疊代式正子掃描影像重建使用邊緣保持事前資訊

Abstract

正子斷層掃描是一種常運用在臨床診斷的核子醫學影像系統，能提供許多醫學影像的資訊來協助醫生們診斷腫瘤、癌症或是腦部疾病，但是正子影像比較少具有結構性的影像資訊，而且相較於其他醫學影像，正子影像的空間解析度較低。如果能夠有效的提高影像的品質，將能提供更多有效的影像資訊來協助醫生診斷。 在本研究中，我們使用OpenGATE軟體來模擬正子斷層掃描系統，並且比較不同正子斷層影像重建方法的效果。在重建的過程中，我們會使用Median root prior 方法計算影像資訊，並且運用全變異最小化方法(Total variation minimization algorithm)搭配Proximal Splitting方法做影像重建，希望藉由此方法能夠提高正子斷層影像的影像品質及解析度。 除了改善正子斷層掃描影像之外，我們也使用了來自台大醫院的阿茲海默症患者的影像，阿茲海默症是一種不可治癒的腦部疾病，阿茲海默症的患者會隨著時間，漸漸的失去記憶及思考的功能，雖然不可以治癒，但是提早治療能夠降低阿茲海默症所造成的傷害，因此我們設計一個多模式影像平台能將正子斷層掃描影像及核磁共振影像對位，並設計可以讓醫生圈出特別的區域，透過這樣的方式，醫生能夠有更多客觀的影像資訊來診斷疾病。我們希望這些方法幫助阿茲海默症的患者能夠即早治療，降低阿茲海默症所帶來的傷害。

Positron emission tomography (PET) is a nuclear medicine technique that can help the doctor to diagnose the disease such as tumor, cancer, or brain disease. However, there are still some disadvantages in PET system. For example, comparing with the different medical imaging modalities, the spatial resolution of PET is relatively poor. If we can improve the image quality, PET can provide more image information for doctors diagnosing disease. In this work, we applied the OpenGATE package to simulate the PET system and compared the efficacy of different reconstruction algorithms. We took advantages of the edge-preserving prior for PET image reconstruction to improve PET image quality. In this work we also employed the medical image data from National Taiwan University Hospital. We developed a multi-modality imaging platform to register functional and anatomical images. This allowed for a quantitative and objective diagnosis of Alzheimer’s disease (AD). AD is an irreversible, progressive brain disease that slowly destroys memory and thinking skills. Early treatment could slow down the damage caused by AD. The knowledge of the anatomical features in MR and CT images can be exploited to better evaluate the tracer dynamics in PET images by first registering anatomical information from MRI with CT images and then with functional information contained in PET. With all of methods, an early diagnosis of may be anticipated.