DUSTER: dynamic contrast enhance up-sampled temporal resolution analysis method

详细信息    查看全文

• 作者：Gilad Liberman^a ; ^b ; Yoram Louzoun^c ; Moran Artzi^a ; ^d ; Guy Nadav^a ; ^e ; James R. Ewing^f ; ^g ; ^h ; Dafna Ben Bashat^a ; ^d ; ⁱ ; ^{dafnab@tlvmc.gov.il" class="auth_mail" title="E-mail the corresponding author} ; ^{dafnabenbashat@gmail.com" class="auth_mail" title="E-mail the corresponding author}
• 关键词：Dynamic contrast enhancement ; Temporal super resolution ; Bolus arrival time ; DUSTER
• 刊名：Magnetic Resonance Imaging
• 出版年：2016
• 出版时间：May 2016
• 年：2016
• 卷：34
• 期：4
• 页码：442-450
• 全文大小：2541 K

文摘

Dynamic contrast enhanced (DCE) MRI using Tofts' model for estimating vascular permeability is widely accepted, yet inter-tissue differences in bolus arrival time (BAT) are generally ignored. In this work we propose a method, incorporating the BAT in the analysis, demonstrating its applicability and advantages in healthy subjects and patients. A method for DCE Up Sampled TEmporal Resolution (DUSTER) analysis is proposed which includes: baseline T₁ map using DESPOT₁ analyzed with flip angle (FA) correction; preprocessing; raw-signal-to-T₁-to-concentration time curves (CTC) conversion; automatic arterial input function (AIF) extraction at temporal super-resolution; model fitting with model selection while incorporating BAT in the pharmacokinetic (PK) model, and fits contrast agent CTC while using exhaustive search in the BAT dimension in super-resolution. The method was applied to simulated data and to human data from 17 healthy subjects, six patients with glioblastoma, and two patients following stroke. BAT values were compared to time-to-peak (TTP) values extracted from dynamic susceptibility contrast imaging. Results show that the method improved the AIF estimation and allowed extraction of the BAT with a resolution of 0.8 s. In simulations, lower mean relative errors were detected for all PK parameters extracted using DUSTER compared to analysis without BAT correction (v_p:5% vs. 20%, K^trans: 9% vs. 24% and Kep: 8% vs. 17%, respectively), and BAT estimates demonstrated high correlations (r = 0.94, p < 1e^− 10) with true values. In real data, high correlations between BAT values were detected when extracted from data acquired with high temporal resolution (2 s) and sub-sampled standard resolution data (6 s) (mean r = 0.85,p < 1e^− 10). BAT and TTP values were significantly correlated in the different brain regions in healthy subjects (mean r = 0.72,p = < 1e^− 3), as were voxel-wise comparisons in patients (mean r = 0.89, p < 1e^− 10). In conclusion, incorporating BAT in DCE analysis improves estimation accuracy for the AIF and the PK parameters while providing an additional clinically important parameter.