Joerg Peter

SPECT breast imaging combining horizontal and vertical axes of rotation

Vertical-axis-of-rotation (VAOR) geometries, in which a gamma camera orbits a pendulous breast of a prone patient, have been proposed for SPECT breast imaging. Dedicated VAOR systems would be advantageous since they allow for minimal radius of rotation around the breast, which improves resolution-sensitivity tradeoffs, and there is less attenuation and scatter between the breast and the camera. However, the torso may not be viewed at enough angles to estimate contamination from torso activity. Also, VAOR geometries may involve camera tilt to scan for tumors near the chest wall. These tilted VAOR projections may be insufficient to establish even breast activity alone. Thus, we investigated the utility of designing VAOR systems to function inside some general-purpose horizontal-axis-of-rotation (HAOR) scanners, with the HAOR scanner orbiting a 192/spl deg/ angular range posteriorly, alleviating the insufficient-data problems. We simulated the expected projection data from six various-sized lesions with different activity concentrations in the breast and axilla, additionally considering the contribution from background activity from the torso. The data were then statistically reconstructed considering 360/spl deg/ HAOR, VAOR plus small additional arcs and combined HAOR-VAOR orbits, and the results were compared with the known activity concentration distribution. Substantial improvement in the noise versus bias was observed in the breast and axilla, an important and otherwise difficult region to image.

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.

Integrating kinetic models for Simulating tumor growth in Monte Carlo Simulation of ECT systems

We have developed an integrated framework for linking tumor growth models directly into a Monte Carlo simulation algorithm for PET and SPECT systems. Tumors are approximated by arbitrarily complex analytically defined five-dimensional (x,y,z,t/sub geometry/,t/sub activity/) compartments which can be placed into tomographic or mathematical phantoms. Various models for tumor growth have been developed or are implemented such as sigmoidal growth according to the Gompertz equation or compartment models for heterogeneous metastatic tumors, as well as model extensions that account for various therapy strategies. With this approach, Monte Carlo simulation studies can be performed repetitively in static or dynamic acquisition mode at any given time of projected tumor growth. This novel simulation technique provides the basis for deriving allometric relationships between growth rate and tumor representation in a sequence of simulated tomographic images.

Quantifiability and image quality in noncontact fluorescence tomography

Non-contact detection schemes for optical or fluorescence tomography offer several advantages compared to classic approaches, most importantly the ability to obtain images with a CCD in the absence of a matching fluid or fiber optics. This allows the acquisition of high density datasets, as well as simplified experimental procedures. Herein we create a unified framework for contact and non-contact detection procedures and present experimental results that show the ability of the non-contact method to quantify the concentration of fluorochromes hidden in turbid media as well as the improvement in image quality between conventional and non-contact detection.

PET-MOT - a novel concept for simultaneous positron and optical tomography in small animals

We present a Monte Carlo feasibility study of a device designed to simultaneously image positron labeled substrates and optical molecular probes in small animals such as mice and rats. The acquisition arrangement constitutes an entirely novel imaging system considering the dual-modality feature as well as the integrated non-contact optical detection principle. In our proposal, the optical scanner does not employ gantry-mounted CCD cameras for detecting projection data. Instead, it uses a cylindrical lattice of micro lens arrays which form an inner optical detector ring while PET detector blocks are mounted in radial extension. The most innovative aspect of the proposed design is the possibility to perform unified simultaneous acquisition, reconstruction, and tracer/molecular-kinetic modeling of dual-labeled emitters. Since regional distribution and time variation of the underlying multi-variate photon distributions are acquisition and subject specific and diversified by variations thereof, combined and simultaneous imaging carries advantageous potential for translational research. Further advantages are less subject encumbrance and identical imaging geometries. The proposed nuclear-optical tomographic imaging system might potentially foster the development of optical image reconstruction algorithms using PET images as a prior to quantify fluorescence and bioluminescence distributions in heterogeneous media in vivo.

Observer studies of cardiac lesion detectability with triple-head 360/spl deg/ versus dual-head 180/spl deg/ SPECT acquisition using simulated projection data

Evaluates cardiac lesion detectability with triple-head 360/spl deg/ versus dual-head 180/spl deg/ myocardial perfusion SPECT scans with equal acquisition time. The channelized Hotelling observer (CHO) and human observers were used. A male and a female voxelized cardiac-torso phantoms were used to generate the 360/spl deg/ and the 180/spl deg/ projection data. A cold lesion was placed in eight different locations of myocardium and had a lesion contrast of 25%. Sixteen time frames of the cardiac cycle were averaged to create the cardiac motion blurring to simulate the clinical ungated scan. Sufficient Poisson noise was added to set the area under the receiver operating characteristic (ROC) curve (A/sub z/) to be between 0.75 and 0.85 in a pilot study and to simulate the clinical case where dual-head 180/spl deg/ and triple-head 360/spl deg/ scans both have the same total acquisition time. For each lesion location, multiple realizations of lesion-present and of lesion-absent data were generated. Five-iteration ordered subsets expectation maximization with eight subsets was used to reconstruct the data. There was no attenuation correction (AC) in reconstruction with images used in the human study, while CHO was applied on images both reconstructed with and without AC. A three-dimensional Hann filter with 0.7 times the Nyquist frequency was used to smooth the reconstructed images. For images reconstructed without AC, both the CHO and the human observer study showed better detection performance for the 180/spl deg/ scan, especially for the female phantom. For images reconstructed with AC, the CHO study showed better detection performance for the 360/spl deg/ scan, especially for the female phantom. The CHO results demonstrated the detection performance with 360/spl deg/ scan was improved more than with 180/spl deg/ scan after AC.

Hybrid phantom representation for simulation of CT systems using intrinsic shapes, tomographic volumes and higher-order surfaces

We have extended our previously introduced hybrid phantom model to allow for phantom representation by higher-order surfaces (point meshes, polygonal meshes) in addition to any combination of intrinsic shapes (cubes, quadrics, super-quadrics) and voxel data originating from tomography (segmented CT/MRI volumes). Higher-order surfaces can be represented by uniform and non-uniform triangles obtained by means of Delaunays algorithm. In order to optimize computational performance for intersection calculations, a variety of efficient rejection tests has been introduced and a novel numerical algorithm which relies on a strategy similar to cue optimization is proposed. The proposed phantom model allows for unlimited complexity and number of phantom compartments. Various hybrid phantom representation examples including associated Monte Carlo simulation studies are presented and demonstrate the enormous versatility of our simulation framework. Compared to previous implementations computational speed has improved for simulation of PET and SPECT systems by more than 40% on the same workstation with the same hardware.

Pinhole Trajectories for SPECT Imaging of the Breast, Axilla, and Upper Chest

Several methods have been proposed for dedicated SPECT imaging of the breast. Two key characteristics achieved by these methods are shortened distance and lessened attenuation between the SPECT collimator and the breast. However, many of these methods are unable to image the important regions of the axilla and upper chest. Partial-Circle pinhole trajectories can image into the axilla, but only over a narrow axial range. Here we propose X and X-Line trajectories that can effectively image a broad axial range of the breast, axilla, and upper chest. The X and X-Line trajectories provide a third dimension of pinhole displacement, which is key to expanding the range of sufficient sampling. In order to investigate spatial variations in the imaging characteristics of different pinhole trajectories, phantoms were constructed with hot lesions and Defrise slabs distributed throughout the breast and torso. Projection data were computer simulated, and images were reconstructed by OSEM. These studies indicate that all three trajectories can sample activity well into the axilla. However, the axial range of good sampling is narrow with the Partial-Circle trajectory. The X and X-Line trajectories provide good sampling well into the axilla and over a broad axial range. The convex hulls of the X and X-Line trajectories encompass the breast and much of the axilla and upper chest, thereby satisfying Tuys condition for sufficient sampling of those regions, truncation issues aside. The spatial extent of the convex hull is tunable. At the same time, the X and X-Line trajectories achieve two other key characteristics of dedicated SPECT breast imaging: Shortened average distance and lessened average attenuation between the collimator, i.e. the pinhole, and the breast and other nearby tissue of interest.

Time-resolved monte carlo simulation of optical and isotopic photons in hybrid phantoms

We have developed a novel Monte Carlo simulation code for time-resolved simultaneous migration of optical and isotopic photons through 4D hybrid phantom media. Phantoms consist of parameterized regions with spatial varying material, activity, and wavelength-dependent optical properties (scatter, attenuation, anisotropy) which can be represented by implicit and higher-order surfaces as well as voxelized tomographic data. Since interaction frequency and subsequently mean free path lengths for optical photons differ significantly from isotopic photon propagation, two alternative ray-tracing strategies are proposed. This new approach allows for direct investigation and comparison not only of nuclear (PET, SPECT) and optical (ODT) imaging modalities but also for direct comparative analysis of simulation strategies such as Monte Carlo versus finite element simulation because of identical source and absorber geometries. The proposed algorithm has been used in our group to support small animal instrumentation development and to generate flux/projection images to advance reconstruction strategies, particular for tomographic approaches in optical imaging. The framework can also be used to develop and optimize acquisition protocols due to the ability of direct investigation of tracer/marker kinetic distributions including dual-modality labeling protocols