SU-E-I-02: Size-Dependent Computed Tomography Histogram Analysis: Towards Breast Tissue Segmentation.

PURPOSE: To examine the effects of object size on scatter-corrected CT histograms to be used in threshold-based tissue segmentation. METHODS: A polyethylene cone filled with various concentrations of water and methanol mixtures, simulating glandular and adipose tissues, were imaged with our quasi-monochromatic dedicated breast CT, and scatter corrected using beam-stop array measurements. Images were reconstructed using iterative OSC, and individual coronal slices along the central axis of the cone were analyzed, with radii ranging from 3.25-6.25cm. Histograms from each slice were fit with two Gaussians (fluid filling and cone material) using nonlinear least squares methods, and the corresponding standard deviation and peak centroid of each filled material were evaluated as a function of object radius. Identical methods were applied to dedicated breast CT images of four patients for clinical comparison. RESULTS: Analysis of phantoms and breast data indicates low correlation between the standard deviation and object diameter. The centroids of the Gaussian peaks demonstrate an inverse linear relationship with increasing object size, independent of object material. The clinical datasets show a similar linear relationship between centroids and breast radius. CONCLUSIONS: Data indicate that using a linear combination of Gaussian distribution functions to segment breast tissue in scatter corrected, quasi-monochromatic cone beam dedicated breast CT is possible. An object size-independent variation of attenuation values may allow for consistent restraints on initial fit parameters, resulting in improved confidence using Gaussian curve fitting. Appropriate scaling of the size-dependent volume slices, or independent slice analysis, is necessary to minimize binning variability for accurate tissue segmentation using Gaussian curve fitting. This work is supported by National Institutes of Health (R01-CA096821) and NIH Training Grant (T32-EB007185). MPT is the inventor of this hybrid breast imaging technology, and is named as an inventor on the patent for this technology assigned to Duke (US Pat. #7,609,808). If this technology becomes commercially successful, MPT and Duke could benefit financially.

Characterizing the contribution of cardiac and hepatic uptake in dedicated breast SPECT using tilted trajectories.

A small field of view, high resolution gamma camera has been integrated into a dedicated breast, single photon emission computed tomography (SPECT) device. The detector can be flexibly positioned relative to the breast and image beyond the chest wall, allowing the system to capture direct views of the heart and liver. The incomplete sampling of these organs creates artifacts in reconstructed images, complicating lesion detection. To understand the limits imposed on a 3D acquisition trajectory, sequential tilted trajectories at increasing polar tilt are utilized to collect data of anthropomorphic phantoms filled with aqueous (99m)Tc in a clinically realistic concentration ratio. The counts collected per projection between different scans and the SNR, contrast and resolution (FWHM) of two hot lesions were compared. As expected, the counts per projection increased when the camera had direct views of the heart and liver, but remained relatively constant at other angles. The SNR, contrast and FWHM were more affected by the insufficient sampling of the data by the large polar angles than by the cardiac and hepatic activity. An upper bound on polar tilt for each azimuthal position reduces the artifacts in the reconstructed images. Such trajectories were implemented to show artifact-free reconstructed images.

Observer detection limits for a dedicated SPECT breast imaging system.

An observer-based contrast-detail study is performed in an effort to evaluate the limits of object detectability using a dedicated CZT-based breast SPECT imaging system under various imaging conditions. A custom geometric contrast-resolution phantom was developed that can be used for both positive ('hot') and negative contrasts ('cold'). The 3 cm long fillable tubes are arranged in six sectors having equal inner diameters ranging from 1 mm to 6 mm with plastic wall thicknesses of <0.25 mm, on a pitch of twice their inner diameters. Scans of the activity filled tubes using simple circular trajectories are obtained in a 215 mL uniform water filled cylinder, varying the rod:background concentration ratios from 10:1 to 1:10 simulating a large range of biological uptake ratios. The rod phantom is then placed inside a non-uniformly shaped 500 mL breast phantom and scans are again acquired using both simple and complex 3D trajectories for similarly varying contrasts. Summed slice and contiguous multi-slice images are evaluated by five independent readers, identifying the smallest distinguishable rod for each concentration and experimental setup. Linear and quadratic regression is used to compare the resulting contrast-detail curves. Results indicate that in a moderately low-noise 500 mL background, using the SPECT camera having 2.5 mm intrinsic pixels, the mean detectable rod was approximately 3.4 mm at a 10:1 ratio, degrading to approximately 5.2 mm with the 2.5:1 concentration ratio. The smallest object detail was observed using a 45 degrees tilted trajectory acquisition. The complex 3D projected sine wave acquisition, however, had the most consistent combined intra- and inter-observer results, making it potentially the best imaging approach for consistent results.

Evaluation of tilted cone-beam CT orbits in the development of a dedicated hybrid mammotomograph.

A compact dedicated 3D breast SPECT-CT (mammotomography) system is currently under development. In its initial prototype, the cone-beam CT sub-system is restricted to a fixed-tilt circular rotation around the patient's pendant breast. This study evaluated stationary-tilt angles for the CT sub-system that will enable maximal volumetric sampling and viewing of the breast and chest wall. Images of geometric/anthropomorphic phantoms were acquired using various fixed-tilt circular and 3D sinusoidal trajectories. The iteratively reconstructed images showed more distortion and attenuation coefficient inaccuracy from tilted cone-beam orbits than from the complex trajectory. Additionally, line profiles illustrated cupping artifacts in planes distal to the central plane of the tilted cone-beam, otherwise not apparent for images acquired with complex trajectories. This indicates that undersampled cone-beam data may be an additional cause of cupping artifacts. High-frequency objects could be distinguished for all trajectories, but their shapes and locations were corrupted by out-of-plane frequency information. Although more acrylic balls were visualized with a fixed-tilt and nearly flat cone-beam at the posterior of the breast, 3D complex trajectories have less distortion and more complete sampling throughout the reconstruction volume. While complex trajectories would ideally be preferred, negatively fixed-tilt source-detector configuration demonstrates minimally distorted patient images.

Experimental spectral measurements of heavy K-edge filtered beams for x-ray computed mammotomography.

A dual modality computed mammotomography (CmT) and single photon emission computed tomography (SPECT) system for dedicated 3D breast imaging is in development. Using heavy K-edge filtration, the CmT component narrows the energy spectrum of the cone-shaped x-ray beam incident on the patient's pendant, uncompressed breast. This quasi-monochromatic beam is expected to improve discrimination of tissue with similar attenuation coefficients while restraining absorbed dose to below that of dual view mammography. Previous simulation studies showed the optimal energy that maximizes dose efficiency for a 50/50% adipose/glandular breast is between 30 and 40 keV. This study experimentally validates these results using pre-breast and post-breast spectral measurements made under tungsten tube voltages between 40 and 100 kVp using filter materials with K-edge values ranging from 15 to 70 keV. Different filter material thicknesses are used, approximately equivalent to the 200th and 500th attenuating value layer (VL) thickness. Cerium (K = 40.4 keV) filtered post-breast spectra for 8-18 cm breasts are measured for a range of breast compositions. Figures of merit include mean beam energy, spectral full-width at tenth-maximum, beam hardening and dose for the range of breast sizes. Measurements corroborate simulation results, indicating that for a given dose, a 200th VL of cerium filtration may have optimal performance in the dedicated mammotomography paradigm.

Performance of dedicated emission mammotomography for various breast shapes and sizes.

We evaluate the effect of breast shape and size and lesion location on a dedicated emission mammotomography system developed in our lab. The hemispherical positioning gantry allows ample flexibility in sampling a pendant, uncompressed breast. Realistic anthropomorphic torso (which includes the upper portion of the arm) and breast phantoms draw attention to the necessity of using unique camera trajectories (orbits) rather than simple circular camera trajectories. We have implemented several novel three-dimensional (3D) orbits with fully contoured radius-of-rotation capability for compensating for the positioning demands that emerge from different breast shapes and sizes. While a general orbit design may remain the same between two different breasts, the absolute polar tilt range and radius-of-rotation range may vary. We have demonstrated that using 3D orbits with increased polar camera tilt, lesions near the chest wall can be visualized for both large and small sized breasts (325 ml to 1,060 ml), for a range of intrinsic contrasts (three to ten times higher activity concentration in the lesion than breast background). Overall, nearly complete 3D acquisition schemes yield image data with relatively high lesion SNRs and contrasts and with minimal distortion of the uncompressed breast shape.

Modeling the axial extension of a transmission line source within iterative reconstruction via multiple transmission sources.

Reconstruction algorithms for transmission tomography have generally assumed that the photons reaching a particular detector bin at a particular angle originate from a single point source. In this paper, we highlight several cases of extended transmission sources, in which it may be useful to approach the estimation of attenuation coefficients as a problem involving multiple transmission point sources. Examined in detail is the case of a fixed transmission line source with a fan-beam collimator. This geometry can result in attenuation images that have significant axial blur. Herein it is also shown, empirically, that extended transmission sources can result in biased estimates of the average attenuation, and an explanation is proposed. The finite axial resolution of the transmission line source configuration is modeled within iterative reconstruction using an expectation-maximization algorithm that was previously derived for estimating attenuation coefficients from single photon emission computed tomography (SPECT) emission data. The same algorithm is applicable to both problems because both can be thought of as involving multiple transmission sources. It is shown that modeling axial blur within reconstruction removes the bias in the average estimated attenuation and substantially improves the axial resolution of attenuation images.

An EM algorithm for estimating SPECT emission and transmission parameters from emissions data only.

A maximum-likelihood (ML) expectation-maximization (EM) algorithm (called EM-IntraSPECT) is presented for simultaneously estimating single photon emission computed tomography (SPECT) emission and attenuation parameters from emission data alone. The algorithm uses the activity within the patient as transmission tomography sources, with which attenuation coefficients can be estimated. For this initial study, EM-IntraSPECT was tested on computer-simulated attenuation and emission maps representing a simplified human thorax as well as on SPECT data obtained from a physical phantom. Two evaluations were performed. First, to corroborate the idea of reconstructing attenuation parameters from emission data, attenuation parameters (mu) were estimated with the emission intensities (lambda) fixed at their true values. Accurate reconstructions of attenuation parameters were obtained. Second, emission parameters lambda and attenuation parameters mu were simultaneously estimated from the emission data alone. In this case there was crosstalk between estimates of lambda and mu and final estimates of lambda and mu depended on initial values. Estimates degraded significantly as the support extended out farther from the body, and an explanation for this is proposed. In the EM-IntraSPECT reconstructed attenuation images, the lungs, spine, and soft tissue were readily distinguished and had approximately correct shapes and sizes. As compared with standard EM reconstruction assuming a fix uniform attenuation map, EM-IntraSPECT provided more uniform estimates of cardiac activity in the physical phantom study and in the simulation study with tight support, but less uniform estimates with a broad support. The new EM algorithm derived here has additional applications, including reconstructing emission and transmission projection data under a unified statistical model.

Analytical versus voxelized phantom representation for Monte Carlo simulation in radiological imaging.

Monte Carlo simulations in nuclear medicine, with accurately modeled photon transport and high-quality random number generators, require precisely defined and often detailed phantoms as an important component in the simulation process. Contemporary simulation models predominantly employ voxel-driven algorithms, but analytical models offer important advantages. We discuss the implementation of ray-solid intersection algorithms in analytical superquadric-based complex phantoms with additional speed-up rejection testing for use in nuclear medicine imaging simulations, and we make comparisons with voxelized counterparts. Comparisons are made with well-known cold rod:sphere and anthropomorphic phantoms. For these complex phantoms, the analytical phantom representations are nominally several orders of magnitude smaller in memory requirements than are voxelized versions. Analytical phantoms facilitate constant distribution parameters. As a consequence of discretizing a continuous surface into finite bins, for example, time-dependent voxelized phantoms can have difficulties preserving accurate volumes of a beating heart. Although virtually no inaccuracy is associated with path calculations in analytical phantoms, the discretization can negatively impact the simulation process and results. Discretization errors are apparent in reconstructed images of cold rod:sphere voxel-based phantoms because of a redistribution of the count densities in the simulated objects. These problems are entirely avoided in analytical phantoms. Voxelized phantoms can accurately model detailed human shapes based on segmented computed tomography (CT) or magnetic resonance imaging (MRI) images, but analytical phantoms offer advantages in time and accuracy for evaluation and investigation of imaging physics and reconstruction algorithms in a straightforward and efficient manner.

Intraoperative probes and imaging probes.

Intraoperative probes have been employed to assist in the detection and removal of tumors for more than 50 years. For a period of about 40 years, essentially every detector type that could be miniaturized had been tested or at least suggested for use as an intraoperative probe. These detectors included basic Geiger-Müller (GM) tubes, scintillation detectors, and even state-of-the-art solid state detectors. The radiopharmaceuticals have progressed from (32)PO(4)(-) injections for brain tumors to sophisticated monoclonal antibodies labeled with iodine-125 for colorectal cancers. The early work was mostly anecdotal, primarily interdisciplinary collaborations between surgeons and physical scientists. These collaborations produced a few publications, but never seemed to result in an ongoing clinical practice. In the mid 1980s, several companies offered basic gamma-detecting intraoperative probes as products. This led to the rapid development of radioimmunoguided surgery (RIGS) and sentinel node detection as regularly practiced procedures to assist in the diagnosis and treatment of cancer. In recent years intraoperative imaging probes have been developed. These devices add the ability to see the details of the detected activity, giving the potential of using the technique in a low-contrast environment. Intraoperative probes are now established as clinical devices, they have a commercial infrastructure to support their continued use, and there is ongoing research, both commercial and academic, that would seem to ensure continued progress and renewed interest in this slowly developing field.