成像体模
背景(考古学)
图像质量
核医学
心室
PET-CT
医学
迭代重建
心脏宠物
断层摄影术
重建算法
放射科
计算机科学
正电子发射断层摄影术
人工智能
内科学
图像(数学)
古生物学
生物
作者
Michel Hesse,Sébastien Marchandise,Bernhard Gerber,Véronique Roelants
标识
DOI:10.1097/mnm.0000000000001694
摘要
Background Evaluation of left atrium (LA) remodeling is becoming increasingly relevant in understanding several pathological cardiac conditions. 18 F-FDG-PET/computed tomography (CT), the current gold standard for metabolic evaluation of the left ventricle, could be extended to LA using the latest PET technologies. We sought to perform a phantom study to optimize the reconstruction algorithm in this context. Methods The liver, heart cavity and walls of an anthropomorphic phantom were filled with typical patient 18 F-FDG activity concentrations. Acquisitions were performed on an analog and on a digital TOF-PET/CT, and reconstructed with and without resolution recovery (RR). The Richardson-Lucy RR method was used, either through a third-party software or through the PET/CT manufacturer algorithm. Activity recoveries in the atria and ventricles and signal-to-noise ratios were evaluated to identify the best reconstruction and RR parameters. The same methodology was applied on a patient cardiac study. Results Analog PET/CT with the third-party RR cannot improve the activity recovery without markedly degrading the image quality. For the digital PET/CT, the optimal algorithm was the manufacturer RR reconstruction using four iterations and 15 subsets combined with five RR iterations. This reconstruction improved the LA activity recovery from 58% to 70% while preserving images of diagnostic quality. Similar results were obtained for the patient study. Conclusion The digital TOF-PET/CT with the identified optimal reconstruction can be used to quantitatively analyze the LA uptake in 18 F-FDG cardiac studies while still preserving image reading quality. This may lead to more precise cardiovascular disease status evaluation, especially when atria are concerned.
科研通智能强力驱动
Strongly Powered by AbleSci AI