| The total-body PET has brought infinite possibilities to the development of PET,and the improvement of timing resolution has greatly improved PET image quality.However,these also bring following challenges to PET manufacturing and image reconstruction.(1)The number of lines of response will increase rapidly with the increase of axial field of view,increasing the computation cost of reconstruction using the existing ray-tracing based reconstruction algorithm.(2)With the introduction of TOF(time-of-flight),the computation cost and memory consumption of PET reconstruction will increase.Moreover,with the improvement of timing resolution,the computing resources consumption of TOF-PET image reconstruction will increase further.(3)PET with long axial field of view needs more crystal detectors,leading to a significant increase in manufacturing cost.In this paper,we researched on an efficient TOF-PET reconstruction and PET with sparse detectors,mainly including following aspects.(1)We implemented Direct Image Reconstruction for TOF(DIRECT)on GPU,enabling the framework to process the system kernel function changing with the spatial position,and optimized the forward-projection by sparsity of data.(2)We simulated PET data with different timing resolution using GATE,effect of timing resolution on image reconstruction is analyzed and verified.(3)We proposed that benefitting from improved timing resolution,modern clinical PET system with sparse detector configurations and high timing resolution may attain comparable image quality to that with a dense detector configuration and low timing resolution.The above conjecture is verified by GATE simulation experiment.Compared with PET with dense detector,sparse detector configurations can reduce the number of crystal detector to half,greatly reducing the manufacturing cost,or double the axial extension of original PET with the same number of detectors to obtain a low-cost long-axis PET with sparse detectors. |
