflection off the breast surface in approximately 12-16 s. The 3D surface was characterized and modeled via principal component analysis. The resulting surface model was combined with a previously developed 2D model of projected compressed breast shapes to generate a full 3D model. Data from ten patients were discarded due to technical problems during image acquisition. The maximum breast thickness (found at the chest-wall) had an average value of 56 mm, and decreased 13% towards the nipple (breast tilt angle of 5.2°). The portion of the breast not in contact with the compression paddle or the support table extended on average 17 mm, 18% of the chest-wall to nipple distance. The outermost point along the breast surface lies below the midline of the total thickness. A complete 3D model of compressed breast shapes was created and implemented as a software application available for download, capable of generating new random realistic 3D shapes of breasts undergoing compression. Accurate characterization and modeling of the breast curvature and shape was achieved and will be used for various image processing and clinical tasks.

Segmented separable footprint projector for digital breast tomosynthesis and its application for subpixel reconstruction

PURPOSE

Digital forward and back projectors play a significant role in iterative image reconstruction. The accuracy of the projector affects the quality of the reconstructed images. Digital breast tomosynthesis (DBT) often uses the ray-tracing (RT) projector that ignores finite detector element size. This paper proposes a modified version of the separable footprint (SF) projector, called the segmented separable footprint (SG) projector, that calculates efficiently the Radon transform mean value over each detector element. The SG projector is specifically designed for DBT reconstruction because of the large height-to-width ratio of the voxels generally used in DBT. This study evaluates the effectiveness of the SG projector in reducing projection error and improving DBT reconstruction quality.

METHODS

We quantitatively compared the projection error of the RT and the SG projector at different locations and their performance in regular and subpixel DBT reconstruction. Subpixel reconstructions used finer voxels in the imaged volume than the detector pixel size. Subpixel reconstruction with RT projector uses interpolated projection views as input to provide adequate coverage of the finer voxel grid with the traced rays. Subpixel reconstruction with the SG projector, however, uses the measured projection views without interpolation. We simulated DBT projections of a test phantom using CatSim (GE Global Research, Niskayuna, NY) under idealized imaging conditions without noise and blur, to analyze the effects of the projectors and subpixel reconstruction without other image degrading factors. The phantom contained an array of horizontal and vertical line pair patterns (1 to 9.5 line pairs/mm) and pairs of closely spaced spheres (diameters 0.053 to 0.5 mm) embedded at the mid-plane of a 5-cm-thick breast tissue-equivalent uniform volume. The images were reconstructed with regular simultaneous algebraic reconstruction technique (SART) and subpixel SART using different projectors. The resolution and contrast of the test objects in the reconstructed images and the computation times were compared under different reconstruction conditions.

RESULTS

The SG projector reduced the projector error by 1 to 2 orders of magnitude at most locations. In the worst case, the SG projector still reduced the projection error by about 50%. In the DBT reconstructed slices parallel to the detector plane, the SG projector not only increased the contrast of the line pairs and spheres but also produced more smooth and continuous reconstructed images, whereas the discrete and sparse nature of the RT projector caused artifacts appearing as patterned noise. For subpixel reconstruction, the SG projector significantly increased object contrast and computation speed, especially for high subpixel ratios, compared with the RT projector implemented with accelerated Siddon's algorithm. The difference in the depth resolution among the projectors is negligible under the conditions studied. Our results also demonstrated that subpixel reconstruction can improve the spatial resolution of the reconstructed images, and can exceed the Nyquist limit of the detector under some conditions.

CONCLUSIONS

The SG projector was more accurate and faster than the RT projector. The SG projector also substantially reduced computation time and improved the image quality for the tomosynthesized images with and without subpixel reconstruction.

Improvements of an objective model of compressed breasts undergoing mammography: Generation and characterization of breast shapes

PURPOSE

To develop a set of accurate 2D models of compressed breasts undergoing mammography or breast tomosynthesis, based on objective analysis, to accurately characterize mammograms with few linearly independent parameters, and to generate novel clinically realistic paired cranio-caudal (CC) and medio-lateral oblique (MLO) views of the breast.

METHODS

We seek to improve on an existing model of compressed breasts by overcoming detector size bias, removing the nipple and non-mammary tissue, pairing the CC and MLO views from a single breast, and incorporating the pectoralis major muscle contour into the model. The outer breast shapes in 931 paired CC and MLO mammograms were automatically detected with an in-house developed segmentation algorithm. From these shapes three generic models (CC-only, MLO-only, and joint CC/MLO) with linearly independent components were constructed via principal component analysis (PCA). The ability of the models to represent mammograms not used for PCA was tested via leave-one-out cross-validation, by measuring the average distance error (ADE).

RESULTS

The individual models based on six components were found to depict breast shapes with accuracy (mean ADE-CC = 0.81 mm, ADE-MLO = 1.64 mm, ADE-Pectoralis = 1.61 mm), outperforming the joint CC/MLO model (P ≤ 0.001). The joint model based on 12 principal components contains 99.5% of the total variance of the data, and can be used to generate new clinically realistic paired CC and MLO breast shapes. This is achieved by generating random sets of 12 principal components, following the Gaussian distributions of the histograms of each component, which were obtained from the component values determined from the images in the mammography database used.

CONCLUSION

Our joint CC/MLO model can successfully generate paired CC and MLO view shapes of the same simulated breast, while the individual models can be used to represent with high accuracy clinical acquired mammograms with a small set of parameters. This is the first step toward objective 3D compressed breast models, useful for dosimetry and scatter correction research, among other applications.

Estimating scatter from sparsely measured primary signal

Scatter radiation severely degrades the image quality. Measurement-based scatter correction methods sample the scatter signal at sparsely distributed points, from which the scatter profile is estimated and deterministically removed from the projection image. The estimation of the scatter profile is generally done through a spline interpolation and the resulting scatter profile is quite smooth. Consequently, the noise is intact and the signal-to-noise ratio is reduced in the projection image after scatter correction, leading to image artifacts and increased noise in the reconstruction images. We propose a simple and effective method, referred to as filtered scatter-to-primary ratio ([Formula: see text]-SPR) estimation, to estimate the scatter profile using the sparsely sampled scatter signal. Using the primary sampling device and the stationary digital tomosynthesis systems previously developed in our lab, we evaluated and compared the [Formula: see text]-SPR method in comparison with existing methods in terms of contrast ratio, signal difference-to-noise ratio, and modulation transfer function. A significant improvement in image quality is observed in both the projection and the reconstruction images using the proposed method.