Friday 1:30- 3:00, Pier IV & V & Lobby

Chair: David S. Lalush, The University of North Carolina at Chapel Hill

V.Y. Chepel¹, M. Lopes¹, J. van der Marel¹, P. Crespo¹, D. Santos², L. Janeiro¹, V.N. Solovov¹, R. Ferreira Marques¹, A.J. Policarpo¹

^{1LIP-Coimbra and Physics Dept. of the Univ. of Coimbra, 3000 Coimbra, Portugal, 2Dept. of Electronics and Telecommunications of the Univ. of Aveiro, 3810 Aveiro, Portugal}

The liquid xenon multiwire chamber with detection of both ionisation and scintillation signals for Positron Emission Tomography (PET), proposed by our group several years ago, is being studied. Previously, some of its basic parameters were measured, such as transaxial position resolution (about 1 mm, fwhm), coincidence time resolution (1.5 ns, fwhm) and interaction depth resolution (<5 mm). Here, we report on the measurements of the detection efficiency for 511 keV gamma rays and its dependence on the discrimination level and event acceptance/rejection algorithm. In particular, the efficiency of detection of the scintillation and that of the ionisation signal on the anode wires are separately measured. The absolute values on efficiency are obtained by comparison with the efficiency of a 5" NaI(Tl) scintillation crystal for which tabulated data exist. Furthermore, we also investigated the uniformity of the scintillation light collection in the cell and the possibility of using the light signal for the measurement of the deposited energy.

J.A. McIntyre¹, J.T. Paulson¹, J. Aguiar¹, R.D. Allen¹

^{1Physics Dept., Texas A&M Univ., College Station, Texas}

A procedure has been developed for decreasing the sampling interval for a digital positron emission tomograph (PET). This PET provides 4 radial, as well as an angular, address for each scintillation detector in the tomograph ring. By tilting each detector toward its neighbor, the four radial segments subtend an angle that is one quarter of the angle between the untilted detectors. Four times as many gamma ray angles can then be determined in the tomograph ring. The utility of the improved sampling has been demonstrated by measuring the projection events for a 1 mm source placed at two locations 4 mm apart near the axis of the digital PET. The sampling interval was 1.25 mm with the untilted detectors and 1.25/4 = 0.31 mm with the tilted detectors. Each projection peak then had about 12, instead of about 3, sampling points and the peaks could easily be identified. To avoid an increase in statistical fluctuations, the events in the four neighboring sampling points were combined together: (1,2,3,4), (2,3,4,5), (3,4,5,6), etc.

Y. Shao¹, S.R. Cherry¹, A. Boutefnouchet¹

^{1Crump Inst. for Biological Imaging, UCLA School of Medicine}

We have investigated a scheme to determine DOI by using a Position Sensitive PMT at one end of an LSO crystal array, and a single channel PMT coupled to the other end through bent optical fibers with the ratio of the two signals to determine DOI. The PS-PMT can provide superior crystal identification, the single channel PMT is relatively cheap and easy to use. The design permits a relatively simple and practical solution to measure DOI. We have successfully bent fibers with bending angles ranging from 90 to 130 degrees and with a radius of curvature from 5 to 20 mm. The light loss due to fiber bending is less than 10% for these fibers. We tested the feasibility of the design with one end of a 2x2x21 mm LSO crystal connected to a PS-PMT directly, and the other end to a single channel PMT through a 90 degrees bent with a radius of curvature at 10 mm double clad fiber. The measured energy and coincidence time resolution are 17% and 2.3 ns. Preliminary measurements reveal a promising DOI measurement capability with DOI position uncertainty from 4.2 to 6.0 mm for most DOI positions.

R. Lecomte¹, A. Saoudi¹, D. Rouleau¹, H. Dautet², D. Waechter², M. Andreaco³, M. Casey³, L. Eriksson³, R. Nutt³

^{1Université de Sherbrooke, 2EG&G Optoelectronics, 3CTI PET Systems}

A detector module allowing individual crystal identification without analog coding is proposed. The basic cell is made of a 2&#61620;2 array of scintillators having different decay times that can be identified by pulse shape discrimination. A 16 pixel module consisting of a 2&#61620;2 array of these quad scintillator cells coupled to a 2&#61620;2 avalanche photodiode (APD) array was assembled and tested. In this design, the signal-to-noise ratio can be optimized in two ways: a) the electronic noise in individual APD channels is minimized by avoiding light sharing between cells; b) light collection efficiency is improved by eliminating light septa within cells to allow scintillation light propagation across crystals. All four crystals in a BGO/LSO/YSO/CsI(Tl) assembly can be clearly separated and individually gated for energy. An energy resolution better than 13% is obtained in all pixels for 511 keV gamma-rays.

D. Clément^1,2, R. Frei¹, J. Loude¹, C. Morel²

^{1Inst. of Nuclear Physics, Univ. of Lausanne, CH-1015 Lausanne, 2Division of Nuclear Medicine, Geneva Univ. Hospitals, CH-1211 Geneva 4}

Image resolution in positron emission tomography is currently limited by the unknown depth-of-interaction of the annihilation photons in the detector block. We present a three-dimensional position sensitive scintillation detector using neural networks. Light output of the detector block is mapped for known irradiation positions of a collimated photon beam and used to train a set of three independent multilayer neural networks. Once trained, the neural networks provide the three-dimensional coordinates of any photon impinging on the scintillation detector. Spatial resolution obtained for a cubic 25 x 25 x 25 mm3 CsI(Tl) crystal fully covered with 24 Si PIN photodiodes is better than 2 mm FWHM in the three directions over the entire volume of the detector block. For comparison, the spatial resolution reconstructed in two dimensions from a simple Anger logic based on the photodetectors coupled to the rear side of the crystal is about 3.5 mm FWHM for the same experimental setup.

N. Ferreira¹, R. Trebossen¹, B. Bendriem¹

^{1Service Hospitalier Frederic Joliot - CEA, Orsay, France}

Most of the algorithms used to correct for non uniformity of detection in PET compare the total number of coincidences recorded by each detector in a ring, when a symmetric source is centered in the field of view. They assume that the sum of the true counts recorded by all the detectors in coincidence with one detector (fan sum) is proportional to the detector efficiency. However, such methods fail to correctly calculate the efficiency of the detectors in coincidence with one defective detector or block, since the fan sum calculated for such detectors is different from the average fan sum over the ring. In this work we propose a correction for such effect. We first group all the detectors according to their position inside a block and identify the bad detectors in each group. The fan sum is computed for all the detectors except those in coincidence with bad detectors. Then, for these detectors, we calculate a fan sum, substituting the coincidence counts between each detector and a bad detector by the average of all the equivalent lines of response joining 2 good detectors. Using simulated data, we found that the average of the absolute errors in the calculated efficiency was 2% using the fan sum and less than 0.05% using the method proposed in this work.

S. Siegel¹, J. Vaquero¹, J. Seidel¹, W.R. Gandler¹, M.V. Green¹

^{1National Institutes of Health, Bethesda MD}

The CAMAC standard offers flexibility by providing power and a data bus for various modules, but is limited to a 1 Mword/sec bandwidth. LeCroy Research CAMAC modules with an auxiliary data bus, FERA, provide a 10 Mword/sec data transfer without CAMAC controller intervention. We have used a National Instruments digital I/O board (PCI-DIO-32HS) as a FERA bus-to-host bridge. The board provides hardware hand-shaking, a 20 Mword/sec bandwidth, bus master scatter-gather DMA, and can control up to 2 FERA busses asynchronously. Multiple boards may reside on the same PCI or Compact PCI bus. A 200 MHz Pentium PC w/16 MB RAM running Windows 95 sustains >2 Mword/sec throughput in 8255 emulation mode. These capabilities are being exploited in our prototype small animal planar and PET imaging system where 32 ADC channels (16 bits each) and 3 scaler channels (32 bits each) define an event.

D. Visvikis¹, T.D. Fryer², S. Downey¹

^{1Wolfson Brain Imaging Centre, Univ. of Cambridge Clinical School, Cambridge, UK, 2Dept. of Medical Physics, Addenbrooke's Hospital, Cambridge, UK}

NEC analysis is the standard method for optimising PET acquisitions. We have used this analysis for FDG imaging with a gamma camera PET system (Picker 2000XP). This system can acquire data with and without septa. The large axial acceptance angle of septaless acquisition makes this mode particularly susceptible to the detection of events originating from out of field activity. To investigate the influence of out of field activity on NEC performance, a combination of phantoms has been used to model activity in the brain, heart and bladder. Experiments were performed with and without septa to evaluate the total and randoms counts rates and scatter fractions for various energy windows. The results reveal that septaless acquisition gives the optimum NEC performance for brain imaging, but septa are required for body imaging.

S.D. Wollenweber¹, B.M. Tsui¹, T.D. Rempel², V. Simcic²

^{1The Univ. of North Carolina at Chapel Hill, 2Siemens Medical Systems}

The purpose of this work is to provide an initial investigation of coincidence imaging characteristics including spatial and energy resolution, sensitivity and scatter fraction for the new Siemens ECAM+ coincidence imaging system. The system uses lead slat collimators to reduce detected scattered photons. From the reconstructed images obtained using single-slice rebinning, the measured radial resolution of 4.6 mm was nearly constant for all detector separations (DS) and source locations. Tangential resolution showed a nearly linear increase and varied from 4.5 (at center) to 11.4 mm (20 cm from center) and showed similar trends for all DS. The axial resolution varied from 4.5 (at center) to 12.2 mm (20 cm from center) for maximum DS and was slightly better with the same trend for smaller DS. Sensitivity was 9063, 9948 and 12614 [counts/sec]/[mCi/mL] for the 20%, 30% and 300-600 keV energy windows. Scatter fraction was measured as 12.3% using the NEMA protocol. The percent energy resolution averaged 7.13, 7.59 and 8.90% FWHM for 511 keV photons at low, medium and high dead times.

G.L. Zeng¹, G.T. Gullberg¹

^{1Radiology, Univ. of Utah, Salt Lake City, UT}

Whole body SPECT with a helical scanning orbit may be useful in diagnosing cancer. In order to correct for attenuation and avoid truncation of projection data, the use of asymmetric cone-beam imaging geometries for simultaneous transmission and emission data acquisition is being investigated. A data sufficiency condition was developed for a helical orbit for any asymmetric cone-beam geometry that includes the center of rotation in all projection views. At least two detectors are required. If the two detectors have opposite geometries, the data sufficiency condition is such that when a detector rotates "pi plus fan-angle," it should shift axially by a, where a is the detector size projected on the axis of rotation. Computer simulations were performed to verify this helical pitch requirement. Simulated projections of a long Defrise phantom, with a length longer than the height of the cone-beam detector, was used to verify the image reconstruction using the iterative ML-EM algorithm.

C. Lyckman¹, S. Dale¹, M. Persson¹, D. Bone¹

^{1Dept. of Medical engineering, Karolinska institute, Novum, S-141 86 Huddinge, Sweden}

The aim of this study was to investigate the feasibility of performing cerebral blood flow (CBF) scintigraphy using the 3D method of Ectomography. Ectomography is a limited view angle method and has been implemented on a mobile system which can be used bedside for acute studies. Simulations (ideal conditions), phantom experiments and six patient studies have been performed with this system using a 40 rotating slant hole collimator (general purpose). Reconstructed slices from simulations and patient studies have been compared to SPECT. Ectomographic and SPECT simulations give comparable images. Phantom experiments using a 3D brain phantom show that the optimal acquisition geometry is with the detector parallel to the orbito-meatal plane. In patient studies, Ectomographic images were similar to SPECT images obtained using a collimator with ultra high resolution. There were however differences which have to be investigated further. In conclusion, the results obtained indicate the feasibility of using Ectomography for CBF studies, especially if a high resolution collimator is used.

M.P. Tornai¹, D.R. Gilland¹, R.J. Jaszczak¹, R.E. Coleman¹, F. Coutand², A. Wattellier², Y. Ooie³, M. Taguchi³, G. Enos³

^{1Duke Univ. Medical Center, 2Medasys Digital Systems, 3Hitachi Medical Corporation}

A recently developed large FOV (51cmx37cm) dual head, 9mm thick NaI(Tl) SPECT system was investigated for transmission CT (TCT) attenuation compensation (AC) with various transmission sources. Each head has three degrees of freedom (transaxial tilt, vertical and off-isocenter travel). For TCT, the system utilizes a fixed line source at the 77cm focus of a symmetric fan-beam collimator. Different size FOVs (34~50cm dia) can be configured when the TCT head is tilted and translated to accommodate variations in patient size. Thus, the TCT detector with large FOV more easily avoids truncation artifacts without sacrificing resolution, compared with other TCT geometries. Line source energies from 100 to 165 keV (Gd-153,Tc-99m,Te-123m,Ce-139) were utilized with rod/sphere and anthropomorphic phantoms. The TCT images were reconstructed with OSEM, while the emission reconstructions utilized FBP (without/with iterative and multiplicative Chang AC) and OSEM (without/with AC). These line sources demonstrate the potential for suitably compensated images and myocardial activity profiles.

M.J. Flynn^1,2, S.E. Wilderman²

^{1X-ray Imaging Res. Lab., Henry Ford Health Sys.,Detroit, MI, 2Nucl. Engr. & Rad. Sci., Univ. of Michigan, Ann Arbor, MI}

We describe a unified method to evaluate the performance of direct digital radiography detectors using Monte Carlo analysis. Detector elements are treated as integrating volumes which sample the deposited charge. From first order and second order statistics associated with of each history, the frequency dependent detective quantum efficiency, Dqe(E,f), is evaluated. The method is not limited with respect to geometry, materials, or interactions in a history. We report results for direct conversion radiography detectors using Se or PbI2 semiconductor films. Complex energy deposition distributions result from the small size of the detection volumes and the variety of materials near the detection region. The effects of substrate materials and absorptive backing materials are shown to significantly influence resolution without contributing to the spatial correlation of noise. A significant low frequency drop in Dqe(E,f) resulting in contrast reduction is shown to exist which varies dependent on the input energy and conversion material.

J. Vaquero¹, S. Siegel¹, J. Seidel¹, M.V. Green¹

^{1National Institutes of Health, Bethesda MD}

Equilibrium-gated blood pool imaging of the heart is a common diagnostic imaging procedure for visualizing and quantifying cardiac function in human subjects. Recently, this procedure has been modified to evaluate cardiac function in mice. The high heart rate encountered in these animals (often greater than 400 beats/min) can confound R-wave trigger devices, acquisition systems and image processing software containing embedded default conditions tailored specifically to the lower heart rates of human studies. In order to determine whether data acquisition and processing components of a commercial or self-generated gated blood pool imaging procedure are performing properly, input of known timing and imaging signals that mimic those generated during high heart rate gated blood pool imaging is required. Here we describe an inexpensive phantom that is suitable for initial evaluation of an unknown system and for ongoing QC of a previously verified system.

E.L. Yeh¹, S. Huang¹, J.R. Barrio¹, E.J. Hoffman¹

^{1Dept. of Molecular and Medical Pharmacology, UCLA School of Medicine}

We are developing a method for physical simulation of dynamic PET scans in a physical phantom. As a model for physical phantoms that can simulate realistic tissue kinetics, we investigated the use of ion exchange resins to imitate tracer uptake and clearance in tissue. We developed a unit cell perfusion system to demonstrate feasibility for such an approach. The system utilizes competitive binding of 13NH4+ and Na+ for cation exchange resin sites. Due to the high binding affinity of 13NH4+ to the cation exchange resin, efficient tracer uptake can be completed, followed by gradual washout of the tracer from binding sites by an aqueous solution of NaCl. The clearance rate of the tracer can be modified to achieve desired temporal clearance characteristics by changing either NaCl eluent concentration or flow rate. The PET phantom would incorporate one or more of these cells and would include other structures to provide realistic backgrounds. The cells would have well calibrated characteristics that would test the capability of PET to measure quantitative kinetics from image data.

M. Dallimore¹, I.D. Jupp¹, P.T. Durrant¹, D. Ramsden¹

^{1Univ. of Southampton}

A novel design for a 3D X-ray microscope based on the coded-aperture imaging technique has been designed for the in vivo monitoring the distribution of radio-pharmaceuticals within small animal specimens. Imaging in the depth plane is realised through the implementation of the cross-correlation reconstruction technique. this technique allows fast image retrieval but is degraded by systematic artefacts. Simulations have been performed to determine both the optimum detector geometry and whether a more advanced iterative reconstruction technique is necessary to eliminate image artefacts. These simulations have been verified using a high resolution X-ray image intensifier combined with a (HURA) hexagonal uniformly redundant array coded aperture. Designs of practical imaging systems based upon an X-ray image intensifier and the newly developed M-HPD are discussed.

S. Weber¹, H. Herzog², H. Coenen³, R. Engels¹, F. Kehren², R. Reinartz¹, P. Reinhart¹, F. Rongen¹, F. Sonnenberg¹, H. Halling¹

^{1Zentrallabor fuer Elektronik, Forschungszentrum Juelich, 2Inst. fuer Medizin, Forschungszentrum Juelich, 3Inst. fuer Nuklearchemie, Forschungszentrum Juelich}

The design and preliminary performance characteristics of the TierPET, a high resolution small animal PET scanner, were presented at the previous IEEE Medical Imaging Conferences. The scanner uses two orthogonal pairs of detectors consisting of arrays of small YAP crystals coupled to position sensitive photomultipliers. The source-detector distance may be varied to improve either sensitivity or spatial resolution. The proposed spatial resolution of about 2 mm over the whole field-of-view could be confirmed by our measurements. We will point out the experimental results obtained with the TierPET scanner, including further performance measurements according to the NEMA protocol and the effects of correction methods. A fully 3D iterative reconstruction algorithm and a 2D iterative algorithm will be compared in respect of the quality of reconstruction results and convergence rate. First investigations in rats have been performed. We will present the results and discuss the experiences with the scanner and its limitations as well as future developments.

L. Ploux¹, L. Pinot¹, R. Mastrippolito¹, P. Laniece², Y. Charon², F. Pain¹, L. Valentin²

^{1Inst. de Physique Nucléaire, Université de Paris 11, Orsay Cedex, France, 2Groupe Modelisation Physique et Interfaces Biologie, Université de Paris 7, 2 place Jussieu, Paris, France}

We have proposed a new approach for in vivo tomography adapted to small animal models. This original tomograph, TOHR (French acronym for TOmographe Haute Rsolution) stresses high resolution while maintaining high efficiency. We develop a 1.4 mm resolution prototype of TOHR based on an icosahedral focusing collimator (15 over the 20 triangular faces of an icosahedron). We evaluate here a new method for centring the different modules and present a new acquisition system based on time coding of energy. The noise characterisation will be detailed. We have measured the performance of the system in single and double photon modes. Phantom images will be presented and finally, as an example of applications, TOHR performances will be related to the neurosciences field. Preliminary in vivo experiments with TOHR will be realized in that context, notably on rat model of dopaminergic neurons degeneration using 123I-Epidpride.

J.A. Correia^1,2, C.A. Burnham^1,2, D. Kaufman¹, A.J. Fischman^1,2

^{1Massachusetts General Hospital, 2Harvard Medical School}

Small animal imaging is a situation where positron range effects and sampling dominate. Simulation studies support the idea that resolution close to 1mm can be achieved with 18F. The purpose of the work reported here was to design and construct a prototype high resolution PET instrument to study the feasibility of 1mm imaging. The detector consists of a single ring of 360 1x5x4.5mm LSO crystals organized into optically isolated blocks of 12 crystals each viewed by two photomultipliers in the manner of Wong. Crystal-specific boundary and energy lookup tables are used to identify the crystals. Upon reciept of a coincidence in the hardware the 30 PMT signals are transfered in to a PC for processing. The instrument is currently operating at 1.5-2mm and is bring optimized for higher resolution. Ongoing work includes improvement of the detector-block design and crystal identification, study of depth-of- interaction correction and implementation of corrections for physical effects such as field non-uniformity, attenuation, scatter and randoms.

M.C. Wu¹, M.W. Dae¹, H. Tang¹, K. Iwata¹, A.J. Da Silva¹, B.H. Hasegawa¹

^{1Univ. of California, San Francisco}

We are developing an imaging system incorporating a scintillation camera with a pinhole collimator to acquire ecg-gated myocardial perfusion studies in a mouse using technetium-99m sestamibi. Our design goal is to achieve 0.5 mm FWHM spatial resolution for the 5 mm diameter murine myocardium with cardiac gating at 600 beats per minute. Planar projection images will be acquired with a 0.5 mm pinhole located 20 cm from the detector surface. Our analysis shows that an object located 2 cm from the pinhole will achieve the desired spatial resolution in which neither pinhole magnification nor the intrinsic resolution of the scintillation camera dominates. In addition, the design should be able to achieve a sensitivity of 3000 counts per second, allowing a study with one million counts to be acquired in less than ten minutes.

S. Yamamoto¹, T. Matusda², I. Kanno²

^{1Kobe City College of Tech., 2Research Inst. for Brain and Blood Vessels, Akita}

A digital signal processor(DSP) based position calculation circuit was developed and tested for beta camera. The previous position calculation circuit which employed flash analog to digital(A-D) converters for A-D conversion and ratio calculation occurred a significant line artifacts in the image due to the differential non-linearity of the A-D converters. The new position calculation circuit uses four A-D converters for only A-D conversion of the four analog signals from the position sensitive photomultiplier tube(PSPMT). The DSP reads the four A-D signals and calculates the ratio of Xa / (Xa + Xb) and Ya / (Ya + Yb) in event by event basis. The DSP also magnifies the image to fit the useful field of view(FOV) and rejects the events out of the FOV. The line artifacts in the image was almost eliminated. The count rate performance of the new circuit was also evaluated.

J.E. Lees¹, J.F. Pearson¹, G.W. Fraser¹, J.M. Hales¹, P. Richards¹

^{1Univ. of Leicester, England}

We describe a pre-commercial prototype autoradiography system which can be used to image a variety of radiolabelled biological samples, ranging from lymphocyte proliferation assays to semi-thin whole body tissue slices. These samples can be imaged with high spatial resolution (~ 70 microns) and high sensitivity (eg. 3H - minimum detectable activity ~0.001Bq in 20 hours) over a wide (5-6 orders of magnitude) dynamic range. Specimens are automatically presented, through a vacuum load-lock, to an imaging microchannel plate (MCP) detector operating in high (~10**(-6)mbar) vacuum. We present images obtained from tritium labelled 2-D protein electophoresis gels and immunological assays and discuss the importance of the new system for a variety of biological techniques.

F. Pain¹, P. Laniece², R. Mastrippolito¹, L. Ploux¹, Y. Charon², L. Valentin²

^{1Inst. de Physique Nucleaire, Universite de Paris 11, Orsay cedex, France, 2Groupe Modelisation Physique et Interfaces Biologie, Universite de Paris 7, Paris, France}

Although high resolution tomographs provide a new approach which strongly simplifies the measurement of in vivo tracer biodistribution and kinetics in small animals, they suffer from a major drawback: the need of an animal anesthesia which restricts the neurophysiological investigations. Moreover, quantitative in vivo experiments realized on brain sometimes require only a simple and local measurement of the radioactivity achieved on few points and do not necessary imply the use of a tomograph, detector of high cost. Those constraints led us to the development of an intracerebral beta sensitive probe, SIC (French acronym of intracerebral probe) allowing chronic measurements of the neurophysiological activity in awake and free of movement small animals. Spatial selectivity of the probe and noise contributions were evaluated by Monte-Carlo simulations and will be exposed. Characterizations of a first prototype based on a small scintillating fiber (500 m diameter and 500 m length) fused to a same diameter clear optical fiber coupled to a photomultiplier will be also presented.

K. Murthy¹, D. Jolly¹, C.J. Thompson¹, A. Loutfi², R. Lisbona^1,2, J.H. Gagnon²

^{1Montreal Neurological Inst., McGill Univ., 2Royal Victoria Hospital, Montreal, Quebec}

We have previously shown that our Positron Emission Mammography-1(PEM-1) system can successfully detect small(< 2 cm) breast tumors. Images are judged qualitatively for the presence of focal uptake of FDG. Quantification of PEM-1 images is essential if the technique is to be applied to track the response of malignant tumors to chemotherapy. To this end we have built a breast phantom and developed a novel technique for fabricating very small radioactive hot-spots without walls. The hot-spots are made by adding F18-FDG to a solution made by dissolving 25 mg agarose/cc water. The heated solution is poured into spherical molds and upon congealing separated from it. Contrast resolution experiments have been performed with 12 and 15 mm diameter hot-spots in the breast phantom containing water with various amounts background activity.In both cases, the observed contrast agrees well with theoretical values (for 12-mm case, mean observed contrast 864% theoretical contrast). Tests to evaluate the effect of multiple hot-spots on the system performance are currently underway.

W.A. Worstell¹, H.A. Kudrolli¹, V.G. Zavarzin¹

^{1Boston Univ. Physics Dept.}

We are developing a portable system for PET imaging of single organs and research animals with fixed dual-plane detector geometry. The device uses wavlength-shifting fiber readout to obtain fine spatial resolution (3mm FWHM), good depth-of-interaction sensitivity, and a relatively large field-of-view (25cm x 25cm). This work builds upon our earlier experience with 11cm x 11cm prototypes which also incorporated wavelength-shifting fiber readout and which were developed for Positron Emission Mammography (PEM). Iterative 3-D PET reconstruction permits tomographic imaging with fixed dual-plane detector geometry. Such a detector geometry permits imaging without a complete detector ring or a complex gantry system. It has utility for breast imaging and for several other applications. Imaging results for phantoms and laboratory animals using our new device and its prototype predecessors are presented and discussed.

G.J. Gruber¹, W.W. Moses¹, S.E. Derenzo¹

^{1LBNL, Univ. of California}

We describe Monte Carlo simulation results for breast tumor imaging properties when using a compact, discrete scintillation camera. The program is to be used to analyze camera design, especially pixel size and collimator configuration. In addition to pixel and collimator geometries, tumor images are affected by lesion size, imaging distance, tumor-to-background uptake ratio, energy resolution, energy discrimination, acquisition time, and the detection of scattered photons originating outside the breast. We have begun analyzing these influences for 12 camera geometries, including different pixel sizes (1.5 mm and 3.0 mm), collimator sensitivities, and collimator hole shapes (hexagonal and square). Results suggest that for 5-15mm diameter tumors imaged at depths of 10-60 mm, pixel size and collimator geometry have a minor impact on both the spatial resolution and peak tumor-to-background contrast. Considerations such as statistical noise (suggesting high sensitivity collimators) and the density of the readout electronics will likely be more important to compact gamma camera design.

L. Zhang¹, B.K. Horn², R.C. Lanza¹

^{1Dept. of Nuclear Eng., MIT, NW13-213, Cambridge, MA , 2Dept. of Electrical Eng. and Computer Sci., Artificial Intelligence Laboratory, MIT, NE43-715A, Cambridge, MA}

We have studied coded aperture patterns composed of cyclic difference set uniformly redundant arrays such that the system point-spread-function is a delta function and no artifacts are introduced by the spatial multiplexing process. Although there are some analytical methods for designing some patterns with such a property, the theory is quite incomplete thus requires a different approach. We have searched a complete list of such patterns for a range of 1-D and 2-D dimensions, and have obtained some rules. The available results can be used as practical patterns, or as future research reference for coded aperture pattern design.

We summarize the 3-D coded aperture imaging methods through a single view (coded aperture laminography, moving object coded aperture imaging) and through multiple views (modular coded aperture technology, coded aperture tomography), describe the image reconstruction and data processing algorithms, and discuss and compare the performance of different coded aperture imaging methods. Simulation and experimental results have been provided for verification.

D.L. Gunter¹, K.P. Matthews¹, C.E. Ordonez¹

^{1Rush-Presbyterian-St.Luke's Medical Center}

Collimators, which are just periodic arrays of holes in lead, are used on gamma cameras in nuclear medicine to create images. The collimator hole pattern is not visible in clinical images because the holes are separated by distances smaller than the intrinsic resolution of the gamma camera. However, new (CZT) solid-state detectors are characterized by a regular grid of detector pixels. The interaction between the collimator-hole lattice and the detector grid produces Moire patterns unless the two grids are commenserate. Unfortunately, collimators with hole spacings commensurate with the pixel size are far from optimal in nuclear medicine. A mathematical analysis of the imaging process in such systems is provided and computer simulations are used to demonstrate the effects. Two new strategies for minimizing the effects are examined: (1) increasing the gap between the collimator and the detector and (2) constructing the collimator with numerous small segments having displaced grid structures. Both strategies have serious drawbacks, but are better than current alternatives.

L. Zhang¹, R.C. Lanza¹, B.K. Horn², R.E. Zimmerman³

^{1Dept. of Nuclear Eng., MIT, Cambridge, MA , 2Dept. of Electrical Eng. and Computer Sci., Artificial Intelligence Laboratory, MIT, NE43-715A, Cambridge, MA , 3Dept. of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA}

Standard nuclear medicine imaging uses photon collimation and thus suffer from very low sensitivity, especially if high energy (>511 KeV) isotopes are to be imaged. Coded aperture techniques use a coded pattern mask instead of a collimator to encode the photon source distribution. It significantly improves the system sensitivity without degrading the spatial resolution of the reconstructed images.

We have developed coded aperture arrays for near field imaging; used a cyclic difference set uniformly redundant array as the coded aperture pattern; conducted imaging experiments for point, 2-D, and 3-D sources of 140 KeV and 511 KeV through phantoms; and compared the experimental results with those from collimator systems. The experiments have been conducted using a Siemens E.CAM gamma/SPECT camera. Our results have shown significantly improved sensitivity for a coded aperture imaging system over collimator systems. We have demonstrated the possibility of using coded apertures and a conventional gamma camera to image gamma rays of high energy, such as 511 KeV, in nuclear medicine.

C.E. Ordonez¹, W. Chang¹, A.I. Bolozdynya²

^{1Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL, 2Case Western Reserve Univ., Cleveland, OH}

In two-detector Compton cameras, the incident direction of gamma rays are localized to the surface of a cone, the axis of which is determined by the interaction positions detected in the two detectors. The cone angle is determined by the energies measured in the two detectors. In this paper, a general expression that relates angular uncertainty in Compton cameras to detector geometry and spatial resolution is derived. The properties of this expression are investigated and numerical estimates for a few simple Compton camera configurations are presented.