Development of Small Animal PET Systems

Physics and engineering research at the Imaging Research Laboratory is clustered around two main topics: Developing a small animal PET imaging system optimized to image mice and improving the imaging capabilities for clinical PET and PET/CT scanners for humans. In addition we provide simulation and imaging tools.

Development of Small Animal PET Systems

Supported by NIH grants R01-EB002117 (Lewellen),   R24-CA88194 (Lewellen),   R21/R33- EB0001563 (Miyaoka), and P01-CA42045 (Krohn).

The Small Animal PET System research cluster has three components:

• Developing the four-ring micro crystal element scanner (MiCES) to be used for functional and biochemical in vivo imaging of mice and other small animals
• Testing of a single ring prototype version, QuickPET II
• Development of next-generation detector technologies for cost-effective construction of small animal PET scanners

The micro crystal element scanner (MiCES)

The micro crystal element scanner (MiCES) will be used for functional and biochemical in vivo imaging of mice and other small animals. Driven by advances in genomics and molecular biology, small animal models, in particular genetically engineered mice, are increasingly recognized as powerful tools to study human disease. In the field of oncology, PET imaging is used for tumor detection (i.e., tumorigenesis and metastasis), disease staging and monitoring, tumor characterization (e.g., hypoxia and estrogen receptor status), response to therapy, and cell trafficking studies. Small animal PET imaging is also being used to study gene expression; to study normal organ uptake; and to study brain function.

The specific aims of the MiCES design were to achieve an image resolution of less than 1 mm and an absolute detection sensitivity of 6%. Each of these design goals is pushing the state of the art for small animal PET imaging systems. The key design component of the system is our micro crystal element (MiCE) detector. Each MiCE detector consists of a 22x22 array of polished 0.8x0.8x10 mm mixed lutetium silicate (MLS) scintillation crystals. The crystals are placed within a grid made of a highly reflective polymer film material (i.e., radiant mirror film, 3M). The grid serves three purposes: 1) it optically isolates the crystals; 2) it functions as a reflective wrap; and 3) it provides structural support for the crystal array. The crystal array is directly mounted to a position sensitive photomultiplier tube (PMT) using an optical coupling compound. The PMT converts and amplifies the light produced by the scintillation crystals into an electrical signal. A sample crystal array and position-sensitive PMT is pictured in figure 1 .

Figure 1. Micro crystal element detector (MiCE) and position sensitive photomultiplier tube next to a dime.

The four ring version of the scanner, named MiCES, is under construction and will be operational by summer of 2004. The system will consist of 72 detector modules housed in 18 detector cassettes. Each detector cassette will house 4 detector modules, as illustrated in figure 2 . MiCES will have ~9.5 cm of axial FOV and ~8 cm transverse FOV. The increase of the in-plane FOV size is due to different data acquisition electronics. To fully sample the imaging FOV, each detector cassette will be axially offset from its neighbor, as illustrated in figure 2 , and the gantry will be continuously rotated during the study.

Figure 2. Neighboring rows of MiCE detectors (detector cassettes 1 and 2) are axially offset to fully sample the imaging FOV when the detector system is rotated.

The UW IRL is designing all of the system electronics with the exception of an amplifier/digitizer board that was developed under a technology sharing agreement between the University of Washington and Concorde Microsystems (Knoxville, Tennessee). A picture of the jointly developed board is shown in figure 3 . Each card supports a detector cassette (i.e., 4 detector modules). The board amplifies and sums the incoming analog signals. The board also assigns a digital time stamp corresponding to the arrival time of each event. The input signals are digitized using very high speed analog to digital converters (ADCs). Processing of the digital data is handled by two powerful field programmable gate array (FPGA) chips located on the board. When a coincidence event is detected, the digitized data is passed to the data acquisition electronics. The data acquisition system is based upon the IEEE 1394a (Firewire) standard and can support coincidence rates in excess of 400 kcps with minimal deadtime.

Figure 3. UW modified CPS analog signal board (ASB).    The 16 analog signals enter at the top and are processed by 4 Concorde ASICs.   The ASIC analog outputs are digitized and integrated by the on-board ADCs and FPGAs.   The ASIC time stamps are transferred to the FPGAs.   Final data is transferred to a DSB.

To enable rotating the detector ring, the gantry is mounted on a mechanical slip ring. All DC voltage signals are supported by the main slip ring. The system also utilizes an optical slip ring to provide high throughput of the digital information.

The data acquisition computer is a Macintosh Xserve running OS X. Data can be transferred to a cluster of five additional Xserves to support image reconstruction and data analysis.

Recent publications related to this project:

Lewellen, T., C. Laymon, R. Miyaoka, M. Janes, B. Park, K. Lee, and P. Kinahan. Development of a data acquisition system for the MiCES small animal PET scanner. in 2002 IEEE Nuclear Science Symposium and Medical Imaging Conference. 2002. Norfolk p. 1066-1070.

Lewellen, T., R. Miyaoka, M. Janes, B. Park, and P. Kinahan System Electronics for the MiCES Small Animal PET Scanner. in IEEE Nuclear Science Symposium and Medical Imaging Conference. 2003. Portland, Or. p. in press.

Miyaoka, R., C. Laymon, M. Janes, K. Lee, P. Kinahan, and T. Lewellen. Recent progress in the development of a micro crystal element (MiCE) PET system. in 2002 IEEE Nuclear Science Symposium and Medical Imaging Conference. 2002. Norfolk p. 1287-1291.

Laymon, C., R. Miyaoka, B. park, and T. Lewellen, Simplified FPGA-Based Data Acquisition System for PET. IEEE Trans. Nucl. Sci., 2003. 50(5): p. 1483-1486

[top]

The QuickPET II prototype scanner

A single ring version of the system, pictured in figure 4 , has been operational since June, 2003. It consists of 18 detector modules and has a 12.8 cm ring diameter and a 1.98 cm axial field of view (FOV). The system collects fully 3D data. Lead end shields reduce the animal port to 10 cm. The in-plane imaging FOV is 5.76 cm. The detector ring is mounted on a bearing that allows partial rotation of the detectors (+/- ~20 degrees). This allows the lines of response associated with the gaps between modules to get sampled. The imaging table (Summit Medical Equipment, Oregon) was designed to allow delivery and exhaust of gas anesthesia from a single end of the table. The green tube is for gas delivery while channels in the table allow the exhaust to be collected from the black tube at the end of the table. Three linear stages and one vertical stage allow for accurate table placement. The table has a linear travel range of ~9 cm, enough for whole body scanning of a mouse. An image resolution of 1.1 mm full width at half maximum (FWHM) has been measured for a reconstructed line source.

Figure 4. Single-ring QuickPET II prototype of the MiCES small animal PET system.

Mouse imaging studies have been conducted for a range of radiopharmaceuticals, including FDG, FLT, FES and F-Annexin. Maximum pixel projection images of mice imaged with 18F-flurodeoxyglucose (FDG) are illustrated in figures 5 and 6 . Figure 5 is of a transgenic mouse that develops spontaneous mammary tumors. Figure 6 is of a mouse with induced skin tumors. The image in the blue box, figure 6 , is a transverse slice through the mouse's heart. The bright ring is the mouse's left ventricle. The faint ring is the mouse's right ventricle. This image illustrates the extremely high spatial resolution of the scanner.

Figure 5. Sample FDG image of a mouse with mammary tumors.

Figure 6. Maximum pixel projection of mouse with induced skin tumors.   Image in the blue box is the transverse slice through the heart represented by the blue dotted line.

Along with building the MiCES scanner, the UW IRL is investigating statistical (i.e., iterative) image reconstruction techniques to improve image quality. A pragmatic iterative reconstruction utilizing a factorized system model of the MiCES scanner is being developed. This work focuses on the characterization and incorporation of accurate models of the MiCES scanner in the reconstruction method. The current implementation includes a model of the system's detector response function. Images of a mouse heart reconstructed with a standard analytic technique (i.e., filtered back projection, FBP); a standard statistical technique (i.e., ordered subsets expectation maximization, OSEM); and our statistical method that includes a model of the detector blurring response function (OSEM+DB) are illustrated in figure 7 . Our OSEM+DB reconstruction improves spatial resolution without significantly increasing the noise texture relative to standard OSEM. In addition to including a model of the detector response characteristics, we are working on models for 18F and 11C positron range to incorporate in our statistical reconstruction.

Figure 7. Transverse sections of an image volume of a mouse heart FDG uptake reconstructed with: (a) FBP, (b) OSEM and (c) OSEM+DB.

Recent publications related to this project:

Miyaoka, R., M. Janes, B. park, K. Lee, P. Kinahan , and T. Lewellen. Toward the Development of a Micro Crystal Element Scanner (MiCES): quickPET II. in IEEE Nuclear Science Symposium and Medical Imaging Conference. 2003. Portland, OR: IEEE p. in press.

Miyaoka, R., M. Janes, and T. Lewellen. Optimization of Mounting Large Crystal Arrays to Photomultiplier Tubes. in IEEE Nuclear Science Symposium and Medical Imaging Conference. 2003. Portland, OR.: IEEE p. in press.

Lee, K., P. Kinahan , J. Fessler, R. Miyaoka, and T. Lewellen. Pragmatic Image Reconstruction for the MiCES Fully-3D Mouse Imaging PET Scanner. in IEEE Nuclear Science Symposium and Medical Imaging Conference. 2003. Portland, OR.: IEEE p. in press.

Lee, K., P. Kinahan, R. Miyaoka, J. Kim, and T. Lewellen, Impact of system design parameters on image figures of merit for a mouse scanner. IEEE Trans. Nucl. Sci., 2004: p. in press.

[top]

Micro Detector Systems Research

To further improve our capabilities for small animal imaging, two research projects are underway to improve the basic detector technology. A fundamental goal for all of our development is to design cost effective detector designs that can easily be adapted to the basic MiCES scanner gantry and electronics. We have taken care to design the gantry and electronics so that new detector module designs can be integrated with a minimum of modifications.

A refinement of the MiCE module is to add depth-of-interaction (DOI) capability (dMiCE) as well higher sensitivity (longer crystals). Other designs we are testing focus on reading out light from both ends of the crystal or using different crystals stacked in layers and using differences in risetime to identify the layer of interaction. Both of these approaches require more electronic channels and light sensors, increasing the cost of implementation. Our basic approach is to use a single-ended light collection system and control the sharing of light between adjacent crystals to provide the DOI information. This approach is inherently less expensive than the other designs. The basic approach is shown in Figure 8 and preliminary results are depicted in figure 9 . To make a practical version of dMiCE with small cross section crystals (e.g., 1x1x25 mm) we are developing an approach with laser cutting patterns into the same 3M reflective polymer used for the MiCE module. We can then fabricate the modules as a laminate of layers of crystals and properly cut reflector. Currently we are working on the ideal pattern versus crystal size and testing a variety of glues and laminate techniques to assure adequate light collection.

Figure 8.   UW DOI detector concept. (a) DOI detector unit.   (b)   Ratio plots.   A significant amount of light is shared when an event is detected near the entrance face of the detector unit.   Less sharing occurs for interactions near the PMT interface.

Figure 9.   Ratio peak and FWHM values of ratio plot versus DOI for a GSO crystal pair with unpolished surfaces and the coupling scheme of Figure 1.   The black lines are estimates of DOI uncertainty for each depth position.

The second detector design is based on small, continuous blocks of scintillator (cMiCE). This design would dramatically reduce the cost of module construction by eliminating most of the crystal and polishing costs. The challenge is in obtaining good spatial resolution uniformly throughout the crystal. Conventional weighted summing techniques are inadequate. Our group has developed a statistical positioning approach that greatly improves the ability to obtain high spatial resolution throughout the crystal volume. Currently, simulation work is in progress to explore optimal arrangement of photomultiplier tube elements. Test crystals (25x25 and 48x48 mm) are being fabricated with a variety of thicknesses. Initial simulation results indicate that using new technology flat PMTs provide excellent performance with intrinsic resolutions of 1 mm with crystal thicknesses of 8 and 10 mm. The challenge is to achieve such performance in real detectors with crystal thicknesses of up to 20 mm.

The laboratory has recently been expanded with a new 64 channel ADC system and motorized x-y-z stages to allow proper experimental evaluation of the cMiCE and dMiCE detector modules. We will develop both detector modules to determine which design will provide the best price/performance ratio for our next small animal PET scanner.

Recent publications related to this project:

Miyaoka, R., T. Lewellen, H. Yu, and D. McDaniel, A depth of interaction PET detector module. J. Nucl. Med, 1998. 39(5): p. 170P.

Miyaoka, R.S. and T.K. Lewellen, Effect of Detector Scatter on the Decoding Accuracy of a DOI Detector Module. IEEE Trans. Nucl. Sci., 2000. 47(4): p. 1614-19.

Joung, J., R. Miyaoka, and T. Lewellen, cMiCE: A High Resolution Animal PET Using Continuous LSO with a Statistics Based Positioning Scheme. Nucl. Inst. Methods, 2002: p. in press

[top]