3022

Comparison of 2D BLADE and Spin-Echo Echo­-Planar Diffusion-Weighted Brain MRI at 3 Tesla: Preliminary Experience in Children
Aaron S McAllister1, Lacey J. Lubeley1, Bhavani Selvaraj1, Ning Jin2, Kun Zhou3, Mark Smith1, Ramkumar Krishnamurthy1, and Houchun Harry Hu1

1Radiology, Nationwide Children's Hospital, Columbus, OH, United States, 2Siemens Medical Solutions, Cleveland, OH, United States, 3Siemens Shenzhen Magnetic Resonance Ltd., Shenzhen, China

Synopsis

We describe our preliminary experience using a GRASE (gradient-echo and spin-echo hybrid) based DWI-BLADE pulse sequence in 53 pediatric patients at 3T. On a 4-point scale for rating diagnostic image quality and impact of artifacts, 1 (best) – 4 (worst), a neuroradiologist scored conventional spin-echo EPI DWI 2.4±0.7 whilst BLADE scored 1.1±0.3 (p<0.01). Overall, DWI-BLADE exhibited less geometric distortion at the periphery of the brain, and reduced signal pile-ups at areas of high susceptibility. The pulse sequence is particularly useful in patients with shunts and dental fixtures and is a viable alternative to conventional spin-echo EPI DWI.

Introduction / Background

In comparison to conventional diffusion-weighted (DW) MRI using single-shot spin-echo (SE) echo-planar (EPI) trajectories, turbo-spin-echo (TSE) (1) and GRASE-based (2) DW BLADE imaging has the advantages of reduced sensitivity to B0 inhomogeneity and decreased image blurring caused by T2 blurring along the readout echo train. Additionally, TSE and GRASE-based BLADE acquisitions have minimal sensitivity to geometric distortions and resultant signal pile-ups. While SE-EPI sequences remain the clinical workhorse for brain DWI, the advantages of BLADE can be exploited in certain patients. In this work, we compare a prototype 2D GRASE non-­Cartesian BLADE sequence with a conventional Cartesian SE-EPI acquisition in a group of pediatric patients. We hypothesize that the BLADE technique will yield brain images of comparable diagnostic quality to conventional imaging in cases where artifact-causing materials are not present, and that it will be potentially more beneficial in instances where shunts and irremovable orthodontia are present.

Materials and Methods

Data collection and scoring: All data were obtained on a 3T clinical scanner (MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany). Fifty-three patients (10.4±7.9 years, 25F/28M) underwent both SE-EPI DW imaging and the prototype BLADE sequence. A neuroradiologist evaluated the severity of image artifacts and whether their presence affected diagnostic image quality (IQ) in a non-blinded fashion. The radiologist answered whether BLADE or ­SE-EPI was preferred in each case. A 4-­point Likert score was given for each sequence based on the presence of artifacts and signal pile-­up near air tissue interfaces and in the presence of shunts/orthodontia: 1: none, 2: mild­ly diagnostic, IQ not affected, 3: moderate­, IQ partially affected, and 4: significant artifacts,­ IQ heavily affected. Using a similar 1-4 scale, the neuroradiologist also rated SE-EPI and BLADE in the category of motion and noise.

% Geometric Distortion: Additionally, anterior-­posterior and right-­left dimensions of the brain were measured at co-registered locations on both BLADE and SE-EPI DWI sequences and compared to reference measurements obtained from a reference Cartesian 3D T1­-weighted inversion ­recovery gradient echo scan by two image analysts. A percent geometric distortion value was computed for each case, in both directions.

Parameters for SE-­EPI were: 1.5x1.5mm in-plane resolution, 4mm slices, TR/TE=4100/81ms, fat suppression, GRAPPAx2 with 40 references lines, partial-Fourier phase encoding, 1446Hz/pixel bandwidth, 0.8ms echo spacing, an EPI factor=192, four diffusion directions, and two signal averages for b=0 and three signal averages for b=1000, scan time ~2min. For the prototype GRASE BLADE sequence: 1.3x1.3mm resolution, 4mm slices, TR/TE=5200/41ms, fat suppression, no GRAPPA and no partial-Fourier encoding, no signal averaging, 520Hz/pixel bandwidth, 11ms echo spacing, EPI factor of 3 for GRASE, and a turbo factor of 11 with reduced refocusing flip angles along the echo train to minimize SAR, scan time ~4min.

Results

Ratings: In 46% of the cases, BLADE was preferred over SE-EPI; in 45% of the cases, both sequences were rated equally. Average scores for SE-­EPI in the 53 patients were 2.4±0.7 vs. BLADE 1.1±0.3 (p<0.01) in diagnostic quality and artifact impact. Scores in the category of motion were 1 for all SE-EPI cases and 1.04±0.3 for BLADE (p=0.16). Motion artifacts were minimal on both sequences in general. The neuroradiologist did notice that ADC maps from BLADE were slightly more noisy than those from SE-EPI, with scores of 1.02±0.13 for SE-EPI and 2.02±0.45 for BLADE (p<0.01).

% Geometric Distortion: In the frontal and temporal lobes near the anterior surface of the brain, the percent geometric distortion in the A/P direction varied widely from -­38.8% to 70.1% for ­SE-EPI (0.4%±16.4%), whereas for BLADE it was significantly lower, from ­-8.6% to 17.5%, while the average was similar -0.8%±5.12%. However, in the R/L direction, the distortion varied from -27.5% to 59.5% for SE-EPI (average: 7.0%±15.7%), whereas for BLADE it was significantly lower, from -­7.6% to 20.9% (1.3%±4.8%). More posteriorly in the occipital lobe, the percent geometric distortion in the A/P direction varied from -52.2% to 21.0% for ­EPI (3.3%±13.0%) and -32.8 to 7.1% (-1.8%±5.5%) for BLADE. In the R/L direction, -13.9% to 21.3% for ­EPI (0.8%±7.2%) and –11.5 to 15.7% (-0.1%±5.2%) for BLADE.

Conclusion

BLADE DWI is feasible in pediatric patients at 3T. Despite longer scan times than a conventional SE-EPI approach, it exhibits less distortion near air tissue interfaces and in the presence of shunts/orthodontia. One future direction of work is to compare BLADE DWI to RESOLVE DWI. A limitation of the existing data analysis and study is that we did not directly compare ADC values computed from EPI vs. BLADE. Overall, ­EPI exhibited more edge distortions and more signal pile­-ups in this preliminary cohort of pediatric patients.

Acknowledgements

The Department of Radiology at Nationwide Children's Hospital thanks Siemens Medical Solutions for research support and acknowledge the time and effort of the hospital's MRI technologists.

References

(1) Pokorney AL, et al. Comparison of 2D single-shot turbo-spin-echo and spin-echo echo-planar diffusion weighted brain MRI at 3.0 Tesla: preliminary experience in children. Clinical Imaging 2017 42:152-157.

(2) Zhou K, et al. Non-CPMG PROPELLER diffusion imaging: comparison of phase insensitive preparation with split acquisition. Proc. ISMRM 2018, #5320.

(3) Henninger B, Kremser C. Diffusion weighted imaging for the detection and evaluation of Cholesteatoma. World J Radiol 2017; 9:217-222.

Figures

Figure 1: Images from a 17-years-old boy with multiple malignancies, including T-cell lymphoma, low grade glioma, and adenocarcinoma. DWI images are shown for (A) SE-EPI and (B) BLADE. Corresponding ADC maps are shown in (C) for SE-EPI and (D) BLADE. Near the sinuses, noticeable geometric distortions and signal pile-ups (dashed area, arrows) seen in (A) conventional imaging are less noticeable in (B). In the middle row near the superior aspects of the brain, the two sequences appear comparable. Another comparison of (E,G) SE-EPI and (F,H) BLADE show signal pile-ups in (E, arrows) that obscure the margins of the large left parietal lobe tumor along the.

Figure 2: Exemplary results in a 14-years-old female with a history of intracranial infection leading to empyema, cavernous sinus thrombosis, left middle-cerebral-artery infarct, seizures, and dysarthria. DWI source images are shown for (left) conventional SE-EPI based sequence and (right) BLADE. Note signal distortion and pile-up seen in the SE-EPI data (arrow, dashed area) are not seen in the BLADE. However, BLADE data appear low in signal-to-noise ratio centrally (dotted region). ADC maps are shown in the lower row for a separate slice highlighting hemorrhage (open arrow) and have similar appearances.

Figure 3: Exemplary results in a 5-years-old male with right frontal lobe craniotomy for epileptic focus resection. DWI source images are shown for (A) conventional SE-EPI based sequence and (B) BLADE. Note signal distortion and pile-up seen in the SE-EPI data obscures part of the resected area (arrows). This is not seen in the BLADE data. ADC maps are shown (C, D) in the lower row for a separate slice highlighting hemorrhage along the resection margins and within the extra-axial spaces (open arrow). In the ADC maps here, the BLADE data appears less grainy centrally than the SE-EPI data (dotted region).

Figure 4: Representative data in a 1-week-old girl (top row),evaluated for birth asphyxia and hypoxic ischemic encephalopathy. DWI BLADE data in (B, D) show better detail and more accurate geometric representation of the patient’s brain than conventional SE-EPI data (A, C). Additional data in the lower row of a 10-years-old girl with braces. BLADE data demonstrates reduced susceptibility and related artifacts.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3022