Abstract
308
Objectives: We are developing a stationary high-performance Dynamic Cardiac SPECT (DC-SPECT) system using scintillator-based detectors and pinhole collimators. The main objective of this project is to provide simultaneous high-resolution images and high system sensitivity for diagnostic and theranostics applications by using large detector coverage where detectors have converging pixel volumes to avoid depth of interaction (DOI) blurring. It is known that to reconstruct a high-resolution image of photon-starved organs such as heart in a short acquisition time, high system sensitivity and many simultaneously acquired projections is required. Due to a number of cost-performance tradeoffs CZT-based SPECT systems (D-SPECT, DNM530) have relatively small detector coverage with large collimator holes compared with that of the dual head gamma cameras (DHGCs) to increase system sensitivity. This result in poor image resolution when no resolution recovery applied. METHODS: Current design of the DC-SPECT is based on using 80 high intrinsic detector resolution (IDR) CsI:Tl modules each coupled with a photodetector array and backed by a pinhole collimator acquiring an image of the heart. Detector-collimator pairs are arranged in a body-contouring geometry around the patient and are all viewing an 18 cm dia. sphere as human heart. Therefore, the collimator-CFOV distance varies among the detector modules. The CsI:Tl detectors are laser-processed with 2.0x2.0x10.0 mm3 pixels volumes with large thickness to stop high-energy photons beyond 140 keV. Given the DOI blurring issue inherent with pinhole collimators, crystal pixels in our detectors have converging pixel volume with a focal length equal to the detector-pinhole distance. We optimized system design and altered geometrical values of the collimators and distance to CFOV with respect to 3 main criteria: a) achievable sensitivity @15mm resolution matching DHGC’s, b) achievable spatial resolution @0.01% sensitivity matching DHGC’s, and c) achievable sensitivity @10 mm resolution. We also have laser-processed multiple CsI:Tl crystals using a picosecond laser. We coupled a 17x4x17.4x6.0 mm3 crystal to a 4x4 array of 6 mm MPPC (HPK S14161-6050HS-04) as well as a 50x50x8 mm3 crystals to an 8x8 array of 6 mm SiPMs (SensL ArrayJ-60035-64P-PCB). The detector modules were connected to a 64-ch CAEN V1740D waveform digitizer through custom readout board. We exposed the crystals to a point Co-57 source (emitting 122 keV photons) to generate flood maps, energy spectra, and line profiles.
Results: Geometry optimization show that @15 mm resolution, DC-SPECT’s sensitivity is 15x larger than that of a DHGC. At 0.01% system sensitivity, 4.1 mm resolution can be achieved which is close to whole-body PET. Also @10 mm resolution, system sensitivity of ~0.07% can be achieved. Note that the resolution values are without employing resolution recovery techniques. Flood map of 17.4x17.4 mm2 detector coupled to HPK MPPC array shows that all pixel-like volumes can be resolved even with unfavorable geometry of the converging pixels. Most of the CsI:Tl pixels in the 5x5 cm2 crystal coupled to SensL SiPM are also resolved, but as expected, the 20% QE of SensL SiPM array at 550nm (peak emitting wavelength of CsI:Tl) compared with 35% of the HPK MPPC array results in deteriorated flood map quality and lower energy resolution (~25% at 122 keV). CONCLUSION: Simulation results show that DC-SPECT outperforms other SPECT systems due to large detector coverage (~225 degrees), and the use of detectors with high IDR that enables many simultaneous projections of the FOV. We experimentally showed that laser-processing of thick CsI:Tl with converging pixel volume to avoid DOI blurring is possible. As shown in data, using detector-collimator combo acquiring projection data is ongoing. We will provide phantom images of DC-SPECT detector at the SNMMI conference. This work was supported in part by the NIH under Grant No. R01HL145160.