Magnetic resonance imaging最新文献

{"title":"Combination of deep learning reconstruction and quantification for dynamic contrast-enhanced (DCE) MRI","authors":"Juntong Jing ,&nbsp;Anthony Mekhanik ,&nbsp;Melanie Schellenberg ,&nbsp;Victor Murray ,&nbsp;Ouri Cohen ,&nbsp;Ricardo Otazo","doi":"10.1016/j.mri.2024.110310","DOIUrl":"10.1016/j.mri.2024.110310","url":null,"abstract":"<div><div>Dynamic contrast-enhanced (DCE) MRI is an important imaging tool for evaluating tumor vascularity that can lead to improved characterization of tumor extent and heterogeneity, and for early assessment of treatment response. However, clinical adoption of quantitative DCE-MRI remains limited due to challenges in acquisition and quantification performance, and lack of automated tools. This study presents an end-to-end deep learning pipeline that exploits a novel deep reconstruction network called DCE-Movienet with a previously developed deep quantification network called DCE-Qnet for fast and quantitative DCE-MRI. DCE-Movienet offers rapid reconstruction of high spatiotemporal resolution 4D MRI data, reducing reconstruction time of the full acquisition to only 0.66 s, which is significantly shorter than compressed sensing's order of 10 min-long reconstructions, without affecting image quality. DCE-Qnet can then perform comprehensive quantification of perfusion parameter maps (K^trans, v_p, v_e), and other parameters affecting quantification (T1, B1, and BAT) from a single contrast-enhanced acquisition. The end-to-end deep learning pipeline was implemented to process data acquired with a golden-angle stack-of-stars k-space trajectory and validated on healthy volunteers and a cervical cancer patient against a compressed sensing reconstruction. The end-to-end deep learning DCE-MRI technique addresses key limitations in DCE-MRI in terms of speed and quantification robustness, which is expected to improve the performance of DCE-MRI in a clinical setting.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110310"},"PeriodicalIF":2.1,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142877514","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Reliability of post-contrast deep learning-based highly accelerated cardiac cine MRI for the assessment of ventricular function","authors":"Makoto Orii ,&nbsp;Momoko Sugawara ,&nbsp;Tsuyoshi Sugawara ,&nbsp;Kunihiro Yoshioka","doi":"10.1016/j.mri.2024.110313","DOIUrl":"10.1016/j.mri.2024.110313","url":null,"abstract":"<div><h3>Objective</h3><div>The total examination time can be reduced if high-quality two-dimensional (2D) cine images can be collected post-contrast to minimize non-scanning time prior to late gadolinium-enhanced imaging. This study aimed to assess the equivalency of the pre-and post-contrast performance of 2D deep learning-based highly accelerated cardiac cine (DL cine) imaging by evaluating the image quality and the quantification of biventricular volumes and function in the clinical setting.</div></div><div><h3>Material and methods</h3><div>Thirty patients (20 men, mean age 53.7 ± 17.8 years) underwent cardiac magnetic resonance on a 1.5 T scanner for clinical indications, and pre- and post-contrast DL cine images were acquired with a short-axis view. Image-quality was scored according to three main criteria: the blood-to-myocardial contrast, endocardial edge delineation, and presence of motion artifacts throughout the cardiac cycle.</div><div>Biventricular end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV), ejection fraction (EF), and left ventricular mass (LVM) were analyzed and compared between the pre- and post-contrast DL cine images.</div></div><div><h3>Results</h3><div>The actual median time of 2D DL cine acquisition was 38.4 ± 9.1 s. There were no significant differences in the image quality scores between pre- and post-contrast DL cine images (<em>p</em> &gt; 0.05). In the volume and functional analysis, there was no significant difference in terms of biventricular EDV, ESV, SV, EF, and LVM (<em>p</em> &gt; 0.05).</div></div><div><h3>Conclusions</h3><div>The performance of 2D DL cine is equivalent before and after contrast injection for the assessment of image quality and ventricular function in the clinical setting.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110313"},"PeriodicalIF":2.1,"publicationDate":"2024-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142872533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning radiomics nomograms predict Isocitrate dehydrogenase (IDH) genotypes in brain glioma: A multicenter study","authors":"Darui Li ,&nbsp;Wanjun Hu ,&nbsp;Laiyang Ma ,&nbsp;Wenxia Yang ,&nbsp;Yang Liu ,&nbsp;Jie Zou ,&nbsp;Xin Ge ,&nbsp;Yuping Han ,&nbsp;Tiejun Gan ,&nbsp;Dan Cheng ,&nbsp;Kai Ai ,&nbsp;Guangyao Liu ,&nbsp;Jing Zhang","doi":"10.1016/j.mri.2024.110314","DOIUrl":"10.1016/j.mri.2024.110314","url":null,"abstract":"<div><h3>Purpose</h3><div>To explore the feasibility of Deep learning radiomics nomograms (DLRN) in predicting IDH genotype.</div></div><div><h3>Methods</h3><div>A total of 402 glioma patients from two independent centers were retrospectively included, and the data from center I was randomly divided into a training cohort (<em>n</em> = 239) and an internal validation cohort (<em>n</em> = 103) on a 7:3 basis. Center II served as an independent external validation cohort (<em>n</em> = 60). We developed a DLRN for IDH classification of gliomas based on T2 images. This hybrid model integrates deep learning features, radiomics features, and clinical features most relevant to IDH genotypes and finally classifies them using multivariate logistic regression analysis. We used the area under the curve (AUC) of the receiver operating characteristic (ROC) to evaluate the performance of the model and applied the DLRN score to the survival analysis of some of the follow-up glioma patients.</div></div><div><h3>Results</h3><div>The proposed model had an area under the curve (AUC) of 0.98 in an externally validated cohort, and DLRN scores were significantly associated with the overall survival of glioma patients.</div></div><div><h3>Conclusions</h3><div>Deep learning radiomics nomograms performed well in non-invasively predicting IDH mutation status in gliomas, assisting stratified management and targeted therapy for glioma patients.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110314"},"PeriodicalIF":2.1,"publicationDate":"2024-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142872530","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multiple b value diffusion-weighted MRI of liver: A novel respiratory frequency-modulated continuous-wave radar-trigger technique and comparison with free-breathing technique","authors":"Kai Liu ,&nbsp;Caizhong Chen ,&nbsp;Tingting Shen ,&nbsp;Xixi Wen ,&nbsp;Mengsu Zeng ,&nbsp;Pengju Xu","doi":"10.1016/j.mri.2024.110312","DOIUrl":"10.1016/j.mri.2024.110312","url":null,"abstract":"<div><h3>Objective</h3><div>The aim of this study was to evaluate a novel respiratory frequency-modulated continuous-wave radar-trigger (FT) technique for multiple -b-value diffusion-weighted imaging (DWI) of liver and compare it with conventional free breathing (FB) DWI technique.</div></div><div><h3>Material and methods</h3><div>39 patients with focal liver lesions underwent both frequency-modulated continuous-wave radar-trigger (FT) and conventional free-breathing (FB) multi-b-value diffusion-weighted imaging (DWI,b = 0,50,400,800 s/mm²⁾. Two abdominal radiologists independently assessed the quality of liver DWI images obtained using both techniques, measured and compared liver signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) at different b-values, as well as apparent diffusion coefficient (ADC) values calculated from all b-values.</div></div><div><h3>Results</h3><div>In terms of image quality, the FT technique is superior to the conventional FB technique, with overall image quality scores (Reader 1, 3.56 ± 0.50 and Reader 2, 3.59 ± 0.55)vs (Reader 1, 2.90 ± 0.75 and Reader 2, 2.97 ± 0.71), respectively. The liver SNR (at b-values of 50,400,and 800 s/mm²) obtained by FT was (138.5 ± 43.48, 96.67 ± 31.95, 71.54 ± 22.03), respectively, which was significantly higher than that obtained by conventional FB (110.90 ± 39.28, 80.86 ± 29.13, 60.43 ± 18.61, <em>P</em> &lt; 0.05). The lesion CNR with FT was significantly higher than that with conventional FB (258.99 ± 151.38 vs 174.60 ± 99.90; 164.56 ± 87.25 vs 111.12 ± 42.43; 118.83 ± 68.76 vs 76.01 ± 35.48, <em>P</em> &lt; 0.001). There was no significant difference in ADC values of liver and lesions between the two techniques: ADCliver-L and ADCliver-R: (FT 1479.3 ± 270.0 vs FB 1529.3 ± 275.5 and FT 1219.6 ± 127.4 vs FB 1248.7 ± 168.2, <em>P</em> &gt; 0.05); ADC lesion:FT(969.0 ± 261.3) vs FB (1017.5 ± 240.4, <em>P</em> &gt; 0.05).</div></div><div><h3>Conclusion</h3><div>For multi-b-value liver diffusion-weighted imaging, FT technique has higher image quality and better lesion visibility than conventional FB technique and there is no significant difference in ADC values of liver and lesions between the two techniques.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110312"},"PeriodicalIF":2.1,"publicationDate":"2024-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142854345","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Segmental myocardial tissue remodeling and atrial arrhythmias in hypertrophic cardiomyopathy: Findings from T1-mapping MRI","authors":"Danqing Liu ,&nbsp;Hong Luo ,&nbsp;Changjing Feng ,&nbsp;Yufei Lian ,&nbsp;Zhenyu Pan ,&nbsp;Xiaojuan Guo ,&nbsp;Qi Yang","doi":"10.1016/j.mri.2024.110311","DOIUrl":"10.1016/j.mri.2024.110311","url":null,"abstract":"<div><h3>Background</h3><div>Myocardial fibrosis of the left ventricle (LV) has been associated with atrial fibrillation and other arrhythmias in individuals with hypertrophic cardiomyopathy (HCM). However, few studies have quantitatively examined the segmental relationship between diffuse LV fibrosis and atrial arrhythmias in HCM using T1 mapping and extracellular volume fraction (ECV). The aim of this study is to explore this relationship through T1 mapping, offering imaging insights into the pathophysiology of HCM with atrial arrhythmia.</div></div><div><h3>Methods</h3><div>A total of 38 patients with HCM were classified into two groups—those with atrial arrhythmia and those without—based on electrocardiographic and Holter monitor recordings. A covariance analysis was conducted to compare T1 mapping parameters between the two groups, adjusting for wall thickness (WT) as a covariate. Analysis was performed collectively for all 16 myocardial segments, as well as for each segment individually.</div></div><div><h3>Results</h3><div>Native T1 values were elevated in the entire LV myocardium and in segments S1–3 in patients with HCM with atrial arrhythmias compared to those without (<em>P</em> &lt; 0.001; <em>P</em> &lt; 0.05, 1316.0 ms ± 15.9 vs 1263.1 ms ± 13.6, 1350.5 ms ± 14.2 vs 1311.9 ms ± 11.7, 1305.7 ms ± 2.5 vs 1271.5 ms ± 10.6, respectively). Notably, the basal anterior segment (S1) and basal inferotseptal segment (S3) exhibited prolonged ECV and elevated native T1 values in patients with HCM and atrial arrhythmia (<em>P</em> &lt; 0.05). Multivariable binary logistic regression analysis identified myocardial native T1 values in the basal anteroseptal segment (S2) as a predictor of atrial arrhythmia presence in HCM, with values exceeding 1350 ms correlating with an increased likelihood of arrhythmia development. No significant difference in WT was observed between the groups in hypertrophic myocardial regions (<em>P</em> &gt; 0.05), while non-hypertrophic myocardium in individuals with HCM and atrial arrhythmias exhibited reduced wall thickness (7.7 mm ± 3.0 vs 9 mm ± 3.0, <em>P</em> &lt; 0.001) compared to those without arrhythmias.</div></div><div><h3>Conclusion</h3><div>Fibrosis in the basal septal and anterior regions of the left ventricle plays a crucial role in myocardial tissue remodeling, contributing to the development of atrial arrhythmia in HCM. Elevated native T1 values in the basal anteroseptal segment may may serve as a significant marker for the concurrent occurrence of atrial arrhythmias in individuals with HCM.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110311"},"PeriodicalIF":2.1,"publicationDate":"2024-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142847024","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting molecular subtypes of breast cancer based on multi-parametric MRI dataset using deep learning method","authors":"Wanqing Ren ,&nbsp;Xiaoming Xi ,&nbsp;Xiaodong Zhang ,&nbsp;Kesong Wang ,&nbsp;Menghan Liu ,&nbsp;Dawei Wang ,&nbsp;Yanan Du ,&nbsp;Jingxiang Sun ,&nbsp;Guang Zhang","doi":"10.1016/j.mri.2024.110305","DOIUrl":"10.1016/j.mri.2024.110305","url":null,"abstract":"<div><h3>Purpose</h3><div>To develop a multi-parametric MRI model for the prediction of molecular subtypes of breast cancer using five types of breast cancer preoperative MRI images.</div></div><div><h3>Methods</h3><div>In this study, we retrospectively analyzed clinical data and five types of MRI images (FS-T1WI, T2WI, Contrast-enhanced T1-weighted imaging (T1-C), DWI, and ADC) from 325 patients with pathologically confirmed breast cancer. Using the five types of MRI images as inputs to the ResNeXt50 model respectively, five base models were constructed, and then the outputs of the five base models were fused using an ensemble learning approach to develop a multi-parametric MRI model. Breast cancer was classified into four molecular subtypes based on immunohistochemical results: luminal A, luminal B, human epidermal growth factor receptor 2-positive (HER2-positive), and triple-negative (TN). The whole dataset was randomly divided into a training set (<em>n</em> = 260; 76 luminal A, 80 luminal B, 50 HER2-positive, 54 TN) and a testing set (<em>n</em> = 65; 20 luminal A, 20 luminal B, 12 HER2-positive, 13 TN). Accuracy, sensitivity, specificity, receiver operating characteristic curve (ROC) and area under the curve (AUC) were calculated to assess the predictive performance of the models.</div></div><div><h3>Results</h3><div>In the testing set, for the assessment of the four molecular subtypes of breast cancer, the multi-parametric MRI model yielded an AUC of 0.859–0.912; the AUCs based on the FS-T1WI, T2WI, T1-C, DWI, and ADC models achieved respectively 0.632–0. 814, 0.641–0.788, 0.621–0.709, 0.620–0.701and 0.611–0.785.</div></div><div><h3>Conclusion</h3><div>The multi-parametric MRI model we developed outperformed the base models in predicting breast cancer molecular subtypes. Our study also showed the potential of FS-T1WI base model in predicting breast cancer molecular subtypes.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110305"},"PeriodicalIF":2.1,"publicationDate":"2024-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142837200","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Conditional generative diffusion deep learning for accelerated diffusion tensor and kurtosis imaging","authors":"Phillip Martin ,&nbsp;Maria Altbach ,&nbsp;Ali Bilgin","doi":"10.1016/j.mri.2024.110309","DOIUrl":"10.1016/j.mri.2024.110309","url":null,"abstract":"<div><h3>Purpose</h3><div>The purpose of this study was to develop DiffDL, a generative diffusion probabilistic model designed to produce high-quality diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) metrics from a reduced set of diffusion-weighted images (DWIs). This model addresses the challenge of prolonged data acquisition times in diffusion MRI while preserving metric accuracy.</div></div><div><h3>Methods</h3><div>DiffDL was trained using data from the Human Connectome Project, including 300 training/validation subjects and 50 testing subjects. High-quality DTI and DKI metrics were generated using many DWIs and combined with subsets of DWIs to form training pairs. A UNet architecture was used for denoising, trained over 500 epochs with a linear noise schedule. Performance was evaluated against conventional DTI/DKI modeling and a reference UNet model using normalized mean absolute error (NMAE), peak signal-to-noise ratio (PSNR), and Pearson correlation coefficient (PCC).</div></div><div><h3>Results</h3><div>DiffDL showed significant improvements in the quality and accuracy of fractional anisotropy (FA) and mean diffusivity (MD) maps compared to conventional methods and the baseline UNet model. For DKI metrics, DiffDL outperformed conventional DKI modeling and the UNet model across various acceleration scenarios. Quantitative analysis demonstrated superior NMAE, PSNR, and PCC values for DiffDL, capturing the full dynamic range of DTI and DKI metrics. The generative nature of DiffDL allowed for multiple predictions, enabling uncertainty quantification and enhancing performance.</div></div><div><h3>Conclusion</h3><div>The DiffDL framework demonstrated the potential to significantly reduce data acquisition times in diffusion MRI while maintaining high metric quality. Future research should focus on optimizing computational demands and validating the model with clinical cohorts and standard MRI scanners.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110309"},"PeriodicalIF":2.1,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142828896","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Longitudinal DTI analysis of microstructural changes in lumbar nerve roots following Interspinous process device placement","authors":"L. Monti ,&nbsp;M. Bellini MD ,&nbsp;M. Alberti ,&nbsp;E. Piane ,&nbsp;T. Casseri ,&nbsp;G. Sadotti ,&nbsp;S. Marcia ,&nbsp;J.A. Hirsc ,&nbsp;F. Ginanneschi ,&nbsp;A. Rossi","doi":"10.1016/j.mri.2024.110306","DOIUrl":"10.1016/j.mri.2024.110306","url":null,"abstract":"<div><div>Diffusion tensor imaging (DTI) and its parameters such as fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), radial diffusivity (RD) are increasingly being used to assess peripheral nerve integrity alongside nerve conduction studies. This pilot study aims to compare DTI values of lumbar spinal nerve roots before (T0) and after (T1) treatment with an interspinous process device (IPD). Seven patients (5 females, 2 males; mean age: 68) suffering from neurogenic claudication and lumbar spinal canal and foraminal stenosis were evaluated. Visual Analog Scale (VAS) for perceived pain, Oswestry Disability Index (ODI), and DTI parameters were assessed between T0 and T1. No significant difference in FA was found in treated roots, while MD (<em>p</em> = 0.0015), RD (<em>p</em> = 0.0032), and AD (<em>p</em> = 0.0221) were significantly altered. At untreated levels, all DTI parameters showed highly significant differences (<em>p</em> &lt; 0.0001) between T0 and T1. In treated roots, FA values significantly increased in the intraforaminal segment(<em>p</em> = 0.0229), while MD(<em>p</em> = 0.0124), AD(<em>p</em> = 0.0128), and RD (<em>p</em> = 0.0143) values decreased in the pre-foraminal segment. In untreated roots, FA significantly increased in pre(<em>p</em> = 0.0039)and intraforaminal(<em>p</em> = 0.0003) segments, and MD, AD, and RD decreased in all segments (p &lt; 0.0001). VAS (p &lt; 0.0001) also decreased between T0 and T1. This pilot study aims to clarify the biomechanical impact of interspinous spacers through microstructural analysis of both treated and adjacent untreated nerve roots. To our knowledge, no studies have examined the short- to medium-term changes in DTI values of lumbar nerve roots before and after IPD placement, or compared changes between treated and untreated roots.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110306"},"PeriodicalIF":2.1,"publicationDate":"2024-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142822227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A lightweight adaptive spatial channel attention efficient net B3 based generative adversarial network approach for MR image reconstruction from under sampled data","authors":"Penta Anil Kumar,&nbsp;Ramalingam Gunasundari","doi":"10.1016/j.mri.2024.110281","DOIUrl":"10.1016/j.mri.2024.110281","url":null,"abstract":"<div><div>Magnetic Resonance Imaging (MRI) stands out as a notable non-invasive method for medical imaging assessments, widely employed in early medical diagnoses due to its exceptional resolution in portraying soft tissue structures. However, the MRI method faces challenges with its inherently slow acquisition process, stemming from the sequential sampling in k-space and limitations in traversal speed due to physiological and hardware constraints. Compressed Sensing in MRI (CS-MRI) accelerates image acquisition by utilizing greatly under-sampled k-space information. Despite its advantages, conventional CS-MRI encounters issues such as sluggish iterations and artefacts at higher acceleration factors. Recent advancements integrate deep learning models into CS-MRI, inspired by successes in various computer vision domains. It has drawn significant attention from the MRI community because of its great potential for image reconstruction from undersampled k-space data in fast MRI. This paper proposes a lightweight Adaptive Spatial-Channel Attention EfficientNet B3-based Generative Adversarial Network (ASCA-EffNet GAN) for fast, high-quality MR image reconstruction from greatly under-sampled k-space information in CS-MRI. The proposed GAN employs a U-net generator with ASCA-based EfficientNet B3 for encoder blocks and a ResNet decoder. The discriminator is a binary classifier with ASCA-based EfficientNet B3, a fully connected layer and a sigmoid layer. The EfficientNet B3 utilizes a compound scaling strategy that achieves a balance amongst model depth, width, and resolution, resulting in optimal performance with a reduced number of parameters. Furthermore, the adaptive attention mechanisms in the proposed ASCA-EffNet GAN effectively capture spatial and channel-wise features, contributing to detailed anatomical structure reconstruction. Experimental evaluations on the dataset demonstrate ASCA-EffNet GAN's superior performance across various metrics, surpassing conventional reconstruction methods. Hence, ASCA-EffNet GAN showcases remarkable reconstruction capabilities even under high under-sampling rates, making it suitable for clinical applications.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110281"},"PeriodicalIF":2.1,"publicationDate":"2024-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142822224","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing thin slice 3D T2-weighted prostate MRI with super-resolution deep learning reconstruction: Impact on image quality and PI-RADS assessment","authors":"Kaori Shiraishi ,&nbsp;Takeshi Nakaura ,&nbsp;Naoki Kobayashi ,&nbsp;Hiroyuki Uetani ,&nbsp;Yasunori Nagayama ,&nbsp;Masafumi Kidoh ,&nbsp;Junji Yatsuda ,&nbsp;Ryoma Kurahashi ,&nbsp;Tomomi Kamba ,&nbsp;Yuichi Yamahita ,&nbsp;Toshinori Hirai","doi":"10.1016/j.mri.2024.110308","DOIUrl":"10.1016/j.mri.2024.110308","url":null,"abstract":"<div><h3>Purposes</h3><div>This study aimed to assess the effectiveness of Super-Resolution Deep Learning Reconstruction (SR-DLR) —a deep learning-based technique that enhances image resolution and quality during MRI reconstruction— in improving the image quality of thin-slice 3D T2-weighted imaging (T2WI) and Prostate Imaging-Reporting and Data System (PI-RADS) assessment in prostate Magnetic Resonance Imaging (MRI).</div></div><div><h3>Methods</h3><div>This retrospective study included 33 patients who underwent prostate MRI with SR-DLR between November 2022 and April 2023. Thin-slice 3D-T2WI of the prostate was obtained and reconstructed with and without SR-DLR (matrix: 720 × 720 and 240 × 240, respectively). We calculated the contrast and contrast-to-noise ratio (CNR) between the internal and external glands of the prostate, as well as the slope of pelvic bone and adipose tissue. Two radiologists evaluated qualitative image quality and assessed PI-RADS scores of each reconstruction.</div></div><div><h3>Results</h3><div>The final analysis included 28 male patients (age range: 47–88 years; mean age: 70.8 years). The CNR with SR-DLR was significantly higher than without SR-DLR (1.93 [IQR: 0.79, 3.83] vs. 1.88 [IQR: 0.63, 3.82], <em>p</em> = 0.002). No significant difference in contrast was observed between images with and without SR-DLR (<em>p</em> = 0.864). The slope with SR-DLR was significantly higher than without SR-DLR (0.21 [IQR: 0.15, 0.25] vs. 0.15 [IQR: 0.12, 0.19], <em>p</em> &lt; 0.01). Qualitative scores for contrast, sharpness, artifacts, and overall image quality were significantly higher with SR-DLR than without SR-DLR (<em>p</em> &lt; 0.05 for all). The kappa values for 2D-T2WI and 3D-T2WI increased from 0.694 and 0.640 to 0.870 and 0.827 with SR-DLR for both readers.</div></div><div><h3>Conclusions</h3><div>SR-DLR has the potential to improve image quality and the ability to assess PI-RADS scores in thin-slice 3D-T2WI of the prostate without extending MRI acquisition time.</div></div><div><h3>Summary</h3><div>Super-Resolution Deep Learning Reconstruction (SR-DLR) significantly improved image quality of thin-slice 3D T2-weighted imaging (T2WI) without extending the acquisition time. Additionally, the PI-RADS scores from 3D-T2WI with SR-DLR demonstrated higher agreement with those from 2D-T2WI.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"Article 110308"},"PeriodicalIF":2.1,"publicationDate":"2024-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142818554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}