Christian G. Graff

T2 mapping from highly undersampled data by reconstruction of principal component coefficient maps using compressed sensing

Recently, there has been an increased interest in quantitative MR parameters to improve diagnosis and treatment. Parameter mapping requires multiple images acquired with different timings usually resulting in long acquisition times. While acquisition time can be reduced by acquiring undersampled data, obtaining accurate estimates of parameters from undersampled data is a challenging problem, in particular for structures with high spatial frequency content. In this work, principal component analysis is combined with a model‐based algorithm to reconstruct maps of selected principal component coefficients from highly undersampled radial MRI data. This novel approach linearizes the cost function of the optimization problem yielding a more accurate and reliable estimation of MR parameter maps. The proposed algorithm—reconstruction of principal component coefficient maps using compressed sensing—is demonstrated in phantoms and in vivo and compared with two other algorithms previously developed for undersampled data. Magn Reson Med, 2012.

Compressive sensing in medical imaging

The promise of compressive sensing, exploitation of compressibility to achieve high quality image reconstructions with less data, has attracted a great deal of attention in the medical imaging community. At the Compressed Sensing Incubator meeting held in April 2014 at OSA Headquarters in Washington, DC, presentations were given summarizing some of the research efforts ongoing in compressive sensing for x-ray computed tomography and magnetic resonance imaging systems. This article provides an expanded version of these presentations. Sparsity-exploiting reconstruction algorithms that have gained popularity in the medical imaging community are studied, and examples of clinical applications that could benefit from compressive sensing ideas are provided. The current and potential future impact of compressive sensing on the medical imaging field is discussed.

Rapid water and lipid imaging with T2 mapping using a radial IDEAL-GRASE technique

Three‐point Dixon methods have been investigated as a means to generate water and fat images without the effects of field inhomogeneities. Recently, an iterative algorithm (IDEAL, iterative decomposition of water and fat with echo asymmetry and least squares estimation) was combined with a gradient and spin‐echo acquisition strategy (IDEAL‐GRASE) to provide a time‐efficient method for lipid–water imaging with correction for the effects of field inhomogeneities. The method presented in this work combines IDEAL‐GRASE with radial data acquisition. Radial data sampling offers robustness to motion over Cartesian trajectories as well as the possibility of generating high‐resolution T2 maps in addition to the water and fat images. The radial IDEAL‐GRASE technique is demonstrated in phantoms and in vivo for various applications including abdominal, pelvic, and cardiac imaging. Magn Reson Med, 2009.

Automated registration of sequential breath-hold dynamic contrast-enhanced MR images: A comparison of three techniques

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is increasingly in use as an investigational biomarker of response in cancer clinical studies. Proper registration of images acquired at different time points is essential for deriving diagnostic information from quantitative pharmacokinetic analysis of these data. Motion artifacts in the presence of time-varying intensity due to contrast enhancement make this registration problem challenging. DCE-MRI of chest and abdominal lesions is typically performed during sequential breath-holds, which introduces misregistration due to inconsistent diaphragm positions and also places constraints on temporal resolution vis-à-vis free-breathing. In this work, we have employed a computer-generated DCE-MRI phantom to compare the performance of two published methods, Progressive Principal Component Registration and Pharmacokinetic Model-Driven Registration, with Sequential Elastic Registration (SER) to register adjacent time-sample images using a published general-purpose elastic registration algorithm. In all three methods, a 3D rigid-body registration scheme with a mutual information similarity measure was used as a preprocessing step. The DCE-MRI phantom images were mathematically deformed to simulate misregistration, which was corrected using the three schemes. All three schemes were comparably successful in registering large regions of interest (ROIs) such as muscle, liver, and spleen. SER was superior in retaining tumor volume and shape, and in registering smaller but important ROIs such as tumor core and tumor rim. The performance of SER on clinical DCE-MRI data sets is also presented.

A new, open-source, multi-modality digital breast phantom

An anthropomorphic digital breast phantom has been developed with the goal of generating random voxelized breast models that capture the anatomic variability observed in vivo. This is a new phantom and is not based on existing digital breast phantoms or segmentation of patient images. It has been designed at the outset to be modality agnostic (i.e., suitable for use in modeling x-ray based imaging systems, magnetic resonance imaging, and potentially other imaging systems) and open source so that users may freely modify the phantom to suit a particular study. In this work we describe the modeling techniques that have been developed, the capabilities and novel features of this phantom, and study simulated images produced from it. Starting from a base quadric, a series of deformations are performed to create a breast with a particular volume and shape. Initial glandular compartments are generated using a Voronoi technique and a ductal tree structure with terminal duct lobular units is grown from the nipple into each compartment. An additional step involving the creation of fat and glandular lobules using a Perlin noise function is performed to create more realistic glandular/fat tissue interfaces and generate a Cooper’s ligament network. A vascular tree is grown from the chest muscle into the breast tissue. Breast compression is performed using a neo-Hookean elasticity model. We show simulated mammographic and T1-weighted MRI images and study properties of these images.

A novel physical anthropomorphic breast phantom for 2D and 3D x‐ray imaging

Purpose: Physical phantoms are central to the evaluation of 2D and 3D breast‐imaging systems. Currently, available physical phantoms have limitations including unrealistic uniform background structure, large expense, or excessive fabrication time. The purpose of this work is to outline a method for rapidly creating realistic, inexpensive physical anthropomorphic phantoms for use in full‐field digital mammography (FFDM) and digital breast tomosynthesis (DBT). Methods: The phantom was first modeled using analytical expressions and then discretized into voxels of a specified size. The interior of the breast was divided into glandular and adipose tissue classes using Voronoi segmentation, and additional structures like blood vessels, chest muscle, and ligaments were added. The physical phantom was then fabricated from the virtual model in a slice by slice fashion through inkjet printing, using parchment paper and a radiopaque ink containing 33% (I33%) or 25% (I25%) iohexol by volume. Three types of parchment paper (P1, P2, and P3) were examined. The phantom materials were characterized in terms of their effective linear attenuation coefficients (μeff) using full‐field digital mammography (FFDM) and their energy‐dependent linear attenuation coefficients (μ(E)) using a spectroscopic energy discriminating detector system. The printing method was further validated on the basis of accuracy, print consistency, and the reproducibility of ink batches. Results: The μeff of two types of parchment paper were close to that of adipose tissue, with μeff = 0.61 ± 0.05 cm−1 for P1, 0.61 ± 0.04 cm−1 for P2, and 0.57 ± 0.03 cm−1 for adipose tissue. The addition of the iodinated ink increased the effective attenuation to that of glandular tissue, with μeff = 0.89 ± 0.06 cm−1 for P1 + I25% and 0.94 ± 0.06 cm−1 for P1 + I33% compared to 0.90 ± 0.03 cm−1 for glandular tissue. Spectroscopic measurements showed a good match between the parchment paper and reference values for adipose and glandular tissues across photon energies. Good accuracy was found between the model and the printed phantom by comparing a FFDM of the virtual model simulated through Monte Carlo with a real FFDM of the fully printed phantom. High consistency was found over multiple prints, with 3% variability in mean ink signal across various samples. Reproducibility of ink consistency was very high with <1% variation signal from multiple batches of ink. Imaging of the phantom using FFDM and DBT systems showed promising utility for 2D and 3D imaging. Conclusions: A novel, realistic breast phantom can be created using an analytically defined breast model and readily available materials. The work provides a method to fabricate any virtual phantom in a manner that is accurate, inexpensive, easily accessible, and can be made with different materials or breast models.

The Ideal Observer Objective Assessment Metric for Magnetic Resonance Imaging

The ideal Bayesian observer is a mathematical construct which makes optimal use of all statistical information about the object and imaging system to perform a task. Its performance serves as an upper bound on any observers task performance. In this paper a methodology based on the ideal observer for assessing magnetic resonance (MR) acquisition sequences and reconstruction algorithms is developed. The ideal observer in the context of MR imaging is defined and expressions for ideal observer performance metrics are derived. Comparisons are made between the raw-data ideal observer and image-based ideal observer to elucidate the effect of image reconstruction on task performance. Lesion detection tasks are studied in detail via analytical expressions and simulations. The effect of imaging sequence parameters on lesion detectability is shown and the advantages of this methodology over image quality metrics currently in use in MR imaging is demonstrated.

Correcting partial volume effects in biexponential T2 estimation of small lesions

T2 mapping provides a quantitative approach for focal liver lesion characterization. For small lesions, a biexponential model should be used to account for partial volume effects (PVE). However, conventional biexponential fitting suffers from large uncertainty of the fitted parameters when noise is present. The purpose of this work is to develop a more robust method to correct for PVE affecting small lesions.

A physical breast phantom for 2D and 3D x-ray imaging made through inkjet printing

Physical breast phantoms are used for imaging evaluation studies with 2D and 3D breast x-ray systems, serving as surrogates for human patients. However, there is a presently a limited selection of available phantoms that are realistic, in terms of containing the complex tissue architecture of the human breast. In addition, not all phantoms can be successfully utilized for both 2D and 3D breast imaging. Additionally, many of the phantoms are uniform or unrealistic in appearance, expensive, or difficult to obtain. The purpose of this work was to develop a new method to generate realistic physical breast phantoms using easy to obtain and inexpensive materials. First, analytical modeling was used to design a virtual model, which was then compressed using finite element modeling. Next, the physical phantom was realized through inkjet printing with a standard inkjet printer using parchment paper and specialized inks, formulated using silver nanoparticles and a bismuth salt. The printed phantom sheets were then aligned and held together using a custom designed support plate made of PMMA, and imaged on clinical FFDM and DBT systems. Objects of interest were also placed within the phantom to simulate microcalcifications, pathologies that often occur in the breast. The linear attenuation coefficients of the inks and parchment were compared against tissue equivalent samples and found to be similar to breast tissue. The phantom is promising for use in imaging studies and developing QC protocols.

Anthropomorphic breast phantoms for evaluation of FFDM/DBT and breast CT using inkjet printing

The breast phantoms currently available for evaluating full field digital mammography (FFDM), digital breast tomosynthesis (DBT), and breast CT (bCT) systems often lack the complexity present in real breasts. In this work we present a new methodology for creating physical anthropomorphic breast phantoms for use in FFDM, DBT, and dedicated bCT systems using zinc acetate-doped ink. First, an uncompressed virtual phantom was created through analytical modeling. The model represented a breast with 28% fibroglandular density with 13 tissue classes and contained a 5 mm lesion. The breast was binarized to two tissue classes: adipose and fibroglandular tissue. The phantom was then realized through inkjet printing using dye ink doped with zinc acetate for the fibroglandular components and three candidate materials for the adipose background: parchment paper, organic paper, and office paper. The fabrication process was evaluated in terms of material realism and reproducibility using spectroscopy, a clinical FFDM system, and a benchtop bCT system. The linear attenuation coefficient of the doped ink plus parchment paper and parchment paper alone closely matched those of the fibroglandular and adipose tissues, respectively. A methodology for generating anthropomorphic breast phantoms was developed using a novel inkjet printing technique for use in FFDM/DBT, as well as dedicated breast CT systems. A novel uncompressed breast phantom for bCT was fabricated using inexpensive, easily available materials with realistic tissue properties.