The Low Counts PET Image Reconstruction Using The Anatomical Information As Priors

Positron Emission Tomography(abbreviated as PET)is one of the most sensitive clinical molecular imaging technology in the field of nuclear medicine.By detecting the distribution of the radioactive tracer injected into the organism,it can noninvasively,dynamically and quantitatively assess the metabolic level,biochemical level and functional activity of various organs in the organism.This helps PET make great effects in the application,like the diagnosis and identification of disease,the judgement of the disease,the evaluation of the curative effect,the research of organ function and the development of new drugs.Compared with other medical imaging devices,such as Computed Tomography(abbreviated as CT)and Magnetic Resonance Imaging(abbreviated as MRI),PET has not fully exploited its application potential.The expansion of application for PET is a key point,but also puts forward higher demands to the performance of PET.The purpose of reconstruction is to restore the location and distribution information of radionuclides,and the counting of projection data directly affects the quality of the final image.Because of the imaging mode,specific requirements or design of the system,some needs of the new applications of PET can not be met by the traditional imaging methods.This could result in high noise level,undistinguishable edges and details,low SNR and other shortcomings in the reconstructed image.With the development of the current multimodality medical image equipments such as PET/CT and PET/MRI,PET image and the accurate anatomical image can be obtained.Combining the self-similarity of PET image and the anatomical prior information to do the image reconstruction is the focus of this paper.By employing the general algorithm framework LMROS,this paper puts forward two new algorithms using the regularization and selectively incorporating the anatomical information to do the image reconstruction,named LMROS-NLM-CT and LMROS-BILCT.Preclinical mouse experiments and clinical phantom and body experiments under different counting conditions were performed.And results show that the use of the anatomical prior information in regularization achieves high-quality PET image reconstruction even in the low counts conditions.It is hoped that the work in this paper can provide valuable reference for the research of image reconstruction using regularization combined with the anatomical information.