Feasibility of improving the accuracy of under-sampled cerebral CT perfusion results using logistic fitting algorithm
10.3760/cma.j.cn112271-20240312-00088
- VernacularTitle:应用logistic拟合算法提升低采样颅脑CT灌注结果准确性的可行性研究
- Author:
Xiang ZHAO
1
;
Fengtan LI
;
Wanhui ZHOU
Author Information
1. 天津医科大学总医院医学影像科，天津 300052
- Publication Type:Journal Article
- Keywords:
Low dose;
Cerebral CT perfusion;
Logistic model;
Fitting algorithm
- From:
Chinese Journal of Radiological Medicine and Protection
2025;45(1):63-68
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To evaluate the effects of the logistic fitting algorithm in improving the calculation accuracy of low-sampled cerebral CT perfusion (CTP) and to explore the application value of this algorithm in reducing irradiation doses to cerebral CTP.Methods:Image data from 28 patients who underwent cerebral CTP were retrospectively analyzed. A total of 192 regions of interest (ROIs) were circled in the basal ganglia and the main blood-supplying areas of the anterior, middle, and posterior cerebral arteries. For each ROI, the time-density curve (TDC) was plotted, from which some data points were deleted to decrease the sampling frequency to half of the conventional scan, thus simulating low-dose scanning and obtaining low-dose TDCs. The logistic model was applied to fit and complete the low-dose TDCs. The potential decrease in radiation dose was assessed. Conventional TDCs, low-dose TDCs, and TDCs processed using the logistic fitting algorithm were compared. Perfusion calculations were performed based on these TDCs, and the calculated cerebral blood volume (CBV), cerebral blood flow (CBF), time to peak (TTP), and mean transit time (MTT) were compared and analyzed.Results:The total radiation dose for cerebral CTP examination could be reduced to 52% of the routine dose. The mean correlation coefficient R between the TDCs derived using the logistic fitting algorithm and the conventional TDCs was 0.958 ± 0.03. The CBV, CBF, TTP, and MTT calculated using the logistic fitting algorithm were compared with the conventional result, and the coefficients of determination R2 of linear regressions were determined at 0.943, 0.942, 0.955, and 0.891, respectively, indicating extremely high consistency(ICC > 0.90). Furthermore, the R2 values determined using the logistic fitting algorithm were all higher than those derived without applying the fitting algorithm. Conclusions:Applying the logistic fitting algorithm to under-sampled cerebral CTP can yield calculation result that are highly consistent with those of conventional sampling. By combining under-sampling with the logistic algorithm, the irradiation doses can be reduced while guaranteeing the accuracy of the effective perfusion result, demonstrating high application value.
