论文标题
有关AAPM深度学习光谱CT大挑战的报告
Report on the AAPM deep-learning spectral CT Grand Challenge
论文作者
论文摘要
该特别报告总结了2022年AAPM大挑战，该挑战是深度学习光谱计算机断层扫描（DL-Spectral CT）图像重建。挑战的目的是开发最准确的图像重建算法，用于解决与使用三个组织映射分解的快速KVP开关双能CT扫描相关的反问题。参与者可以选择使用深度学习（DL），迭代或混合方法。从18个研究小组收到了测试阶段提交。获胜和第二名的球队都取得了高度准确的结果，其中RMSE几乎为单浮点精度为零。前十名的结果也达到了高度的准确性。结果，该特别报告概述了这些组中每个组开发的方法。 DL-Spectral CT挑战成功地建立了一个论坛，以基于深度学习的深度学习来开发图像重建算法，该算法解决了与光谱CT相关的重要逆问题。
This Special Report summarizes the 2022 AAPM Grand Challenge on Deep-Learning spectral Computed Tomography (DL-spectral CT) image reconstruction. The purpose of the challenge is to develop the most accurate image reconstruction algorithm for solving the inverse problem associated with a fast kVp switching dual-energy CT scan using a three tissue-map decomposition. Participants could choose to use deep-learning (DL), iterative, or a hybrid approach. Test phase submission were received from 18 research groups. Both the winning and second place teams had highly accurate results where the RMSE was nearly zero to single floating point precision. Results from the top ten also achieved a high degree of accuracy; and as a result this special report outlines the methodology developed by each of these groups. The DL-spectral CT challenge successfully established a forum for developing image reconstruction algorithms based on deep-learning that address an important inverse problem relevant for spectral CT.
