上海科技大学知识管理系统

Objective. The objective is to generate reliable K-i parametric images from F-18-fluorodeoxyglucose (F-18-FDG) total-body positron emission tomography (PET) with clinically acceptable scan durations using Patlak and shallow machine learning algorithms, under conditions of limited computational and data resources. Approach. We proposed a robust and fast algorithm named Patlak-KXD to generate K-i images from dynamic PET images with shortened scan durations. In the training phase, K-means is employed to generate a K-i-balanced training dataset. Subsequently, XGBoost is utilized to learn the mapping relationship between the tissue-to-blood standardized uptake ratio (SUR) time curves and Patlak-based K-i values using this balanced dataset. In the prediction phase, the trained XGBoost can generate K-i images by calculating the K-i values from voxel-based SUR time curves obtained from the dynamic images. We compared the accuracy of K-i images generated by both the Patlak-KXD and the traditional Patlak methods across a range of shortened scan durations, and the differences in K-i images generated by the XGBoost model using static (Patlak-KXS) and dynamic PET inputs. Main results. The K-i images generated by the Patlak-KXD from just a 4 min (56-60 min) dynamic F-18-FDG total-body PET scan are comparable to those generated by the traditional Patlak method using 40 min (20-60 min) dynamic PET images, as demonstrated by a normalized mean square error of 0.13 and a Pearson's correlation coefficient of 0.94 on average. The K-i images generated by the Patlak-KXD is robust to the scan duration, and the quality of K-i images generated from Patlak-KXD is superior to those from Patlak-KXS as scan duration >10 min. Significance. Reliable K-i images can be rapidly generated using shallow machine learning algorithms from dynamic F-18-FDG total-body PET scans with durations as short as four minutes. This total-body K-i parametric imaging method has potential to be used in clinical nuclear medicine and molecular imaging.