Novel deep learning-based noise reduction technique for prostate magnetic resonance imaging

Abstract

Introduction Magnetic resonance imaging (MRI) has played an increasingly major role in the evaluation of patients with prostate cancer, although prostate MRI presents several technical challenges. Newer techniques, such as deep learning (DL), have been applied to medical imaging, leading to improvements in image quality. Our goal is to evaluate the performance of a new deep learning-based reconstruction method, “DLR” in improving image quality and mitigating artifacts, which is now commercially available as AIRTM Recon DL (GE Healthcare, Waukesha, WI). We hypothesize that applying DLR to the T2WI images of the prostate provides improved image quality and reduced artifacts. Methods This study included 31 patients with a history of prostate cancer that had a multiparametric MRI of the prostate with an endorectal coil (ERC) at 1.5 T or 3.0 T. Four series of T2-weighted images were generated in total: one set with the ERC signal turned on (ERC) and another set with the ERC signal turned off (Non-ERC). Each of these sets then reconstructed using two different reconstruction methods: conventional reconstruction (Conv) and DL Recon (DLR): ERCDLR, ERCConv, Non-ERCDLR, and Non-ERCConv. Three radiologists independently reviewed and scored the four sets of images for (i) image quality, (ii) artifacts, and (iii) visualization of anatomical landmarks and tumor. Results The Non-ERCDLR scored as the best series for (i) overall image quality (p < 0.001), (ii) reduced artifacts (p < 0.001), and (iii) visualization of anatomical landmarks and tumor. Conclusion Prostate imaging without the use of an endorectal coil could benefit from deep learning reconstruction as demonstrated with T2-weighted imaging MRI evaluations of the prostate.

Fast T2‐Weighted Imaging With Deep Learning‐Based Reconstruction: Evaluation of Image Quality and Diagnostic Performance in Patients Undergoing Radical Prostatectomy

“…There have been two PCa studies with the use of DLR to our knowledge: one demonstrated the feasibility of prostate MRI with reduced acquisition time by 65% (4 minutes 37 seconds for axial T2WI using conventional protocol vs. 1 minute 38 seconds for axial T2WI with short scan time and DLR) in 30 men 22 and the other showed the image quality improvement on standard MRI for 31 men. 23 In our study, although fast MRI with DLR demonstrated better subjective image scores than fast MRI, it showed significantly lower image quality scores than conventional MRI-contrary to the study by Gassenmaier et al showing better image quality scores with the use of fast MRI with DLR. 22 Since image quality can be affected by various acquisition parameters, including parallel imaging acceleration factor, echo train length, TR, echo time, or number of excitations, 29 the use of different parameters may have attributed to those conflicting results with the study by Gassenmaier et al Optimal modification of these technical parameters for the DLR approach may enable image quality improvement of fast MRI, relative to conventional MRI.…”

“…20,21 For PCa, there have been two feasibility studies using DLR on MRI with short acquisition time (i.e., fast MRI); these have demonstrated improvement in overall image quality, reduction of artifacts, and good lesion detectability. 22,23 However, those were predominantly based on subjective image quality analysis in a small number of patients. Furthermore, they did not evaluate the diagnostic performance with regard to evaluation of tumor size or extraprostatic extension (EPE), which may be closely related to image quality.…”

“…It aids noise reduction and image quality improvement in undersampled MRI and may increase diagnostic performance 20,21 . For PCa, there have been two feasibility studies using DLR on MRI with short acquisition time (i.e., fast MRI); these have demonstrated improvement in overall image quality, reduction of artifacts, and good lesion detectability 22,23 . However, those were predominantly based on subjective image quality analysis in a small number of patients.…”

Accelerated T2-Weighted TSE Imaging of the Prostate Using Deep Learning Image Reconstruction: A Prospective Comparison with Standard T2-Weighted TSE Imaging

“…It was shown in recent studies on prostate MRI that DL can be applied for cancer detection and classification and also for registration with histopathological images [10,11,23,24]. Wang et al demonstrated that DL could also be applied for the omission of endorectal coils in mpMRI without compromising the image quality regarding noise [25]. However, one of the most important recent developments regarding DL is related to MRI acquisition times.…”

Deep Learning Applications in Magnetic Resonance Imaging: Has the Future Become Present?

“…Therefore, DL techniques systematically reducing acquisition times might be the key for overcoming shortness of MR scanner capacities and improve healthcare and patient care. DL image reconstruction can not only be used for reduction of acquisition time, but also for improvement of image quality and improvement of patient comfort as it was presented by Wang et al in prostate MRI [ 39 ].…”