Pub Date : 2025-12-11DOI: 10.1016/j.mri.2025.110593
Di Wu, Zhaobing Tang
Background
Diagnosis of prostate cancer in the PSA gray zone (4–10 ng/mL) and PI-RADS 3 cases remains challenging. Although multiparametric MRI (mpMRI) is widely used, its diagnostic accuracy is limited by inter-reader variability and the lack of integration with clinical indicators. Prostate-specific antigen density (PSAD) is a valuable risk stratifier, but its optimal combination with mpMRI remains unclear.
Methods
We developed a deep-learning model that integrates PSAD with mpMRI—including T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC) maps derived from DWI. A cross-modal attention-guided (CM-AG) fusion module weights the PSAD and mpMRI feature branches. Performance was assessed in the PSA gray zone and the PI-RADS 3 subgroup. Ablation experiments quantified the incremental contributions of PSAD and CM-AG.
Results
The model achieved AUC = 0.89 in the PSA gray-zone cohort and AUC = 0.83 in PI-RADS 3, outperforming single-modality MRI baselines and PI-RADS–based assessment alone (DeLong p < 0.01). In patients with larger prostate volumes, specificity increased by 10.2 %. Ablation results confirmed that both PSAD and CM-AG contributed materially to performance gains.
Conclusion
Fusing PSAD with mpMRI via cross-modal attention improves diagnostic performance, particularly in challenging subgroups (PSA gray zone, PI-RADS 3). This approach may support more consistent risk stratification and earlier detection.
{"title":"Application value of prostate-specific antigen density combined with multiparametric MRI in early diagnosis of prostate cancer","authors":"Di Wu, Zhaobing Tang","doi":"10.1016/j.mri.2025.110593","DOIUrl":"10.1016/j.mri.2025.110593","url":null,"abstract":"<div><h3>Background</h3><div>Diagnosis of prostate cancer in the PSA gray zone (4–10 ng/mL) and PI-RADS 3 cases remains challenging. Although multiparametric MRI (mpMRI) is widely used, its diagnostic accuracy is limited by inter-reader variability and the lack of integration with clinical indicators. Prostate-specific antigen density (PSAD) is a valuable risk stratifier, but its optimal combination with mpMRI remains unclear.</div></div><div><h3>Methods</h3><div>We developed a deep-learning model that integrates PSAD with mpMRI—including T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC) maps derived from DWI. A cross-modal attention-guided (CM-AG) fusion module weights the PSAD and mpMRI feature branches. Performance was assessed in the PSA gray zone and the PI-RADS 3 subgroup. Ablation experiments quantified the incremental contributions of PSAD and CM-AG.</div></div><div><h3>Results</h3><div>The model achieved AUC = 0.89 in the PSA gray-zone cohort and AUC = 0.83 in PI-RADS 3, outperforming single-modality MRI baselines and PI-RADS–based assessment alone (DeLong <em>p</em> < 0.01). In patients with larger prostate volumes, specificity increased by 10.2 %. Ablation results confirmed that both PSAD and CM-AG contributed materially to performance gains.</div></div><div><h3>Conclusion</h3><div>Fusing PSAD with mpMRI via cross-modal attention improves diagnostic performance, particularly in challenging subgroups (PSA gray zone, PI-RADS 3). This approach may support more consistent risk stratification and earlier detection.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110593"},"PeriodicalIF":2.0,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145751972","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}
Pub Date : 2025-12-11DOI: 10.1016/j.mri.2025.110592
Neil Abraham Barnes , Winniecia Dkhar , Rajagopal Kadavigere , Suresh Sukumar , K. Vaishali , Abhimanyu Pradhan , P.S. Priya , Ashwin Prabhu , Nikhil Raj
Background
Degenerative spinal disease is a leading cause of lower back pain worldwide, impairing mobility and quality of life. Early degeneration is increasingly seen in younger adults due to sedentary lifestyles and occupational stress. Conventional MRI, particularly T2-weighted imaging, is limited in detecting early microstructural changes. Diffusion kurtosis imaging (DKI), by quantifying non-Gaussian water diffusion, offers enhanced sensitivity for identifying subtle alterations preceding overt degeneration.
Objective
To evaluate the ability of quantitative DKI parameters to detect early intervertebral disc degeneration by correlating them with the Pfirrmann grading system in the thoracolumbar spine.
Methods
This prospective study included 76 participants, 54 with degenerative spine disease and 22 healthy controls. MRI was performed on a 3 T scanner using sagittal T1-, T2-, and DKI sequences. Parameter maps included mean kurtosis (MK), axial kurtosis (AK), radial kurtosis (RK), fractional anisotropy (FA), and diffusivity indices (MD, AD, RD, KA). Regions of interest were placed in the nucleus pulposus and annulus fibrosus, with subregional analysis of the anterior and posterior annulus. Grade V discs were excluded, yielding (1,287) discs (Grades I–IV). Statistical analyses included group comparisons and Spearman's correlation.
Results
MK, AK, and RK were significantly higher in the degenerative group (p < 0.001), while MD, AD, and RD were substantially lower (p < 0.001). Correlations were strongest in the mid-to-lower lumbar levels, particularly within the nucleus pulposus and anterior annulus.
Conclusion
DKI enables quantitative characterisation of disc microstructure and demonstrates diagnostic potential for differentiating degenerative from non-degenerative discs, supporting its role as an emerging imaging biomarker for early degeneration assessment.
{"title":"Quantitative assessment of early intervertebral disc degeneration with MR diffusion kurtosis imaging: A radiologic correlation with Pfirrmann grade","authors":"Neil Abraham Barnes , Winniecia Dkhar , Rajagopal Kadavigere , Suresh Sukumar , K. Vaishali , Abhimanyu Pradhan , P.S. Priya , Ashwin Prabhu , Nikhil Raj","doi":"10.1016/j.mri.2025.110592","DOIUrl":"10.1016/j.mri.2025.110592","url":null,"abstract":"<div><h3>Background</h3><div>Degenerative spinal disease is a leading cause of lower back pain worldwide, impairing mobility and quality of life. Early degeneration is increasingly seen in younger adults due to sedentary lifestyles and occupational stress. Conventional MRI, particularly T2-weighted imaging, is limited in detecting early microstructural changes. Diffusion kurtosis imaging (DKI), by quantifying non-Gaussian water diffusion, offers enhanced sensitivity for identifying subtle alterations preceding overt degeneration<strong>.</strong></div></div><div><h3>Objective</h3><div>To evaluate the ability of quantitative DKI parameters to detect early intervertebral disc degeneration by correlating them with the Pfirrmann grading system in the thoracolumbar spine.</div></div><div><h3>Methods</h3><div>This prospective study included 76 participants, 54 with degenerative spine disease and 22 healthy controls. MRI was performed on a 3 T scanner using sagittal T1-, T2-, and DKI sequences. Parameter maps included mean kurtosis (MK), axial kurtosis (AK), radial kurtosis (RK), fractional anisotropy (FA), and diffusivity indices (MD, AD, RD, KA). Regions of interest were placed in the nucleus pulposus and annulus fibrosus, with subregional analysis of the anterior and posterior annulus. Grade V discs were excluded, yielding (1,287) discs (Grades I–IV). Statistical analyses included group comparisons and Spearman's correlation.</div></div><div><h3>Results</h3><div>MK, AK, and RK were significantly higher in the degenerative group (<em>p</em> < 0.001), while MD, AD, and RD were substantially lower (p < 0.001). Correlations were strongest in the mid-to-lower lumbar levels, particularly within the nucleus pulposus and anterior annulus<strong>.</strong></div></div><div><h3>Conclusion</h3><div>DKI enables quantitative characterisation of disc microstructure and demonstrates diagnostic potential for differentiating degenerative from non-degenerative discs, supporting its role as an emerging imaging biomarker for early degeneration assessment.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110592"},"PeriodicalIF":2.0,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145751912","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}
Pub Date : 2025-12-09DOI: 10.1016/j.mri.2025.110591
Chenhao Gao, Fan Yang, Zhihao Xue, Junyao Zhang, Zhuo Chen, Sirui Huo, Juan Gao, Yixin Emu, Haiyang Chen, Chenxi Hu
Free-running self-gated 3D cardiac cine imaging is highly desirable for volumetric, high-resolution, breath-hold-free assessment of left ventricular (LV) function. However, its implementation at 3 T remains challenging due to specific absorption rate (SAR) constraints and reduced myocardium-blood contrast. In this study, we propose a novel non-contrast, free-running, self-gated 3D gradient-echo (GRE) cine sequence for 3 T imaging, which acquires multi-slab data using a pseudo-radial Cartesian trajectory with a 1.5 mm slice thickness. To address respiratory motion, a locally low-rank motion-corrected image reconstruction algorithm was developed. Fifteen participants underwent imaging with the proposed multi-slab 3D cine sequence and conventional 2D cine sequences. Additionally, single-slab 3D cine data were acquired in 10 participants. Various image quality metrics (signal-to-noise ratio, contrast-to-noise ratio, myocardial sharpness, and residual artefact) and LV volumetric parameters (end-diastolic volume (EDV), end-systolic volume (ESV), and ejection fraction (EF)) were compared between the different methods. Results demonstrated that the proposed multi-slab 3D cine method provided significantly superior image quality compared to the single-slab 3D approach (myocardial sharpness: 2.82 ± 0.42 vs. 1.58 ± 0.46, P = 0.005; residual artefact: 2.58 ± 0.26 vs. 1.13 ± 0.21, P = 0.005) due to the improvement of image contrast. Furthermore, the multi-slab 3D cine sequence exhibited good agreement and correlation with the reference 2D cine method in terms of volumetric measures (EDV: 140.3 ± 19.9 mL vs. 139.3 ± 20.0 mL, P = 0.357, r = 0.976; ESV: 56.7 ± 11.8 mL vs. 57.7 ± 10.2 mL, P = 0.259, r = 0.964; EF: 59.8 % ± 4.9 % vs. 58.7 % ± 3.7 %, P = 0.073, r = 0.907). In conclusion, the proposed multi-slab 3D cine framework enables free-running 3 T cine imaging with whole-heart coverage and high through-plane resolution. Although myocardium-blood contrast is reduced compared to 2D breath-hold cine, the retained contrast is sufficient to evaluate LV function.
自由运行的自门控3D心脏电影成像是非常理想的容积，高分辨率，无呼吸的左心室（LV）功能评估。然而，由于特定吸收率（SAR）的限制和心肌-血液对比降低，在3t时的实施仍然具有挑战性。在这项研究中，我们提出了一种新的无对比度、自由运行、自门控的3D梯度回波（GRE）序列，用于3t成像，该序列使用1.5 mm层厚的伪径向笛卡尔轨迹获取多层数据。针对呼吸运动，提出了一种局部低秩运动校正图像重建算法。15名参与者接受了拟议的多板3D电影序列和传统的2D电影序列的成像。此外，还获得了10名参与者的单平板三维电影数据。各种图像质量指标（信噪比、对比噪声比、心肌清晰度和残余伪影）和左室容积参数（舒张末期容积（EDV）、收缩末期容积（ESV）和射血分数（EF））在不同方法之间进行了比较。结果表明，由于图像对比度的提高，多板三维成像方法的图像质量明显优于单板三维成像方法（心肌清晰度：2.82±0.42 vs. 1.58±0.46,P = 0.005；残余伪影：2.58±0.26 vs. 1.13±0.21,P = 0.005）。此外，在容积测量方面，多板三维电影序列与参考2D电影方法表现出良好的一致性和相关性（EDV: 140.3±19.9 mL对139.3±20.0 mL, P = 0.357, r = 0.976; ESV: 56.7±11.8 mL对57.7±10.2 mL, P = 0.259, r = 0.964; EF: 59.8%±4.9%对58.7%±3.7%,P = 0.073, r = 0.907）。总之，所提出的多板3D电影框架可以实现全心脏覆盖和高透平面分辨率的自由运行的3t电影成像。虽然与2D屏气片相比，心肌-血液对比降低，但保留的对比足以评估左室功能。
{"title":"Non-contrast free-running high-resolution volumetric multi-slab cardiac cine MRI at 3 T","authors":"Chenhao Gao, Fan Yang, Zhihao Xue, Junyao Zhang, Zhuo Chen, Sirui Huo, Juan Gao, Yixin Emu, Haiyang Chen, Chenxi Hu","doi":"10.1016/j.mri.2025.110591","DOIUrl":"10.1016/j.mri.2025.110591","url":null,"abstract":"<div><div>Free-running self-gated 3D cardiac cine imaging is highly desirable for volumetric, high-resolution, breath-hold-free assessment of left ventricular (LV) function. However, its implementation at 3 T remains challenging due to specific absorption rate (SAR) constraints and reduced myocardium-blood contrast. In this study, we propose a novel non-contrast, free-running, self-gated 3D gradient-echo (GRE) cine sequence for 3 T imaging, which acquires multi-slab data using a pseudo-radial Cartesian trajectory with a 1.5 mm slice thickness. To address respiratory motion, a locally low-rank motion-corrected image reconstruction algorithm was developed. Fifteen participants underwent imaging with the proposed multi-slab 3D cine sequence and conventional 2D cine sequences. Additionally, single-slab 3D cine data were acquired in 10 participants. Various image quality metrics (signal-to-noise ratio, contrast-to-noise ratio, myocardial sharpness, and residual artefact) and LV volumetric parameters (end-diastolic volume (EDV), end-systolic volume (ESV), and ejection fraction (EF)) were compared between the different methods. Results demonstrated that the proposed multi-slab 3D cine method provided significantly superior image quality compared to the single-slab 3D approach (myocardial sharpness: 2.82 ± 0.42 vs. 1.58 ± 0.46, <em>P</em> = 0.005; residual artefact: 2.58 ± 0.26 vs. 1.13 ± 0.21, P = 0.005) due to the improvement of image contrast. Furthermore, the multi-slab 3D cine sequence exhibited good agreement and correlation with the reference 2D cine method in terms of volumetric measures (EDV: 140.3 ± 19.9 mL vs. 139.3 ± 20.0 mL, <em>P</em> = 0.357, <em>r</em> = 0.976; ESV: 56.7 ± 11.8 mL vs. 57.7 ± 10.2 mL, <em>P</em> = 0.259, <em>r</em> = 0.964; EF: 59.8 % ± 4.9 % vs. 58.7 % ± 3.7 %, <em>P</em> = 0.073, <em>r</em> = 0.907). In conclusion, the proposed multi-slab 3D cine framework enables free-running 3 T cine imaging with whole-heart coverage and high through-plane resolution. Although myocardium-blood contrast is reduced compared to 2D breath-hold cine, the retained contrast is sufficient to evaluate LV function.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110591"},"PeriodicalIF":2.0,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145735213","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}
This study aimed to evaluate, qualitatively and quantitatively, cross-sectional black-blood images obtained using T2-prepared phase-sensitive inversion-recovery steady-state free precession (T2PSIR-SSFP), in comparison with conventional double inversion recovery turbo spin-echo (DIR-TSE), in patients with Kawasaki disease (KD), and to assess the feasibility of T2PSIR-SSFP imaging.
Materials and methods
Nine patients (three female and six male; median age, 6.2 years; range, 8 months–14 years) were enrolled. Black-blood imaging was separately analyzed in aneurysmal and regressed aneurysmal regions. Lumen and outer wall boundary image quality was visually graded using a four-point scale. Lumen area (LA) reproducibility measurements were determined using intraclass correlation coefficients (ICCs) between T2PSIR-SSFP and coronary magnetic resonance angiography (MRA) images, as well as between DIR-TSE and MRA. Agreement between T2PSIR-SSFP and MRA was further examined using Bland–Altman analysis.
Results
A total of 22 coronary regions (11 aneurysmal and 11 regressed aneurysmal) were assessed. T2PSIR-SSFP exhibited excellent reproducibility with MRA in both aneurysmal and regressed aneurysmal regions (ICCs = 0.99 and 1.00, respectively). DIR-TSE showed high reproducibility in regressed aneurysmal regions (ICC = 0.93) but poor agreement in aneurysmal regions (ICC = 0.43). Bland–Altman analysis revealed strong agreement between T2PSIR-SSFP and MRA, with no fixed or proportional bias in either region (P > 0.1).
Conclusions
Flow-independent coronary black-blood imaging using T2PSIR-SSFP provided values within the expected range in patients with KD. T2PSIR-SSFP imaging appears suitable for KD follow-up because it can provide accurate cross-sectional images and reproducibility of LA measurements.
{"title":"Coronary artery black-blood imaging via T2-prepared phase-sensitive inversion-recovery steady-state free precession in Kawasaki disease","authors":"Koji Matsumoto , Hajime Yokota , Hiroki Mukai , Ryota Ebata , Kentaro Okunushi , Hiromichi Hamada , Hiroyuki Takaoka , Masami Yoneyama , Takashi Namiki , Takashi Iimori , Takashi Uno","doi":"10.1016/j.mri.2025.110589","DOIUrl":"10.1016/j.mri.2025.110589","url":null,"abstract":"<div><h3>Purpose</h3><div>This study aimed to evaluate, qualitatively and quantitatively, cross-sectional black-blood images obtained using T<sub>2</sub>-prepared phase-sensitive inversion-recovery steady-state free precession (T<sub>2</sub>PSIR-SSFP), in comparison with conventional double inversion recovery turbo spin-echo (DIR-TSE), in patients with Kawasaki disease (KD), and to assess the feasibility of T<sub>2</sub>PSIR-SSFP imaging.</div></div><div><h3>Materials and methods</h3><div>Nine patients (three female and six male; median age, 6.2 years; range, 8 months–14 years) were enrolled. Black-blood imaging was separately analyzed in aneurysmal and regressed aneurysmal regions. Lumen and outer wall boundary image quality was visually graded using a four-point scale. Lumen area (LA) reproducibility measurements were determined using intraclass correlation coefficients (ICCs) between T<sub>2</sub>PSIR-SSFP and coronary magnetic resonance angiography (MRA) images, as well as between DIR-TSE and MRA. Agreement between T<sub>2</sub>PSIR-SSFP and MRA was further examined using Bland–Altman analysis.</div></div><div><h3>Results</h3><div>A total of 22 coronary regions (11 aneurysmal and 11 regressed aneurysmal) were assessed. T<sub>2</sub>PSIR-SSFP exhibited excellent reproducibility with MRA in both aneurysmal and regressed aneurysmal regions (ICCs = 0.99 and 1.00, respectively). DIR-TSE showed high reproducibility in regressed aneurysmal regions (ICC = 0.93) but poor agreement in aneurysmal regions (ICC = 0.43). Bland–Altman analysis revealed strong agreement between T<sub>2</sub>PSIR-SSFP and MRA, with no fixed or proportional bias in either region (<em>P</em> > 0.1).</div></div><div><h3>Conclusions</h3><div>Flow-independent coronary black-blood imaging using T<sub>2</sub>PSIR-SSFP provided values within the expected range in patients with KD. T<sub>2</sub>PSIR-SSFP imaging appears suitable for KD follow-up because it can provide accurate cross-sectional images and reproducibility of LA measurements.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110589"},"PeriodicalIF":2.0,"publicationDate":"2025-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145708504","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}
Pub Date : 2025-12-01DOI: 10.1016/j.mri.2025.110579
Yuan Lian, Yuancheng Jiang, Hua Guo
Purpose
Proper regularization weights are crucial for the reconstruction quality of compressed sensing (CS) MRI. This work aims to develop an automatic and adaptive regularization weights selection method for CS reconstruction
Methods
A statistical model based on Bayesian theory is designed, incorporating prior information about the Gaussian distribution of incoherent noise and the Laplacian distribution of wavelet coefficients in the wavelet transform domain. Using the variance of coefficients and noise, the adaptive regularization weight for achieving optimal reconstruction quality in each iteration step is obtained through a maximum a posteriori estimator. The adaptive regularization weights vary across different subjects, slices, iterations, and wavelet sub-bands
Results
The efficacy of the proposed method was demonstrated through retrospective and prospective studies. Compared to reconstruction results using optimal fixed regularization weights and sparsity-adaptive composite recovery method (SCoRe), the proposed method successfully reduces reconstruction errors and effectively recovers original signals from noise-like incoherent artifacts in the wavelet transform domain. It also saves weight selection time when searching for optimal fixed regularization weights
Conclusion
We propose an adaptive regularization weights selection method for CS-MRI reconstruction. It provides optimal regularization weights for different subjects, slices, and iterations without requiring manual intervention
{"title":"Adaptive regularization weight selection for compressed sensing MRI reconstruction","authors":"Yuan Lian, Yuancheng Jiang, Hua Guo","doi":"10.1016/j.mri.2025.110579","DOIUrl":"10.1016/j.mri.2025.110579","url":null,"abstract":"<div><h3>Purpose</h3><div>Proper regularization weights are crucial for the reconstruction quality of compressed sensing (CS) MRI. This work aims to develop an automatic and adaptive regularization weights selection method for CS reconstruction</div></div><div><h3>Methods</h3><div>A statistical model based on Bayesian theory is designed, incorporating prior information about the Gaussian distribution of incoherent noise and the Laplacian distribution of wavelet coefficients in the wavelet transform domain. Using the variance of coefficients and noise, the adaptive regularization weight for achieving optimal reconstruction quality in each iteration step is obtained through a maximum a posteriori estimator. The adaptive regularization weights vary across different subjects, slices, iterations, and wavelet sub-bands</div></div><div><h3>Results</h3><div>The efficacy of the proposed method was demonstrated through retrospective and prospective studies. Compared to reconstruction results using optimal fixed regularization weights and sparsity-adaptive composite recovery method (SCoRe), the proposed method successfully reduces reconstruction errors and effectively recovers original signals from noise-like incoherent artifacts in the wavelet transform domain. It also saves weight selection time when searching for optimal fixed regularization weights</div></div><div><h3>Conclusion</h3><div>We propose an adaptive regularization weights selection method for CS-MRI reconstruction. It provides optimal regularization weights for different subjects, slices, and iterations without requiring manual intervention</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110579"},"PeriodicalIF":2.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145668927","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}
Pub Date : 2025-11-27DOI: 10.1016/j.mri.2025.110577
Baihe Luo , Aoran Yang , Jialin Li , Chen Pan, Chunli Li, Minghui Zhou, Zhiying Wang, Chengli Gu, Xiaoli Yin, Yun Zhao, Yu Shi
Objective
Tyrosine kinase inhibitors (TKIs), such as sorafenib, are standard therapies for advanced hepatocellular carcinoma (HCC), but their biomechanical impact and the role of magnetic resonance elastography (MRE) in treatment evaluation remain unclear. This study explored whether TKIs reduce tumor stiffness by inhibiting malignant behavior and whether MRE can detect such changes early.
Methods
A prospective animal study was performed using subcutaneous SK-HEP-1 HCC xenografts in 50 nude rats. Forty tumor-bearing rats were randomized to control or sorafenib-treated groups (n = 20 each). Multiparametric 3.0 T MRI included T1- and T2-weighted imaging, T1/T2/T2* mapping, and MRE at 200 Hz and 100 Hz. Imaging was conducted at baseline (∼2 cm3 tumor volume) and on days 1, 2, and 3 post-intervention. Histology involved H&E and immunohistochemistry for VEGFR-1, BRAF, Ki67, and TUNEL. Ex vivo stiffness was measured by atomic force microscopy. Cell behavior was assessed by EdU, Transwell, CCK-8, and Western blot. Statistical analysis included ICC, Bland–Altman, Mann–Whitney U, repeated measures ANOVA, Spearman correlation, and multivariate regression.
Results
TKIs reduced tumor stiffness at cellular (P = 0.02) and tissue (P = 0.004) levels. Stiffness decreased by day 2 at 200 Hz and day 3 at both frequencies. Treated tumors showed reduced cellularity, lower Ki67, and increased apoptosis. Stiffness correlated with cellularity (r = 0.527) and Ki67 (r = 0.623), both predicting MRE stiffness (R2 = 0.537).
Conclusion
TKIs reduce stiffness and malignancy in HCC. MRE is a promising tool for early treatment response evaluation.
{"title":"Tumor stiffness as an imaging biomarker of tyrosine kinase inhibitor response: A preclinical study","authors":"Baihe Luo , Aoran Yang , Jialin Li , Chen Pan, Chunli Li, Minghui Zhou, Zhiying Wang, Chengli Gu, Xiaoli Yin, Yun Zhao, Yu Shi","doi":"10.1016/j.mri.2025.110577","DOIUrl":"10.1016/j.mri.2025.110577","url":null,"abstract":"<div><h3>Objective</h3><div>Tyrosine kinase inhibitors (TKIs), such as sorafenib, are standard therapies for advanced hepatocellular carcinoma (HCC), but their biomechanical impact and the role of magnetic resonance elastography (MRE) in treatment evaluation remain unclear. This study explored whether TKIs reduce tumor stiffness by inhibiting malignant behavior and whether MRE can detect such changes early.</div></div><div><h3>Methods</h3><div>A prospective animal study was performed using subcutaneous SK-HEP-1 HCC xenografts in 50 nude rats. Forty tumor-bearing rats were randomized to control or sorafenib-treated groups (<em>n</em> = 20 each). Multiparametric 3.0 T MRI included T1- and T2-weighted imaging, T1/T2/T2* mapping, and MRE at 200 Hz and 100 Hz. Imaging was conducted at baseline (∼2 cm<sup>3</sup> tumor volume) and on days 1, 2, and 3 post-intervention. Histology involved H&E and immunohistochemistry for VEGFR-1, BRAF, Ki67, and TUNEL. Ex vivo stiffness was measured by atomic force microscopy. Cell behavior was assessed by EdU, Transwell, CCK-8, and Western blot. Statistical analysis included ICC, Bland–Altman, Mann–Whitney U, repeated measures ANOVA, Spearman correlation, and multivariate regression.</div></div><div><h3>Results</h3><div>TKIs reduced tumor stiffness at cellular (<em>P</em> = 0.02) and tissue (<em>P</em> = 0.004) levels. Stiffness decreased by day 2 at 200 Hz and day 3 at both frequencies. Treated tumors showed reduced cellularity, lower Ki67, and increased apoptosis. Stiffness correlated with cellularity (<em>r</em> = 0.527) and Ki67 (<em>r</em> = 0.623), both predicting MRE stiffness (R<sup>2</sup> = 0.537).</div></div><div><h3>Conclusion</h3><div>TKIs reduce stiffness and malignancy in HCC. MRE is a promising tool for early treatment response evaluation.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110577"},"PeriodicalIF":2.0,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145634504","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}
Pub Date : 2025-11-27DOI: 10.1016/j.mri.2025.110578
Kevin Sun Zhang , Philip Alexander Glemser , Christian Jan Oliver Neelsen , Markus Wennmann , Lukas Thomas Rotkopf , Nils Netzer , Clara Meinzer , Thomas Hielscher , Vivienn Weru , Magdalena Görtz , Albrecht Stenzinger , Markus Hohenfellner , Heinz-Peter Schlemmer , David Bonekamp
Objectives
To assess variability of maximum diameter measurements of prostate lesions in MRI assessing patient repositioning, rater and sequence effects.
Methods
Forty-two patients were included retrospectively, who received a clinical bi−/multiparametric prostate MRI examination and agreed to have the T2-weighted (T2WI) and diffusion weighted-imaging (DWI) sequences scanned twice. Maximum diameter measurements of prostate lesions mentioned in the clinical radiologist reports were performed by four readers in multiple reading sessions for determination of inter-sequence (between two DWI sequences), inter-scan (between clinical and additional scan), intra-rater and inter-rater variability. The primary calculated metrics were the repeatability and reproducibility coefficient (RC/RDC), including pooled RC/RDC.
Results
Variability measured by RCs/RDCs was lowest for measurements obtained within the same reading session, with inter-scan RCs up to 5.6 mm/6.5 mm for T2WI/DWI, pooled RCs of 4.8 mm/5.8 mm, respectively, and inter-sequence RDCs of 5.4 mm–5.9 mm, pooled RDC 5.8 mm. Measurements performed in separate reading sessions demonstrated significantly higher variability for both settings in the majority of cases (RCs: up to 10.9 mm/11.7 mm/10.2 mm for T2WI/DWI/inter-sequence, p ≤ 0.002), pooled RCs/RDCs 9.2 mm–9.9 mm. Measurements necessarily generated in different reading sessions, i.e., intra-rater or inter-rater, demonstrated high variability (RCs/RDCs up to 11.4 mm/11.5 mm for T2WI/DWI).
Conclusions
Prostate lesion measurements demonstrate considerable variability. When measured in one reading session by one rater, lesion diameter differences below the pooled RCs of 4.8 mm, 95 %-CI [3.9, 5.6] for T2WI and 5.8 mm, 95 %-CI [4.7, 7.1] for DWI should not necessarily assumed to be true biological change, as these differences may result from measurement- or repositioning-based variability alone. Caution needs to be taken assessing size changes.
目的：评估磁共振成像中前列腺病变最大直径测量的可变性，以评估患者重新定位、排序和序列效应。方法：回顾性分析42例接受临床双参数/多参数前列腺MRI检查的患者，并同意进行2次t2加权（T2WI）和弥散加权成像（DWI）序列扫描。临床放射科医生报告中提到的前列腺病变的最大直径测量由四名读取器在多次读取会话中完成，以确定序列间（两个DWI序列之间）、扫描间（临床和附加扫描之间）、分级内和分级间的变异性。主要计算指标为重复性和再现性系数（RC/RDC），包括合并RC/RDC。结果：在相同的读数过程中，RCs/RDC测量的变变性最低，T2WI/DWI的扫描间RCs高达5.6 mm/6.5 mm，合并RCs分别为4.8 mm/5.8 mm，序列间RDC为5.4 mm-5.9 mm，合并RDC为5.8 mm。在单独的读数过程中进行的测量显示，在大多数情况下，这两种设置的变异性显著更高（T2WI/DWI/序列间的RCs：高达10.9 mm/11.7 mm/10.2 mm, p ≤ 0.002），合并的RCs/RDCs为9.2 mm-9.9 mm。在不同的阅读过程中产生的测量结果，即内部或内部的测量结果，显示出很高的可变性（T2WI/DWI的RCs/ rdc高达11.4 mm/11.5 mm）。结论：前列腺病变测量显示出相当大的可变性。当由一名评估者在一次读数中测量时，T2WI的病变直径差异低于4.8 mm, 95% %- ci [3.9, 5.6]， DWI的病变直径差异低于5.8 mm, 95% %- ci[4.7, 7.1]，这并不一定被认为是真正的生物学变化，因为这些差异可能仅仅是由测量或重新定位的可变性造成的。评估大小变化时需要谨慎。
{"title":"Repeatability and reproducibility of maximum diameter measurements of prostate lesions on MRI with repositioning and variation of imaging sequences: A test-retest study","authors":"Kevin Sun Zhang , Philip Alexander Glemser , Christian Jan Oliver Neelsen , Markus Wennmann , Lukas Thomas Rotkopf , Nils Netzer , Clara Meinzer , Thomas Hielscher , Vivienn Weru , Magdalena Görtz , Albrecht Stenzinger , Markus Hohenfellner , Heinz-Peter Schlemmer , David Bonekamp","doi":"10.1016/j.mri.2025.110578","DOIUrl":"10.1016/j.mri.2025.110578","url":null,"abstract":"<div><h3>Objectives</h3><div>To assess variability of maximum diameter measurements of prostate lesions in MRI assessing patient repositioning, rater and sequence effects.</div></div><div><h3>Methods</h3><div>Forty-two patients were included retrospectively, who received a clinical bi−/multiparametric prostate MRI examination and agreed to have the T2-weighted (T2WI) and diffusion weighted-imaging (DWI) sequences scanned twice. Maximum diameter measurements of prostate lesions mentioned in the clinical radiologist reports were performed by four readers in multiple reading sessions for determination of inter-sequence (between two DWI sequences), inter-scan (between clinical and additional scan), intra-rater and inter-rater variability. The primary calculated metrics were the repeatability and reproducibility coefficient (RC/RDC), including pooled RC/RDC.</div></div><div><h3>Results</h3><div>Variability measured by RCs/RDCs was lowest for measurements obtained within the same reading session, with inter-scan RCs up to 5.6 mm/6.5 mm for T2WI/DWI, pooled RCs of 4.8 mm/5.8 mm, respectively, and inter-sequence RDCs of 5.4 mm–5.9 mm, pooled RDC 5.8 mm. Measurements performed in separate reading sessions demonstrated significantly higher variability for both settings in the majority of cases (RCs: up to 10.9 mm/11.7 mm/10.2 mm for T2WI/DWI/inter-sequence, <em>p</em> ≤ 0.002), pooled RCs/RDCs 9.2 mm–9.9 mm. Measurements necessarily generated in different reading sessions, i.e., intra-rater or inter-rater, demonstrated high variability (RCs/RDCs up to 11.4 mm/11.5 mm for T2WI/DWI).</div></div><div><h3>Conclusions</h3><div>Prostate lesion measurements demonstrate considerable variability. When measured in one reading session by one rater, lesion diameter differences below the pooled RCs of 4.8 mm, 95 %-CI [3.9, 5.6] for T2WI and 5.8 mm, 95 %-CI [4.7, 7.1] for DWI should not necessarily assumed to be true biological change, as these differences may result from measurement- or repositioning-based variability alone. Caution needs to be taken assessing size changes.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"127 ","pages":"Article 110578"},"PeriodicalIF":2.0,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145634519","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}
Magnetic resonance imaging (MRI) is a crucial medical imaging modality. However, long acquisition times remain a significant challenge, leading to increased costs, and reduced patient comfort. Recent studies have shown the potential of using deep learning models that incorporate information from prior subject-specific MRI scans to improve reconstruction quality of present scans. Integrating this prior information requires registration of the previous scan to the current image reconstruction, which can be time-consuming. We propose a novel deep-learning-based MRI reconstruction framework which consists of an initial reconstruction network, a deep registration model, and a transformer-based enhancement network. We validated our method on a longitudinal dataset of T1-weighted MRI scans with 2808 images from 18 subjects at four acceleration factors (R5, R10, R15, R20). Quantitative metrics confirmed our approach’s superiority over existing methods (p 0.05, Wilcoxon signed-rank test). Furthermore, we analyzed the impact of our MRI reconstruction method on the downstream task of brain segmentation and observed improved accuracy and volumetric agreement with reference segmentations. Our approach also achieved a substantial reduction in total reconstruction time compared to methods that use traditional registration algorithms, making it more suitable for real-time clinical applications. The code associated with this work is publicly available at https://github.com/amirshamaei/longitudinal-mri-deep-recon.
{"title":"Enhancing and accelerating brain MRI through deep learning reconstruction using prior subject-specific imaging","authors":"Amirmohammad Shamaei , Alexander Stebner , Salome (Lou) Bosshart , Johanna Ospel , Gouri Ginde , Mariana Bento , Roberto Souza","doi":"10.1016/j.mri.2025.110558","DOIUrl":"10.1016/j.mri.2025.110558","url":null,"abstract":"<div><div>Magnetic resonance imaging (MRI) is a crucial medical imaging modality. However, long acquisition times remain a significant challenge, leading to increased costs, and reduced patient comfort. Recent studies have shown the potential of using deep learning models that incorporate information from prior subject-specific MRI scans to improve reconstruction quality of present scans. Integrating this prior information requires registration of the previous scan to the current image reconstruction, which can be time-consuming. We propose a novel deep-learning-based MRI reconstruction framework which consists of an initial reconstruction network, a deep registration model, and a transformer-based enhancement network. We validated our method on a longitudinal dataset of T1-weighted MRI scans with 2808 images from 18 subjects at four acceleration factors (R5, R10, R15, R20). Quantitative metrics confirmed our approach’s superiority over existing methods (p <span><math><mo><</mo></math></span> 0.05, Wilcoxon signed-rank test). Furthermore, we analyzed the impact of our MRI reconstruction method on the downstream task of brain segmentation and observed improved accuracy and volumetric agreement with reference segmentations. Our approach also achieved a substantial reduction in total reconstruction time compared to methods that use traditional registration algorithms, making it more suitable for real-time clinical applications. The code associated with this work is publicly available at <span><span>https://github.com/amirshamaei/longitudinal-mri-deep-recon</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"126 ","pages":"Article 110558"},"PeriodicalIF":2.0,"publicationDate":"2025-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145570712","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}
To assess the correlation between ASL-derived perfusion parameters and enhancement pattern on DCE in patients with histologically uncharacterized lesions of the peripheral prostatic zone.
Materials and methods
Prospective study: informed consent was obtained. We included 52 patients with peripheral zone lesions ≥4 mm, PI-RADS ≥3. All patients underwent PI-RADS-recommended multiparametric MRI and ASL acquisition using a Flow-sensitive Alternating Inversion Recovery (FAIR) with True Fast Imaging with Steady-State Precession (True-FISP) pulse sequence at 3 T. DCE images were assessed for early focal enhancement (DCE+). Two radiologists independently measured prostatic blood flow (PBF) in the lesion and adjacent normal PZ tissue; the mean of the values was used for further analysis. The PBF ratio (lesion/healthy tissue) was calculated. Statistical analysis included Mann-Whitney U, Wilcoxon tests, and ROC curve analysis.
Results
Fifty-two lesions with median diameter of 8 mm (IQR: 6–12 mm) were included. On DCE, 31/52 (60 %) were DCE+, and 21/52 (40 %) were DCE−. Interobserver agreement for PBF measurements was excellent (ICC 0.980–0.994). Median PBF was significantly higher in target lesions than in healthy tissue (28.03 vs. 15.76 ml/100 g/min, p = 0.0003). DCE+ lesions showed significantly higher PBF than DCE– ones (39.17 vs. 19.04 ml/100 g/min; p = 0.0001), as well as a significantly higher PBF ratio (2.6 vs. 1.09; p < 0.0001). On ROC analysis, PBF ratio showed an AUC of 0.97 (95 % CI: 0.87–0.99; p < 0.0001) in discriminating between DCE+ and DCE− lesions; a cut-off value of >1.66 provided 86 % sensitivity and 100 % specificity in identifying DCE+ lesions.
Conclusions
ASL may serve as a safe, contrast-free alternative for evaluating vascularization of peripheral prostatic zone lesions.
目的探讨前列腺周围区组织学特征不明显病变患者asl灌注参数与DCE增强模式的相关性。材料与方法前瞻性研究：获得知情同意。我们纳入52例外周区病变≥4mm， PI-RADS≥3的患者。所有患者都接受了pi - rads推荐的多参数MRI和ASL采集，使用流量敏感交替反转恢复（FAIR）和真正的快速成像与稳态进动（True- fisp）脉冲序列在3 t时进行DCE图像的早期病灶增强（DCE+）评估。两名放射科医生独立测量病变和邻近正常PZ组织的前列腺血流量（PBF）；取平均值作进一步分析。计算PBF比（病变/健康组织）。统计分析采用Mann-Whitney U检验、Wilcoxon检验和ROC曲线分析。结果共纳入52个中位直径为8 mm （IQR: 6 ~ 12 mm）的病变。在DCE中，31/52（60%）为DCE+， 21/52（40%）为DCE−。观察者间对PBF测量的一致性非常好（ICC 0.980-0.994）。靶病变的中位PBF显著高于健康组织（28.03 vs. 15.76 ml/100 g/min, p = 0.0003）。DCE+病变PBF显著高于DCE -病变（39.17 vs. 19.04 ml/100 g/min, p = 0.0001）， PBF比值显著高于DCE -病变（2.6 vs. 1.09, p < 0.0001）。在ROC分析中，PBF比值显示区分DCE+和DCE -病变的AUC为0.97 (95% CI: 0.87-0.99; p < 0.0001)；截断值>；1.66为鉴别DCE+病变提供了86%的敏感性和100%的特异性。结论sasl可作为一种安全、无对比的评价前列腺周围区病变血管化的方法。
{"title":"Arterial spin labelling as a contrast-free alternative to dynamic contrast enhancement for the evaluation of peripheral zone prostatic lesions on MRI","authors":"Valentina Corato , Vincenzo Vingiani , Bernardo Proner , Petros Martirosian , Armin Pycha , Emanuela Trenti , Alessandro Lanaro , Giovanna Nordio , Riccardo Valletta , Matteo Bonatti","doi":"10.1016/j.mri.2025.110575","DOIUrl":"10.1016/j.mri.2025.110575","url":null,"abstract":"<div><h3>Purpose</h3><div>To assess the correlation between ASL-derived perfusion parameters and enhancement pattern on DCE in patients with histologically uncharacterized lesions of the peripheral prostatic zone.</div></div><div><h3>Materials and methods</h3><div>Prospective study: informed consent was obtained. We included 52 patients with peripheral zone lesions ≥4 mm, PI-RADS ≥3. All patients underwent PI-RADS-recommended multiparametric MRI and ASL acquisition using a Flow-sensitive Alternating Inversion Recovery (FAIR) with True Fast Imaging with Steady-State Precession (True-FISP) pulse sequence at 3 T. DCE images were assessed for early focal enhancement (DCE+). Two radiologists independently measured prostatic blood flow (PBF) in the lesion and adjacent normal PZ tissue; the mean of the values was used for further analysis. The PBF ratio (lesion/healthy tissue) was calculated. Statistical analysis included Mann-Whitney U, Wilcoxon tests, and ROC curve analysis.</div></div><div><h3>Results</h3><div>Fifty-two lesions with median diameter of 8 mm (IQR: 6–12 mm) were included. On DCE, 31/52 (60 %) were DCE+, and 21/52 (40 %) were DCE−. Interobserver agreement for PBF measurements was excellent (ICC 0.980–0.994). Median PBF was significantly higher in target lesions than in healthy tissue (28.03 vs. 15.76 ml/100 g/min, <em>p</em> = 0.0003). DCE+ lesions showed significantly higher PBF than DCE– ones (39.17 vs. 19.04 ml/100 g/min; <em>p</em> = 0.0001), as well as a significantly higher PBF ratio (2.6 vs. 1.09; <em>p</em> < 0.0001). On ROC analysis, PBF ratio showed an AUC of 0.97 (95 % CI: 0.87–0.99; p < 0.0001) in discriminating between DCE+ and DCE− lesions; a cut-off value of >1.66 provided 86 % sensitivity and 100 % specificity in identifying DCE+ lesions.</div></div><div><h3>Conclusions</h3><div>ASL may serve as a safe, contrast-free alternative for evaluating vascularization of peripheral prostatic zone lesions.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"126 ","pages":"Article 110575"},"PeriodicalIF":2.0,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145570714","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}
Pub Date : 2025-11-19DOI: 10.1016/j.mri.2025.110574
Ke Liang , Zhijin Lin , Jie Yang , Teng Yu , Jieru Chi , Feng Liu
In clinical practice, magnetic resonance imaging is essential for disease diagnosis and evaluation, although it generally requires a prolonged scanning duration. Recently, several deep learning-based restoration techniques have been introduced to accelerate MRI acquisition. However, most existing MRI restoration methods struggle to fully capture local texture features and have limitations in fusing global and local features. To address this issue, the paper introduces a dual path Mamba network. This method uses a strategy of multi-scale local feature and global feature fusion for under-sampled single-coil image domain data, significantly improving both the quality and efficiency of image restoration. Specifically, 1) A multi-scale local Mamba block is proposed, which extracts local information from different regions through a multi-scale window mechanism, capturing diverse local features. 2) A new feature fusion block is proposed, which fuses global and local information to enhance the completeness of feature expression. 3) A dual path Mamba network architecture is proposed. This dual-path design significantly improves feature extraction capability and adaptability in complex and dynamic data environments. Through comprehensive experiments on the NAMIC, fastMRI and BraTS datasets, it is shown that the proposed network surpasses current leading methods across various evaluation metrics, particularly excelling in restoring texture details and tissue structures.
{"title":"GL-mamba-net: A magnetic resonance imaging restoration network with global-local mamba","authors":"Ke Liang , Zhijin Lin , Jie Yang , Teng Yu , Jieru Chi , Feng Liu","doi":"10.1016/j.mri.2025.110574","DOIUrl":"10.1016/j.mri.2025.110574","url":null,"abstract":"<div><div>In clinical practice, magnetic resonance imaging is essential for disease diagnosis and evaluation, although it generally requires a prolonged scanning duration. Recently, several deep learning-based restoration techniques have been introduced to accelerate MRI acquisition. However, most existing MRI restoration methods struggle to fully capture local texture features and have limitations in fusing global and local features. To address this issue, the paper introduces a dual path Mamba network. This method uses a strategy of multi-scale local feature and global feature fusion for under-sampled single-coil image domain data, significantly improving both the quality and efficiency of image restoration. Specifically, 1) A multi-scale local Mamba block is proposed, which extracts local information from different regions through a multi-scale window mechanism, capturing diverse local features. 2) A new feature fusion block is proposed, which fuses global and local information to enhance the completeness of feature expression. 3) A dual path Mamba network architecture is proposed. This dual-path design significantly improves feature extraction capability and adaptability in complex and dynamic data environments. Through comprehensive experiments on the NAMIC, fastMRI and BraTS datasets, it is shown that the proposed network surpasses current leading methods across various evaluation metrics, particularly excelling in restoring texture details and tissue structures.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"126 ","pages":"Article 110574"},"PeriodicalIF":2.0,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145570713","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号