Diagnostic advantage of thin slice 2D MRI and multiplanar reconstruction of the knee joint using deep learning based denoising approach

Abstract

The purpose of this study is to evaluate whether thin-slice high-resolution 2D fat-suppressed proton density-weighted image of the knee joint using denoising approach with deep learning-based reconstruction (dDLR) with MPR is more useful than 3D FS-PD multi planar voxel image. Twelve patients who underwent MRI of the knee at 3T and 13 knees were enrolled. Denoising effect was quantitatively evaluated by comparing the coefficient of variation (CV) before and after dDLR. For the qualitative assessment, two radiologists evaluated image quality, artifacts, anatomical structures, and abnormal findings using a 5-point Likert scale between 2D and 3D. All of them were statistically analyzed. Gwet’s agreement coefficients were also calculated. For the scores of abnormal findings, we calculated the percentages of the cases with agreement with high confidence. The CV after dDLR was significantly lower than the one before dDLR (p < 0.05). As for image quality, artifacts and anatomical structure, no significant differences were found except for flow artifact (p < 0.05). The agreement was significantly higher in 2D than in 3D in abnormal findings (p < 0.05). In abnormal findings, the percentage with high confidence was higher in 2D than in 3D (p < 0.05). By applying dDLR to 2D, almost equivalent image quality to 3D could be obtained. Furthermore, abnormal findings could be depicted with greater confidence and consistency, indicating that 2D with dDLR can be a promising imaging method for the knee joint disease evaluation.

Deep learning‐based reconstruction can improve canine thoracolumbar magnetic resonance image quality and reduce slice thickness

“…15 Several human studies have used DLR with slice thickness reduction to evaluate small structures; their results have shown enhanced diagnostic accuracy and improved structural conspicuity. 6,9,17,24 In addition, DLR improves the quantitative SNR and CNR and visual quality of MR images, thus improving clinical diagnosis and management. 6,9,15,25 The results of the present study were consistent with those of previous studies; further compared with conventional 3-mm MR images, DLR-applied MR images exhibited higher or similar SNR and CNR despite slice thickness reduction to 1 mm.…”

“…In few studies, DLR was applied to evaluate small structures such as the pituitary gland, peripheral nerves, and knee joint, 6,9,14,17,18,24 and it demonstrated similar or better diagnostic performance compared with the conventional MR images. In the present study, for identifying nerve roots, the diagnostic performances of thin-slice MRI were higher than those of 3-mm MRI; this finding was similar to those of previous studies.…”

“…6,9,14,26 Nevertheless, the effective denoising performance of DLR compensates for the quality of regional signals despite slice thickness reduction. 6,9,14,17 This implies that the application of DLR in thoracolumbar MRI is effective and that the trade-off factors do not necessarily need to be modified for improving SNR or spatial resolution. Therefore, thin-slice MRI with DLR is particularly useful for the evaluation of small anatomical structures in small breed dogs.…”

“…This is similar to a previous human study where DLR was applied on thin-slice MRI scans to evaluate small anatomical structures such as the pituitary gland, peripheral nerve, and knee joints. 6,9,14,17,18 This suggests that DLR could be a promising technique for improving the visualization of small structures with enhanced image quality.…”

“…6,14,15 In human medicine, owing to the higher diagnostic performance and better image quality obtained by the application of DLR to thin-slice MR images, this method has been used in the diagnosis of a pituitary tumor and knee joint disease. 6,9,14,[16][17][18] However, based on our review of the literature, no study has described the use of DLR in veterinary practice. Previous human studies have reported that thin-slice MRI can be obtained while maintaining image quality by applying DLR.…”