1757
Evaluation of 3D FLAIR Combined Compressed Sense Technology in Cerebrospinal Fluid Flow Artifacts Reduction: Comparison with 2D FLAIR1Departments of Radiology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, WuHan, China, 2Philips Healthcare, BeiJing, China
Synopsis
3D FLAIR TSE imaging can effectively inhibit cerebrospinal fluid flow artifacts, but the scanning time is long, affecting clinical application. Compressed sensing (CS) is a technique to accelerate the acquisition of magnetic resonance imaging (MRI) by using sparsity constraints during readout measurements of k-space. This study aimed to compare the cerebrospinal fluid flow artifacts and overall image quality in CS 3D FLAIR TSE imaging combined and in 2D FLAIR TSE imaging. Results showed that CS 3D FLAIR TSE yielded better image quality, enhanced diagnosis performance, and reduced fluid flow artifacts when compared to the traditional 2D FLAIR TSE sequence.
INTRODUCTION
Cerebrospinal fluid artifacts in conventional 2D FLAIR TSE imaging on 3.0 T MRI sometimes remains a disturbance to diagnosis in clinical scenarios(1). 3D FLAIR TSE imaging can effectively inhibit cerebrospinal fluid flow artifacts, but the scanning time is long, which hampers its clinical applications(2). Compressed sensing (CS) technique can increase the acquisition speed by recovering image information from highly undersampled k-space, taking advantage of the sparse nature of 3D images(3).This study aimed to compare cerebrospinal fluid flow artifacts and the overall image quality in the CS 3D FLAIR TSE imaging and in the regular 2D FLAIR TSE imaging.METHOD AND MATERIALS
Fifty-six patients (mean age 38.2 years, age range 22-75 years, 24 women, 32 males) were prospectively recruited for 3.0 T MR exams (Ingenia CX, Philips Healthcare, Best, the Netherlands) including both CS 3D and 2D FLAIR imaging at consistent spatial resolution. The scan parameters were described in Figure 1. The resulting images were sent to the IntelliSpace Portal (Philips Healthcare, Best, the Netherlands) where two central radiographic experts performed a subjective score on the flow artifact using a five-point Likert scale. SNR, CNR, and contrast-ratio (CR) were compared between the two sequences on the white matter and gray matter where the signal was generally uniform, methods described in Figure 2. The image quality subjective score was compared using Friedman test between conventional 2D FLAIR and 3D FLAIR CS, and the difference between SNR, CNR and CR was compared using two-sample T test. For all tests, values of p <0.05 were considered to indicate statistical significance.RESULTS
The overall imaging quality of CS 3D FLAIR was significantly better than conventional 2D FLAIR images (Figure 3). The inhibitory effect of CS 3D FLAIR on cerebrospinal fluid artifacts was better than that of 2D FLAIR according to the subjective score (U = 3.66, P < 0.01). The SNR, CNR and CR scores of 3D FLAIR CS images were significantly higher than those of 2D FLAIR images.DISCUSSION
Compared with the conventional 2D FLAIR sequence, CS 3D FLAIR has obvious advantages in suppressing cerebrospinal fluid flow artifacts, especially in the front of the brainstem (54 higher scores), and the overall image quality was better given similar scan time. These findings are expected to help the diagnosis cerebral infarction, demyelinating lesions and auditory nerves.Acknowledgements
Thank you very much for Philips Healthcare support. Thank you to JiaZheng Wang,Zhongping Zhang for your great help to this project.References
1.Qian Yinfeng, Zhang Cheng, Wu Jinmin, etc. Cerebrospinal fluid artifacts on 2D and 3D fluid attenuation inversion recovery images of 3.0T MR. Magnetic resonance imaging, 2012,3 (3): 179 ≤ 183.
2.Sharma SD, Fong CL, Tzung BS, Law M, Nayak KS. Clinical image quality assessment of accelerated magnetic resonance neuroimaging using compressed sensing. Invest Radiol.2013;48:638–45.
3.Lustig M, Donoho DL, Santos JM, Pauly JM (2008) Compressed sensing MRI: a look at how CS can improve on current imaging techniques. IEEE Signal Process Mag 25:72–82.
Figures
