James G. Mainprize

Model-based PSF and MTF estimation and validation from skeletal clinical CT images

PURPOSE A method was developed to correct for systematic errors in estimating the thickness of thin bones due to image blurring in CT images using bone interfaces to estimate the point-spread-function (PSF). This study validates the accuracy of the PSFs estimated using said method from various clinical CT images featuring cortical bones. METHODS Gaussian PSFs, characterized by a different extent in the z (scan) direction than in the x and y directions were obtained using our method from 11 clinical CT scans of a cadaveric craniofacial skeleton. These PSFs were estimated for multiple combinations of scanning parameters and reconstruction methods. The actual PSF for each scan setting was measured using the slanted-slit technique within the image slice plane and the longitudinal axis. The Gaussian PSF and the corresponding modulation transfer function (MTF) are compared against the actual PSF and MTF for validation. RESULTS The differences (errors) between the actual and estimated full-width half-max (FWHM) of the PSFs were 0.09 ± 0.05 and 0.14 ± 0.11 mm for the xy and z axes, respectively. The overall errors in the predicted frequencies measured at 75%, 50%, 25%, 10%, and 5% MTF levels were 0.06 ± 0.07 and 0.06 ± 0.04 cycles/mm for the xy and z axes, respectively. The accuracy of the estimates was dependent on whether they were reconstructed with a standard kernel (Toshibas FC68, mean error of 0.06 ± 0.05 mm, MTF mean error 0.02 ± 0.02 cycles/mm) or a high resolution bone kernel (Toshibas FC81, PSF FWHM error 0.12 ± 0.03 mm, MTF mean error 0.09 ± 0.08 cycles/mm). CONCLUSIONS The method is accurate in 3D for an image reconstructed using a standard reconstruction kernel, which conforms to the Gaussian PSF assumption but less accurate when using a high resolution bone kernel. The method is a practical and self-contained means of estimating the PSF in clinical CT images featuring cortical bones, without the need phantoms or any prior knowledge about the scanner-specific parameters.

Quantifying masking in clinical mammograms via local detectability of simulated lesions

PURPOSE High mammographic density is known to be associated with decreased sensitivity of mammography. Recent changes in the BI-RADS density assessment address the effect of masking by densities, but the BI-RADS assessment remains qualitative and achieves only moderate agreement between radiologists. An automated, quantitative algorithm that estimates the likelihood of masking of simulated masses in a mammogram by dense tissue has been developed. The algorithm considers both the effects of loss of contrast due to density and the distracting texture or appearance of dense tissue. METHODS A local detectability (dL) map is created by tessellating the mammograms into overlapping regions of interest (ROIs), for which the detectability by a non-prewhitening observer is computed using local estimates of the noise power spectrum and volumetric breast density (VBD). The dL calculation was validated in a 4-alternative forced-choice observer study on the ROIs of 150 craniocaudal digital mammograms. The dL metric was compared against the inverse threshold contrast, (ΔμT)(-1) from the observer study, the anatomic noise parameter β, the radiologists BI-RADS density category, and a validated measure of VBD (Cumulus). RESULTS The mean dL had a high correlation of r = 0.915 and r = 0.699 with (ΔμT)(-1) in the computerized and human observer study, respectively. In comparison, the local VBD estimate had a low correlation of 0.538 with (ΔμT)(-1). The mean dL had a correlation of 0.663, 0.835, and 0.696 with BI-RADS density, β, and Cumulus VBD, respectively. CONCLUSIONS The proposed dL metric may be useful in characterizing the potential for lesion masking by dense tissue. Because it uses information about the anatomic noise or tissue appearance, it is more closely linked to lesion detectability than VBD metrics.

Towards a Quantitative Measure of Radiographic Masking by Dense Tissue in Mammography

The detection sensitivity of screening mammography is reduced for dense breasts where the appearance of fibroglandular tissue can mask suspicious lesions. A measure of the degree of masking expected for a mammogram could be useful for informing the decision to direct some women to supplemental imaging procedures not affected by density. Here, we present an adaptation of a model observer to estimate the detection task SNR, d local, of a lesion embedded in various portions of the breast to indicate the level of detection difficulty. Rank correlation of mean mammogram d local with density category is ρ=–0.58. Correlation of fractional area of mammograms with low d local < 2 versus density category is ρ=0.61. This suggests that a metric based on d local may be useful in quantifying masking effects of breast density.

Technical Note: Robust measurement of the slice-sensitivity profile in breast tomosynthesis

PURPOSE The purpose of this work is to improve the repeatability of the measurement of the slice-sensitivity profile (SSP) in reconstructed breast tomosynthesis volumes. METHODS A grid of aluminum ball-bearings (BBs) within a PMMA phantom was imaged on breast tomosynthesis systems from three different manufacturers. The full-width half-maximum (FWHM) values were measured for the SSPs of the BBs in the reconstructed volumes. The effect of transforming the volumes from a Cartesian coordinate system (CCS) to a cone-beam coordinate system (CBCS) on the variability in the FWHM values was assessed. RESULTS Transforming the volumes from a CCS to a CBCS before measuring the SSPs reduced the coefficient of variation (COV) in the measurements of FWHM in repeated measurements by 56% and reduced the dependence of the FWHM values on the location of the BBs within the reconstructed volume by 76%. CONCLUSIONS Measuring the SSP in the volumes in a CBCS improves the robustness of the measurement.

Pre-clinical evaluation of tumour angiogenesis with contrast-enhanced breast tomosynthesis

Contrast-enhanced digital breast tomosynthesis (CE DBT) has been proposed to image the effects of tumour angiogenesis. In this work we evaluate the relationship between CE DBT image signal and histopathology in an animal tumour model to provide evidence for the underlying basis for signal enhancement. A VX2 carcinoma was induced in the hind leg of 8 rabbits and grown for up to 3 weeks. Projection images from a 60 s contrast-enhanced CT acquisition were used to reconstruct CE DBT volumes. Fiducial markers implanted in the tumour provided a means for registration between images and stained whole-mount sections. The relationship between CE DBT image signal and angiogenesis marker expression was determined. A correlation was found between CE DBT image signal and dextran extravasation, which strengthened during washout, while no relationship was observed with CD31 staining. These results suggest that for clinical CE DBT, washout phase imaging will provide information on vascular permeability.

Simultaneous deblurring and iterative reconstruction of CBCT for image guided brain radiosurgery

One of the limiting factors in cone-beam CT (CBCT) image quality is system blur, caused by detector response, x-ray source focal spot size, azimuthal blurring, and reconstruction algorithm. In this work, we develop a novel iterative reconstruction algorithm that improves spatial resolution by explicitly accounting for image unsharpness caused by different factors in the reconstruction formulation. While the model-based iterative reconstruction techniques use prior information about the detector response and x-ray source, our proposed technique uses a simple measurable blurring model. In our reconstruction algorithm, denoted as simultaneous deblurring and iterative reconstruction (SDIR), the blur kernel can be estimated using the modulation transfer function (MTF) slice of the CatPhan phantom or any other MTF phantom, such as wire phantoms. The proposed image reconstruction formulation includes two regularization terms: (1) total variation (TV) and (2) nonlocal regularization, solved with a split Bregman augmented Lagrangian iterative method. The SDIR formulation preserves edges, eases the parameter adjustments to achieve both high spatial resolution and low noise variances, and reduces the staircase effect caused by regular TV-penalized iterative algorithms. The proposed algorithm is optimized for a point-of-care head CBCT unit for image-guided radiosurgery and is tested with CatPhan phantom, an anthropomorphic head phantom, and 6 clinical brain stereotactic radiosurgery cases. Our experiments indicate that SDIR outperforms the conventional filtered back projection and TV penalized simultaneous algebraic reconstruction technique methods (represented by adaptive steepest-descent POCS algorithm, ASD-POCS) in terms of MTF and line pair resolution, and retains the favorable properties of the standard TV-based iterative reconstruction algorithms in improving the contrast and reducing the reconstruction artifacts. It improves the visibility of the high contrast details in bony areas and the brain soft-tissue. For example, the results show the ventricles and some brain folds become visible in SDIR reconstructed images and the contrast of the visible lesions is effectively improved. The line-pair resolution was improved from 12 line-pair/cm in FBP to 14 line-pair/cm in SDIR. Adjusting the parameters of the ASD-POCS to achieve 14 line-pair/cm caused the noise variance to be higher than the SDIR. Using these parameters for ASD-POCS, the MTF of FBP and ASD-POCS were very close and equal to 0.7 mm-1 which was increased to 1.2 mm-1 by SDIR, at half maximum.

Segmenting pectoralis muscle on digital mammograms by a Markov random field-maximum a posteriori model

Abstract. Accurate and automatic segmentation of the pectoralis muscle is essential in many breast image processing procedures, for example, in the computation of volumetric breast density from digital mammograms. Its segmentation is a difficult task due to the heterogeneity of the region, neighborhood complexities, and shape variability. The segmentation is achieved by pixel classification through a Markov random field (MRF) image model. Using the image intensity feature as observable data and local spatial information as a priori, the posterior distribution is estimated in a stochastic process. With a variable potential component in the energy function, by the maximum a posteriori (MAP) estimate of the labeling image, given the image intensity feature which is assumed to follow a Gaussian distribution, we achieved convergence properties in an appropriate sense by Metropolis sampling the posterior distribution of the selected energy function. By proposing an adjustable spatial constraint, the MRF-MAP model is able to embody the shape requirement and provide the required flexibility for the model parameter fitting process. We demonstrate that accurate and robust segmentation can be achieved for the curving-triangle-shaped pectoralis muscle in the medio-lateral-oblique (MLO) view, and the semielliptic-shaped muscle in cranio-caudal (CC) view digital mammograms. The applicable mammograms can be either “For Processing” or “For Presentation” image formats. The algorithm was developed using 56 MLO-view and 79 CC-view FFDM “For Processing” images, and quantitatively evaluated against a random selection of 122 MLO-view and 173 CC-view FFDM images of both presentation intent types.

Impact of image acquisition timing on image quality for dual energy contrast-enhanced breast tomosynthesis

Dual-energy contrast-enhanced digital breast tomosynthesis (DE CE-DBT) image quality is affected by a large parameter space including the tomosynthesis acquisition geometry, imaging technique factors, the choice of reconstruction algorithm, and the subject breast characteristics. The influence of most of these factors on reconstructed image quality is well understood for DBT. However, due to the contrast agent uptake kinetics in CE imaging, the subject breast characteristics change over time, presenting a challenge for optimization . In this work we experimentally evaluate the sensitivity of the reconstructed image quality to timing of the low-energy and high-energy images and changes in iodine concentration during image acquisition. For four contrast uptake patterns, a variety of acquisition protocols were tested with different timing and geometry. The influence of the choice of reconstruction algorithm (SART or FBP) was also assessed. Image quality was evaluated in terms of the lesion signal-difference-to-noise ratio (LSDNR) in the central slice of DE CE-DBT reconstructions. Results suggest that for maximum image quality, the low- and high-energy image acquisitions should be made within one x-ray tube sweep, as separate low- and high-energy tube sweeps can degrade LSDNR. In terms of LSDNR per square-root dose, the image quality is nearly equal between SART reconstructions with 9 and 15 angular views, but using fewer angular views can result in a significant improvement in the quantitative accuracy of the reconstructions due to the shorter imaging time interval.

Design and validation of a mathematical breast phantom for contrast-enhanced digital mammography

In contrast-enhanced digital mammography (CEDM) an iodinated contrast agent is employed to increase lesion contrast and to provide tissue functional information. Here, we present the details of a software phantom that can be used as a tool for the simulation of CEDM images, and compare the degree of anatomic noise present in images simulated using the phantom to that associated with breast parenchyma in clinical CEDM images. Such a phantom could be useful for multiparametric investigations including characterization of CEDM imaging performance and system optimization. The phantom has a realistic mammographic appearance based on a clustered lumpy background and models contrast agent uptake according to breast tissue physiology. Fifty unique phantoms were generated and used to simulate regions of interest (ROI) of pre-contrast images and logarithmically subtracted CEDM images using monoenergetic ray tracing. Power law exponents, β, were used as a measure of anatomic noise and were determined using a linear least-squares fit to log-log plots of the square of the modulus of radially averaged image power spectra versus spatial frequency. The power spectra for ROI selected from regions of normal parenchyma in 10 pairs of clinical CEDM pre-contrast and subtracted images were also measured for comparison with the simulated images. There was good agreement between the measured β in the simulated CEDM images and the clinical images. The values of β were consistently lower for the logarithmically subtracted CEDM images compared to the pre-contrast images, indicating that the subtraction process reduced anatomical noise.

A task-based quality control metric for digital mammography

A reader study was conducted to tune the parameters of an observer model used to predict the detectability index (dʹ ) of test objects as a task-based quality control (QC) metric for digital mammography. A simple test phantom was imaged to measure the model parameters, namely, noise power spectrum,modulation transfer function and test-object contrast. These are then used ina non-prewhitening observer model, incorporating an eye-filter and internal noise, to predict dʹ. The model was tuned by measuring dʹ of discs in a four-alternative forced choice reader study. For each disc diameter, dʹ was used to estimate the threshold thicknesses for detectability. Data were obtained for six types of digital mammography systems using varying detector technologies and x-ray spectra. A strong correlation was found between measured and modeled values of dʹ, with Pearson correlation coefficient of 0.96. Repeated measurements from separate images of the test phantom show an average coefficient of variation in dʹ for different systems between 0.07 and 0.10. Standard deviations in the threshold thickness ranged between 0.001 and 0.017 mm. The model is robust and the results are relatively system independent, suggesting that observer model dʹ shows promise as a cross platform QC metric for digital mammography.