Validation of a Denoising Method Using Deep Learning–Based Reconstruction to Quantify Multiple Sclerosis Lesion Load on Fast FLAIR Imaging

Abstract

BACKGROUND AND PURPOSE: Accurate quantification of WM lesion load is essential for the care of patients with multiple sclerosis. We tested whether the combination of accelerated 3D-FLAIR and denoising using deep learning-based reconstruction could provide a relevant strategy while shortening the imaging examination. MATERIALS AND METHODS:Twenty-eight patients with multiple sclerosis were prospectively examined using 4 implementations of 3D-FLAIR with decreasing scan times (4 minutes 54 seconds, 2 minutes 35 seconds, 1 minute 40 seconds, and 1 minute 15 seconds). Each FLAIR sequence was reconstructed without and with denoising using deep learning-based reconstruction, resulting in 8 FLAIR sequences per patient. Image quality was assessed with the Likert scale, apparent SNR, and contrast-to-noise ratio. Manual and automatic lesion segmentations, performed randomly and blindly, were quantitatively evaluated against ground truth using the absolute volume difference, true-positive rate, positive predictive value, Dice similarity coefficient, Hausdorff distance, and F1 score based on the lesion count. The Wilcoxon signed-rank test and 2-way ANOVA were performed. RESULTS:Both image-quality evaluation and the various metrics showed deterioration when the FLAIR scan time was accelerated. However, denoising using deep learning-based reconstruction significantly improved subjective image quality and quantitative performance metrics, particularly for manual segmentation. Overall, denoising using deep learning-based reconstruction helped to recover contours closer to those from the criterion standard and to capture individual lesions otherwise overlooked. The Dice similarity coefficient was equivalent between the 2-minutes-35-seconds-long FLAIR with denoising using deep learning-based reconstruction and the 4-minutes-54-seconds-long reference FLAIR sequence.CONCLUSIONS: Denoising using deep learning-based reconstruction helps to recognize multiple sclerosis lesions buried in the noise of accelerated FLAIR acquisitions, a possibly useful strategy to efficiently shorten the scan time in clinical practice.ABBREVIATIONS: AVD ¼ absolute volume difference; dDLR ¼ denoising using deep learning-based reconstruction; DSC ¼ Dice similarity coefficient; HD ¼ Hausdorff distance; MS ¼ multiple sclerosis; PPV ¼ positive predictive value; TPR ¼ true-positive rate M ultiple sclerosis (MS) is the most common inflammatory disease of the central nervous system affecting young patients, 1 in which demyelination mediated by autoimmune mechanisms is spatially and temporally disseminated. MR imaging plays an essential role not only in the initial diagnosis of MS 2 but also in regular monitoring as a sensitive marker of disease activity for promptly switching therapy if progression is observed. 3 Life-long imaging follow-up is, therefore, required for most patients with MS. A short examination time is necessary to improve the patient's comfort and to cope with the high number of demands in imaging centers.

How far MS lesion detection and segmentation are integrated into the clinical workflow? A systematic review

“… DSC, PPV, spec, TPR, non-parametric Spearman’s correlation coefficient 1st, 2nd ( Chen et al, 2021 ) ISBI 2015 FLAIR, MPRAGE, T2-w, PD-w U-Net b.c., z-score normalisation DSC, PPV, TPR, LFPR, LTPR 1st, 2nd ( Mehta et al, 2021 ) multi-centric 1073 MS T1-w, T2-w, PD-w and FLAIR Bayesian U-Net and another U-Net AUC with TPR and FDR 1st, 2nd ( Opbroek et al, 2014 ) multi-centric 70 MS T1-w, T2-w and FLAIR reduced SVM on 33 features Nyul standardisation relative AVD, average symmetric SD, TPR, FPR 1st, 2nd ( de Oliveira et al, 2022 ) ISBI 2015, MICCAI 2016 FLAIR U-Net anisotropic diffusion filter, normalisation, skull stripping, b.c. DSC, accuracy, precision, sensitivity, specificity 1st, 2nd ( Yamamoto et al, 2022 ) 28 MS FLAIR CNN denoising absolute VD, PPV, TPR, DSC, HD, F1 1st, 2nd ( McKinley et al, 2019 ) multi-centric 139 MS T1-w, T2-w and FLAIR FCNN skull stripping, co-reg. DSC 1st, 2nd ( Zhang et al, 2021b ) multi-centric 200 MS T1-w, T2-w and FLAIR U-Net …”

Deep learning denoising reconstruction enables faster T2‐weighted FLAIR sequence acquisition with satisfactory image quality

“…24 Preliminary work by vendors and studies published to date show no compromise in image quality and therefore diagnostic capability. 5,[7][8][9][10][11][12]16,19,22,25,26 To our knowledge, no prior study has specifically assessed the effect of DLR on 2D adult T2-weighted FLAIR brain images. This study aimed to independently evaluate an accelerated T2-weighted FLAIR sequence that was reconstructed using a commercially available DLR algorithm, by comparing it against the reference standard of our SOC T2-weighted FLAIR sequence.…”

“…There is one prior study which has specifically evaluated the effect of DLR denoising on image quality and lesion conspicuity in a FLAIR sequence, which has been accelerated. 25 The algorithm used in that study was also a commercially available, however from a different vendor with a smaller install-base. Unlike our study, this investigation focussed on a single pathology, multiple sclerosis plaques, and used a 3D instead of a 2D FLAIR sequence.…”

“…DLR techniques, now the dominant form of machine learning, have been recently developed to address this limitation and further accelerate MR image acquisition 24 . Preliminary work by vendors and studies published to date show no compromise in image quality and therefore diagnostic capability 5,7–12,16,19,22,25,26 . To our knowledge, no prior study has specifically assessed the effect of DLR on 2D adult T2‐weighted FLAIR brain images.…”

“…There is one prior study which has specifically evaluated the effect of DLR denoising on image quality and lesion conspicuity in a FLAIR sequence, which has been accelerated 25 . The algorithm used in that study was also a commercially available, however from a different vendor with a smaller install‐base.…”