Ly John, Minarik David, Jögi Jonas, Wollmer Per, Trägårdh Elin
Department of Radiology, Kristianstad Hospital, Kristianstad, Sweden.
Department of Translational Medicine, Lund University, Malmö, Sweden.
EJNMMI Res. 2021 May 11;11(1):48. doi: 10.1186/s13550-021-00788-5.
The aim of the study was to develop and test an artificial intelligence (AI)-based method to improve the quality of [F]fluorodeoxyglucose (FDG) positron emission tomography (PET) images.
A convolutional neural network (CNN) was trained by using pairs of excellent (acquisition time of 6 min/bed position) and standard (acquisition time of 1.5 min/bed position) or sub-standard (acquisition time of 1 min/bed position) images from 72 patients. A test group of 25 patients was used to validate the CNN qualitatively and quantitatively with 5 different image sets per patient: 4 min/bed position, 1.5 min/bed position with and without CNN, and 1 min/bed position with and without CNN.
Difference in hotspot maximum or peak standardized uptake value between the standard 1.5 min and 1.5 min CNN images fell short of significance. Coefficient of variation, the noise level, was lower in the CNN-enhanced images compared with standard 1 min and 1.5 min images. Physicians ranked the 1.5 min CNN and the 4 min images highest regarding image quality (noise and contrast) and the standard 1 min images lowest.
AI can enhance [F]FDG-PET images to reduce noise and increase contrast compared with standard images whilst keeping SUV stability. There were significant differences in scoring between the 1.5 min and 1.5 min CNN image sets in all comparisons, the latter had higher scores in noise and contrast. Furthermore, difference in SUV and SUV fell short of significance for that pair. The improved image quality can potentially be used either to provide better images to the nuclear medicine physicians or to reduce acquisition time/administered activity.
本研究的目的是开发并测试一种基于人工智能（AI）的方法，以提高[F]氟脱氧葡萄糖（FDG）正电子发射断层扫描（PET）图像的质量。
使用来自72例患者的优质（采集时间为6分钟/床位）与标准（采集时间为1.5分钟/床位）或次标准（采集时间为1分钟/床位）图像对来训练卷积神经网络（CNN）。25例患者的测试组用于对CNN进行定性和定量验证，每位患者有5种不同的图像集：4分钟/床位、有和没有CNN处理的1.5分钟/床位，以及有和没有CNN处理的1分钟/床位。
标准的1.5分钟图像与经CNN处理的1.5分钟图像之间的热点最大或峰值标准化摄取值差异无统计学意义。与标准的1分钟和1.5分钟图像相比，经CNN增强的图像中变异系数（即噪声水平）更低。医生对图像质量（噪声和对比度）的评分中，1.5分钟经CNN处理的图像和4分钟图像最高，标准的1分钟图像最低。
与标准图像相比，人工智能可以增强[F]FDG-PET图像，以降低噪声并增加对比度，同时保持SUV稳定性。在所有比较中，1.5分钟图像集与经CNN处理的1.5分钟图像集之间的评分存在显著差异，后者在噪声和对比度方面得分更高。此外，该组SUV和SUV之间的差异无统计学意义。改善后的图像质量可潜在地用于为核医学医生提供更好的图像，或减少采集时间/给药剂量。
