Accelerating Whole-Body Diffusion-weighted MRI with Deep Learning-based Denoising Image Filters.

Zormpas-Petridis, Konstantinos; Tunariu, Nina; Curcean, Andra; Messiou, Christina; Curcean, Sebastian; Collins, David J; Hughes, Julie C; Jamin, Yann et al. · Radiol Artif Intell · 2021

prospective_cohort · Level II

Where this comes from

Abstract

To use deep learning to improve the image quality of subsampled images (number of acquisitions = 1 [NOA<sub>1</sub>]) to reduce whole-body diffusion-weighted MRI (WBDWI) acquisition times. Both retrospective and prospective patient groups were used to develop a deep learning-based denoising image filter (DNIF) model. For initial model training and validation, 17 patients with metastatic prostate cancer with acquired WBDWI NOA<sub>1</sub> and NOA<sub>9</sub> images (acquisition period, 2015-2017) were retrospectively included. An additional 22 prospective patients with advanced prostate cancer, myeloma, and advanced breast cancer were used for model testing (2019), and the radiologic quality of DNIF-processed NOA<sub>1</sub> (NOA<sub>1-DNIF</sub>) images were compared with NOA<sub>1</sub> images and clinical NOA<sub>16</sub> images by using a three-point Likert scale (good, average, or poor; statistical significance was calculated by using a Wilcoxon signed ranked test). The model was also retrained and tested in 28 patients with malignant pleural mesothelioma (MPM) who underwent lung MRI (2015-2017) to demonstrate feasibility in other body regions. The model visually improved the quality of NOA<sub>1</sub> images in all test patients, with the majority of NOA<sub>1-DNIF</sub> and NOA<sub>16</sub> images being graded as either "average" or "good" across all image-quality criteria. From validation data, the mean apparent diffusion coefficient (ADC) values within NOA<sub>1-DNIF</sub> images of bone disease deviated from those within NOA<sub>9</sub> images by an average of 1.9% (range, 1.1%-2.6%). The model was also successfully applied in the context of MPM; the mean ADCs from NOA<sub>1-DNIF</sub> images of MPM deviated from those measured by using clinical-standard images (NOA<sub>12</sub>) by 3.7% (range, 0.2%-10.6%). Clinical-standard images were generated from subsampled images by using a DNIF.<b>Keywords:</b> Image Postprocessing, MR-Diffusion-weighted Imaging, Neural Networks, Oncology, Whole-Body Imaging, Supervised Learning, MR-Functional Imaging, Metastases, Prostate, Lung <i>Supplemental material is available for this article.</i> Published under a CC BY 4.0 license.