3446

Prospective evaluation of motion correction and complex averaging in ultra-high-b-value DWI for image quality and lesion conspicuity in oncologic follow-up examinations
Constantin Dreher1, Tristan Anselm Kuder2, Stefan Windhaber1, Franziska K├Ânig1, Daniel Paech1, Anoshirwan Tavakoli1, Regula Gnirs1, Thomas Benkert3, Heinz-Peter Schlemmer1, and Sebastian Bickelhaupt1

1Radiology, German Cancer Research Center, Heidelberg, Germany, 2Medical Physics in Radiology, German Cancer Research Center, Heidelberg, Germany, 3Siemens Healthcare GmbH, Erlangen, Germany

Synopsis

In oncological, abdominal MRI, the value of motion corrected Diffusion-Weighted Imaging (DWI) including complex averaging with high and ultra-high b-values remains to be defined. This is of special importance regarding the balance between normal tissue suppression and suspicious lesion demarcation. This prospective study investigated 41 patients with an oncologically optimized prototype-DWI (with complex averaging, motion correction between averages, rescaling of motion corrupted averages, background suppression). Image quality, tissue differentiation and lesion detection/characterization were significantly increased in high (b900) as compared to ultra-high (b1500) DWI. At the same time, apparent signal-intensity ratio of lesion/normal tissue was not significantly different.

Introduction

Diffusion-Weighted Imaging (DWI) is of high diagnostic importance in the setting of oncological magnetic resonance imaging (MRI) follow-up examinations. However, DW-MRI is challenged by motion artifacts and little SNR in the retroperitoneum, which is why it has to be optimized [1, 2]. Current developments allow using adapted DWI sequences with improved motion correction between the averages and complex averaging modifications. However, in comparison to DW-MRI of the prostate [3], the influence on the balance of normal tissue suppression and lesion demarcation of suspicious findings in high- and ultra-high b-value DWI remains unclear. The target of this study is therefore to prospectively evaluate an oncologically optimized DWI sequence with improved motion correction between the averages and complex averaging in abdominal MRI for both its general image quality and potential in the presentation of malignant lesions by the use of high and ultra-high b-values.

Methods

This IRB-approved, prospective study included 41 prospectively acquired oncological MRI examinations (mean age: 57 years; male/female: 16/25) of the abdomen, by the use of a 1.5T MRI scanner (MAGNETOM Aera, Siemens Healthcare GmbH, Germany) with a prototypical EPI DWI sequence (b=0,50,900,1500s/mm²), including complex averaging, motion correction between the averages, rescaling of motion corrupted averages, and background suppression. The following parameters were used: TR=7.9 s; TE=57 ms; FOV 480x270 mm²; matrix 164x92; resolution 3x3 mm², interpolated to 1.5x1.5 mm²; slice thickness 5 mm; bandwidth 2540 Hz/Px; b-values 0, 50, 900, 1500 s/mm² (1, 1, 16 and 18 averages); gradient mode 3D-diagonal; 90 slices in 3 steps, 15 min. High b-value DWI with b900 (s/mm²) and ultra-high b-value DWI with b1500 (s/mm²) were compared as follows: Both normal tissue image quality, tissue differentiation parameters and diagnostic confidence for characterization and detection of malignant lesions were rated by two independent, blinded readers using a 5-point Likert Scale [4]. The apparent signal intensities of the malignant lesions and the normal tissue of the same organ were divided, resulting in lesion-to-normal-tissue ratio (LNTR). Statistics included Wilcoxon-signed-rank test and kappa statistic for interreader agreement analysis (p<0.05).

Results

Image quality parameters (contour sharpness right liver, left liver and pancreas) and tissue differentiation for upper abdominal organs and retroperitoneal structures were significantly higher with b900 as compared to b1500 optimized DWI (Figure 1 and 2) (p<0.001). Interreader reliability test showed good agreement (kappa=0.715) (p<0.001). Thirty-seven suspicious lesions were detected (12 lymphatic, 8 hepatopancreatic and 17 lesions of other origins). Reader confidence for characterization/detection of malignant lesions was significantly higher using the b900 (mean: 4.4±0.6) as compared to the b1500 optimized DWI (mean: 4.0±0.8) (p=0.001) with an overall good interreader agreement (kappa=0.723) (p<0.001). The increased confidence of lesion recognition on b900 remained significant in subgroup analysis of lymphatic lesions (p=0.003), but not with regard to hepatopancreatic and lesions of other origins (p=0.102, 0.317) (Figure 1 and 3). Apparent-LNTR of b900 vs. b1500 optimized DWI was not significantly different with regard to all lesions (mean: 3.0±2.3 vs. 2.7±2.2, p=0.067), and the subgroup analysis of lymphatic (mean: 2.2±1.0 vs. 2.0±1.5, p=0.117), hepatopancreatic (mean: 1.8±0.5 vs. 1.8±0.8, p=0.944) and lesions of other origins (mean: 4.1±2.9 vs. 3.7±2.8, p=0.148).

Discussion

This study evaluated the applicability of an adapted DWI sequence in oncologic follow-up examinations using improved motion correction between the averages and complex averaging modification with high- and ultra-high b-values for improved image quality and lesion detection. The findings demonstrated that high b-value provided an improved image quality and lesion conspicuity as compared to ultra-high b-value DWI. These findings were significant with regard to image quality, while at the same time ultra-high b-value DWI was inferior to high b-value DWI concerning the reader confidence of lesion detection and characterization. Consequently, the potential and usage of ultra-high vs. high b-values in oncological examinations of the abdominal region should be critically evaluated in future studies. However, this does not imply any statement about quantitative DWI analysis (e.g. ADC).

Conclusion

This prospective study demonstrates the feasibility of a novel approach of oncologically optimized DWI with improved motion correction between the averages and complex averaging modification in oncological follow-up MRI examinations. The results demonstrated that high b-value DWI (b=900s/mm²) provided better image quality and lesion conspicuity as compared to ultra-high b-values (b=1500s/mm²) and might be preferred for oncologic follow-up abdominal MRI examinations.

Acknowledgements

Special acknowledgments and thanks to Siemens Healthineers for providing the optimized protoype sequence and protocol

References

1. Taron, J., et al., Clinical Robustness of Accelerated and Optimized Abdominal Diffusion-Weighted Imaging. Invest Radiol, 2017. 52(10): p. 590-595.

2. Zhang, T.T., et al., Differentiation of pancreatic carcinoma and mass-forming focal pancreatitis: qualitative and quantitative assessment by dynamic contrast-enhanced MRI combined with diffusion-weighted imaging. Oncotarget, 2017. 8(1): p. 1744-1759.

3. Ueno, Y., et al., Ultra-high b-value diffusion-weighted MRI for the detection of prostate cancer with 3-T MRI. J Magn Reson Imaging, 2013. 38(1): p. 154-60.

4. de Winter, J.C.F., Dodou, D., Five-point Likert items: T test versus Mann–Whitney–Wilcoxon. Practical Assessment, Research & Evaluation, 2010. 15(11): p. 16.

Figures

Figure 1: Investigated image quality and normal tissue differentiation of the right and left liver (A) and the pancreas (B) and investigated lesion characterization and detection of for example a hepatic malignancy (C) with oncologically optimized high b-value (900 s/mm²) (1) and ultra-high b-value (1500 s/mm²) (2) Diffusion-Weighted Imaging (DWI). Increased contour sharpness of the liver and the pancreas (red arrows) by high as compared to ultra-high b-value DWI. Good recognition of the hepatic lesion with well-defined borders (yellow arrows) both by high and ultra-high b-value DWI.

Figure 2: Image quality and tissue differentiation of oncologically optimized Diffusion-Weighted Imaging (DWI) MRI with b900 compared to b1500: Reading with a 5-point Likert-Scale: Analysis of both the contour sharpness of the normal tissue (NT) right liver (NT_RI_LI), left liver (NT_LE_LI) and pancreas (NT_PANC) and the tissue differentiation of the upper abdominal organs and retroperitoneum (NT_ABD). p*** = significant difference with p<0.001.

Figure 3: Diagnostic confidence for lesion characterization and detection of oncologically optimized Diffusion-Weighted Imaging (DWI) MRI with b900 compared to b1500: Reading with a 5-point Likert-Scale. Subgroup analysis for lymphatic malignancies (LYM), hepatopancreatic malignancies (HEP_PANC), and malignancies of other origins (OTH). p** = significant difference with p<0.01.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3446