Tajima Taku, Akai Hiroyuki, Yasaka Koichiro, Kunimatsu Akira, Akahane Masaaki, Yoshioka Naoki, Abe Osamu, Ohtomo Kuni, Kiryu Shigeru
Department of Radiology, International University of Health and Welfare Mita Hospital, 1-4-3 Mita, Minato-ku, Tokyo 108-8329, Japan; Department of Radiology, International University of Health and Welfare Narita Hospital, 852 Hatakeda Narita, Chiba 286-0124, Japan.
Department of Radiology, The Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108-8639, Japan.
Magn Reson Imaging. 2022 Jul;90:76-83. doi: 10.1016/j.mri.2022.04.005. Epub 2022 Apr 30.
T2-weighted imaging (T2WI) is a key sequence of MRI studies of the pancreas. The single-shot fast spin echo (single-shot FSE) sequence is an accelerated form of T2WI. We hypothesized that denoising approach with deep learning-based reconstruction (dDLR) could facilitate accelerated breath-hold thin-slice single-shot FSE MRI, and reveal the pancreatic anatomy in detail.
To assess the image quality of thin-slice (3 mm) respiratory-triggered FSE T2WI (Resp-FSE) and breath-hold fast advanced spin echo with and without dDLR (BH-dDLR-FASE and BH-FASE, respectively) at 1.5 T.
MR images of 42 prospectively enrolled patients with suspected pancreaticobiliary disease were obtained at 1.5 T. We qualitatively and quantitatively evaluated image quality of BH-dDLR-FASE related to BH-FASE and Resp-FSE.
The scan time of BH-FASE was significantly shorter than that of Resp-FSE (30 ± 4 s and 122 ± 25 s, p < 0.001). Qualitatively, dDLR significantly improved BH-FASE image quality, and the image quality of BH-dDLR-FASE was significantly better than that of Resp-FSE; as quantitative parameters, the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of BH-dDLR-FASE were also significantly better than those of Resp-FSE. The BH-dDLR-FASE sequence covered the entire pancreas and liver and provided overall image quality rated close to excellent.
The dDLR technique enables accelerated thin-slice single-shot FSE, and BH-dDLR-FASE seems to be clinically feasible.
T2加权成像（T2WI）是胰腺MRI研究的关键序列。单次激发快速自旋回波（单次激发FSE）序列是T2WI的一种加速形式。我们假设基于深度学习重建（dDLR）的去噪方法可以促进屏气薄层单次激发FSE MRI，并详细显示胰腺解剖结构。
评估1.5T时，使用和不使用dDLR的薄层（3mm）呼吸触发FSE T2WI（Resp-FSE）以及屏气快速进阶自旋回波（分别为BH-dDLR-FASE和BH-FASE）的图像质量。
对42例前瞻性纳入的疑似胰腺胆管疾病患者在1.5T时进行了MR成像。我们对BH-dDLR-FASE与BH-FASE和Resp-FSE相关的图像质量进行了定性和定量评估。
BH-FASE的扫描时间明显短于Resp-FSE（分别为30±4秒和122±25秒，p<0.001）。定性方面，dDLR显著改善了BH-FASE的图像质量，且BH-dDLR-FASE的图像质量明显优于Resp-FSE；作为定量参数，BH-dDLR-FASE的信噪比（SNR）和对比噪声比（CNR）也明显优于Resp-FSE。BH-dDLR-FASE序列覆盖了整个胰腺和肝脏，提供的整体图像质量接近优秀等级。
dDLR技术可实现加速薄层单次激发FSE，且BH-dDLR-FASE在临床上似乎是可行的。
