Efficacy Evaluation of AI-Assisted Compressed Sensing Combined with Deep-Learning Reconstruction in Accelerating Brain T2-Weighted Imaging: A Clinical Feasibility Study

Efficacy Evaluation of AI-Assisted Compressed Sensing Combined with Deep-Learning Reconstruction in Accelerating Brain T2-Weighted Imaging: A Clinical Feasibility Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Efficacy Evaluation of AI-Assisted Compressed Sensing Combined with Deep-Learning Reconstruction in Accelerating Brain T2-Weighted Imaging: A Clinical Feasibility Study Shiwei Lai, Yong Xiang, Shaoxin Xiang, Jiahuan Dai, Zuonian Hu, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7239920/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background While using AI-assisted compressed sensing (ACS) combined with deep-learning reconstruction (DR) techniques has the potential to Shorten the acquisition time of brain T2-weighted imaging (T2WI), its imaging improvements and clinical application in brain imaging remains under explored. Purpose To evaluate the clinical feasibility of DR-ACS for brain T2-weighted imaging (T2WI), focusing on image quality, acquisition efficiency, and diagnostic accuracy, compared with the routine T2WI. Material and Methods A prospective cohort of 110 participants underwent brain MRI using three protocols at a 3.0-T MR scanner: routine T2WI, ACS-T2WI (without DR), and DR-ACS-T2WI. Subjective image quality (overall image quality, motion artifact, and diagnostic confidence, with a 5-point scale) and objective metrics (Signal-to-Noise Ratio (SNR), Contrast-to-Noise Ratio (CNR), and scan time) were compared. Statistical analysis included ANOVA and kappa. Results The overall image quality, motion artifact, and diagnostic confidence scores of DR-ACS-T2WI, assessed by two radiologists, were 4.90±0.30, 4.91±0.29 , 4.92±0.28(Reader 1) and 4.90±0.30 , 4.91±0.28, 4.91±0.30 (Reader 2), higher than those of ACS-T2WI and routine T2WI. DR-ACS-T2WI demonstrated superior SNRs in both white matter (65.06±12.1) and gray matter (97.25±18.52) compared to ACS-T2WI (47.62±8.65 and 71.54±12.05, respectively) and routine T2WI (34.32±6.51 and 51.92±8.62, respectively; P < 0.001 for all). Similarly, the gray-white matter CNR of DR-ACS-T2WI (32.93±12.35) was significantly higher than that of ACS-T2WI (24.29±9.08) and routine T2WI (17.31±6.01; P < 0.001). Additionally, the scan time of DR-ACS-T2WI and ACS-T2WI (both 25.7s) was 57.87% shorter than that of the routine T2WI (61s). Conclusion The ACS combined with DR is clinically feasible for MRI examinations of brain diseases, offering significantly shorter image acquisition time and higher image quality compared with the routine T2WI. AI-assisted compressed sensing deep-learning reconstruction brain MRI Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Magnetic resonance imaging (MRI) is indispensable for neurological diagnostics. T2-weighted sequences remain the clinical cornerstone for detecting fluid-associated pathologies, derived from their enhanced tissue contrast resolution in hydrophilic environments and heightened sensitivity to proton mobility alterations. By highlighting differences in tissue relaxation times, T2WI enables radiologists to identify various pathological changes, such as cerebral infarcts, tumors, and demyelinating lesions, with high sensitivity. 1 Yet the relatively long scan times remain a significant challenge, particularly for uncooperative patients. Traditional acceleration techniques, such as parallel imaging 2 and compressed sensing (CS) 3 , have been developed to address this issue. However, these methods often entail a compromise between acceleration efficiency and image quality. Specifically, the primary drawback of parallel imaging techniques is the reduction in the acquisition of phase encoding lines, which results in decreased signal-to-noise ratio (SNR) and increased noise in the image. Similarly, simply applying compressed sensing techniques for image acquisition also leads to degradation in image quality. The application of these traditional acceleration techniques frequently results in a decrement of SNR or spatial resolution, limiting diagnostic confidence. 4 However, the brain harbors highly intricate structures, including neural nuclei, cerebral vasculature, and cranial nerves. Given the minimal signal contrast between various brain tissues and pathological entities, high - resolution MRI with an elevated SNR and contrast-to-noise ratio (CNR) is essential for clear visualization. The integration of artificial intelligence (AI) into MRI workflows has catalyzed paradigm shifts across acquisition and reconstruction pipelines, redefining clinical imaging capabilities 5 , 6 , 7 . In the field of image acquisition, AI-driven compressed sensing (ACS) emerges from multidimensional integration of parallel imaging, compressed sensing, half-Fourier sampling, and deep neural architectures, achieving scan time reductions without compromising diagnostic precision. 8 , 9 ACS have achieved substantial scan time reductions of over 50% in liver and cardiac, thereby optimizing clinical workflow efficiency and reduce the impact of breathing movements on image quality. 10 , 11 , 12 Concurrently, deep learning reconstruction (DR) technique synchronize k-space manipulation with image enhancement through adaptive noise-signal discrimination, optimizing both signal-to-noise ratio and contrast-to-noise ratio metrics. Deep learning reconstruction can be effectively applied to diffusion-weighted imaging (DWI) across multiple anatomical regions, demonstrating capabilities in mitigating geometric distortion and reducing scan time. 13 , 14 Furthermore, this technique enables optimal image quality enhancement in MRI systems with different magnetic field strengths, including 1.5T, 3T, etc. 15 , 16 These parallel advancements demonstrate how intelligent algorithm design can overcome traditional MRI limitations through data-driven information retrieval and computational signal processing. While extant literature substantiates the neuroimaging efficacy of ACS and DR as standalone modalities, both parallel imaging and compressed sensing techniques result in the loss of k-space information while accelerating the scanning process, leading to degradation of image quality. 17 , 18 , 19 Critical gaps persist in delineating their synergistic effects when implemented within dual-modality AI frameworks, highlighting the need for a systematic evaluation of integrating ACS and DR in brain MRI. 20 Therefore, integrating ACS-accelerated and deep learning technology enables not only accelerated imaging but also enhances overall image quality. 21 Consequently, this study investigates to assess the efficacy of integrating ACS and DR techniques in transcending the inherent limitations of conventional T2-weighted imaging (T2WI). Specifically, the study aims to determine if this integrated approach can achieve significant reductions in scan duration, concomitant enhancements in SNR and CNR, and substantial minimization of motion-induced diagnostic inaccuracies. Materials and methods Study Design and Participants A prospective, single-center study enrolled 110 adults (mean age: 52 ± 14 years; 58 male, comprising 27 cases of intracranial space-occupying lesions) referred for brain MRI at GUIQIAN International General Hospital. The exclusion criteria were as follows: (1) contraindications to MRI; (2) history of prior neurosurgery. Written informed consent was obtained from all participants. Data Acquisition All scans were performed on a 3T scanner (uMR880, United Imaging Healthcare, Shanghai, PR China) using a 48-channel head coil. The scanning sequences included brain ACS-T2WI sequence and the routine T2WI sequence without any acceleration technology, respectively. Following T2WI-ACS acquisition, the DR technique (uAIFI-DeepRecon, United Imaging Healthcare) was applied to generate DR-ACS-T2WI sequences through offline post-processing of the raw ACS data on the magnetic resonance host. The following parameters were used for all sequences: field of view (FOV) = 240×220 mm 2 ; resolution = 400×320; TR/TE=5100/115.64 ms. The other detailed scanning parameters of different sequences are shown in Table 1. Qualitative image analysis Two experienced radiologists (radiologists 1, 6 years of experience, and radiologists 2, 7 years of experience) independently assessed the overall image quality (OIQ), motion artifact, and diagnostic confidence (DC) of the three groups at a post-processing workstation (uOmnispace.MR, United Imaging Healthcare Co., Ltd., Shanghai, China). These qualitative indicators were assessed based on a five-point Likert scale as follows: 1 = severe, 2 = moderate, 3 = mild, 4 = fine, 5 = excellent. The objective evaluation included the SNR for gray matter and white matter, the CNR between gray and white matter, and the scan time. Finally, the formulas used to calculate the SNR and CNR were as follows: where is the meaning of signal Intensity of white matter, gray matter or cerebrospinal fluid , and is the variance of background. Statistical Analysis Statistical analysis was performed using SPSS 26.0 software (IBM SPSS Statistics). The Shapiro-Wilk test was employed to determine if the data followed a normal distribution, for continuous data with a normal distribution are reported as the means ± standard deviation (SD), whereas non-normally distributed data are expressed as the median and interquartile range (IQR). For subjective scores and objective indicators among the three groups, multiple comparisons were performed using the analysis of variance (ANOVA). Interobserver agreement was assessed using Cohen's kappa test. The interpretation criteria for kappa coefficients followed these thresholds: <0.20 (poor); 0.20-0.39 (fair); 0.40-059 (moderate); 0.60-0.79 (substantial), and 0.80-1.00 (almost perfect agreement), with P<0.05 considered statistically significant. Results The scan time of Groups DR-ACS-T2WI and ACS-T2WI (both 25.7s) was 57.87% shorter than that of Group the routine T2WI (61s). In the qualitative evaluation for image quality, Table 2 displays the rating results from the two radiologists. DR-ACS-T2WI achieved the highest ratings across all categories (OIQ, motion artifact, and DC), with scores of 4.90±0.30 , 4.91±0.29 , 4.92±0.28 (Reader 1) and 4.90±0.30 , 4.91±0.28 , 4.91±0.30 (Reader 2). ACS-T2WI ranked second, with scores of 4.83±0.39, 4.81±0.40 , 4.80±0.41 (Reader 1) and 4.82±0.39, 4.80±0.40 , 4.79±0.42 (Reader 2). Routine T2WI showed comparatively lower scores: 4.65±0.48, 4.50±0.50, 4.56±0.50 (Reader 1) and 4.64±0.49, 4.50±0.50, 4.56±0.51 (Reader 2). The two radiologists demonstrated substantial inter-rater agreement in subjective evaluations, with Kappa coefficients all greater than 0.6 (Table 3 and Table 4). Besides, DR-ACS-T2WI demonstrated superior SNRs in both white matter (65.06±12.1) and gray matter (97.25±18.52) compared to ACS-T2WI (47.62±8.65 and 71.54±12.05, respectively) and routine T2WI (34.32±6.51 and 51.92±8.62, respectively; P < 0.001 for all). Similarly, the gray-white matter CNR of DR-ACS-T2WI (32.93±12.35) was significantly higher than that of ACS-T2WI (24.29±9.08) and routine T2WI (17.31±6.01; P < 0.001) (Table 4 and figure 4). Figs. 1-3 show comparisons of images obtained by routine T2WI, ACS-T2WI, and DR-ACS-T2WI. In figure 1, routine T2WI sequences exhibit prominent Gibbs artifacts (green arrows) around the cerebral cortex. In contrast, images utilizing AI-assisted compressed sensing techniques and deep-learning reconstruction algorithms show better artifact suppression while maintaining anatomical integrity and spatial resolution. Figure 2 shows images from a 77-year-old female with cerebral infarction. Routine T2WI shows partial signal loss in the left pontine lesion due to motion artifacts from prolonged scanning, which may obscure diagnostic accuracy. Conversely, both ACS-T2WI and DR-ACS-T2WI preserve homogeneous and intact signal intensity within the lesion. Notably, DR-ACS-T2WI further enhances delineation with sharper lesion margins. Figure 3 presents brain images from a tumor patient, illustrating the limitations of routine T2WI and ACS-T2WI in tumor margin definition, where both sequences exhibit ill-defined boundaries and noise. In the pontine lesion of routine T2WI, partial signal loss is observed, while the signal loss in ACS-T2WI is moderately improved. However, the tumor margins remain relatively blurred. By contrast, DR-ACS-T2WI demonstrates complete tumor signal and clear margins, achieving the best performance. Discussion This present study evaluated the clinical feasibility of DR-ACS for brain T2WI, focusing on subjective image quality (overall image quality, motion artifact, and diagnostic confidence, with a 5-point scale) and objective metrics (SNR, CNR, and scan time), compared with the conventional PI-accelerated T2WI. The results revealed that the scan time of DR-ACS-T2WI and ACS-T2WI (25.7s) was 57.87% shorter than that of the routine T2WI (61s). The overall image quality, motion artifact, and diagnostic confidence scores of DR-ACS-T2WI significantly higher than those of ACS-T2WI and routine T2WI. DR-ACS-T2WI demonstrated superior SNRs in both white matter and gray matter compared to ACS-T2WIand routine T2WI. Similarly, the gray-white matter CNR of DR-ACS-T2WI was significantly higher than that of ACS-T2WI and routine T2WI. In clinical diagnosis, T2WI is an indispensable component of MRI examinations. It plays a crucial and pivotal role in both visualizing the structural manifestations of brain diseases and facilitating subsequent functional diagnosis. By highlighting differences in tissue relaxation times, T2WI enables radiologists to identify various pathological changes, such as cerebral infarcts, tumors, and demyelinating lesions, with high sensitivity. However, the relatively long scan durations associated with T2WI remain a persistent and significant challenge in brain MRI. Prolonged scanning not only increases patient discomfort and the risk of motion artifacts but also reduces the overall efficiency of the imaging workflow. In this study, DR-ACS-T2WI achieved a 57.87% reduction in scan time compared with routine T2WI, enabling the acquisition of T2 structural images in only 25.7 seconds. The 57.87% reduction in scan time aligns with prior musculoskeletal and abdominal studies. 23 , 24 However, what sets this study apart is the significant improvement in both the SNR and CNR compared to the routine T2WI. These enhancements surpasses earlier reports, 25 , 26 indicating that DR-ACS not only speeds up the scanning process but also elevates the overall quality of the acquired images. This is of paramount importance in brain imaging, where subtle differences in tissue contrast can be crucial for accurate diagnosis. For the qualitative assessment of T2WI imaging, noteworthy improvements on the mean gray matter SNR values and white matter SNR values when employing DR-ACS as compared to the ACS and routine T2WI were observed(Figure 4). When compared to conventional T2WI, the gray matter SNR, white matter SNR, and gray-white matter CNR of ACS-T2WI demonstrated approximately 40% improvements, whereas those of DR-ACS-T2WI showed roughly 90% enhancements across all metrics. These findings indicate that both DR and ACS can significantly improve image quality to varying degrees. Notably, the combined application of DR and ACS proposed in this study achieves an additional improvement in image quality beyond the baseline performance of ACS alone. The notable enhancement in DR-ACS-T2WI is primarily driven by the residual U-Net architecture embedded within the framework. This architecture effectively mitigated the noise amplification inherent to CS, preserving fine anatomical details such as cortical ribbon integrity and perivascular spaces (Figure 3). Maintaining these details is essential for a comprehensive understanding of brain anatomy and the accurate identification of potential pathologies. For instance, a recent study demonstrated that combining compressed sensing with deep learning reconstruction (DR) denoising improves knee image quality compared to compressed sensing alone 28 , which aligns with our findings. In the subjective evaluations conducted by two radiologists, DR-ACS-T2WI demonstrated superior performance across all metrics, with quantitative scores reinforcing its diagnostic advantage. Specifically, its image quality score (4.90±0.30 for both readers) surpassed both ACS-T2WI and conventional T2WI, aligning with objective assessment results. Meanwhile, DR-ACS-T2WI achieved superior artifact suppression scores (4.91±0.29 for Reader1 and 4.91±0.28 for Reader2), a critical advantage in neuroimaging where motion artifacts disproportionately degrade visualization of delicate brain structures such as the cortical ribbon and perivascular spaces 27 . This performance is visually corroborated by Figure 1 and Figure 2, which demonstrate marked reduction in susceptibility-induced artifacts within DR-ACS-T2WI compared to baseline modalities. The enhanced artifact suppression directly contributes to improved visualization of anatomical details, which is essential for accurate neurological diagnosis. The diagnostic confidence score of DR-ACS-T2WI reached 4.92±0.28 (Reader 1) and 4.91±0.30 (Reader 2), significantly higher than comparator sequences. This superiority is visually corroborated by Figure 2 and Figure 3, which demonstrate marked reduction in perilesional artifacts and enhanced boundary definition in DR-ACS-T2WI. Such improvements enable clear visualization of subcentimeter lesions previously obscured by motion artifacts in routine T2WI, a finding consistent with liver MRI studies, where DR-based sequences improved lesion detectability in uncooperative patients. 26 The enhanced anatomical clarity directly supports more reliable diagnostic confidence, underscoring the modality’s clinical utility in neurological applications. Despite these promising results, the study has several limitations. The data in this study were derived from a single-center source, which may not be representative of the broader population. Therefore, future research should involve multi-center trials, allowing for the validation of these findings across diverse populations, different MRI platforms, more different MRI sequences, and various clinical settings. 29 Secondly, the current investigation omitted a granular examination of individual intracranial lesion subtypes. Given that pathological diversity across lesion categories could introduce confounding variables affecting result accuracy, subsequent studies should prioritize particular neurological disorders to establish validated protocols for targeted clinical applications. Future investigations should focus on certain specific diseases to explore more refined application scenarios. This will help to ensure the generalizability and reliability of the DR-ACS technique in routine clinical practice. Conclusion Our study demonstrated that brain T2WI using the ACS combined with DR technology could achieve ultra-fast, high-quality imaging with enhanced diagnostic accuracy. By reducing motion artifacts and scan time, this technology holds promise for improving patient throughput and clinical outcomes. Abbreviations ACS AI-assisted compressed sensing DR Deep-learning reconstruction CS Compressed sensing T2WI T2-weighted imaging SD Standard deviation AI Artificial intelligence CNR Contrast-to-noise-ratio SNR Signal-to-noise-ratio ROI Region of interest MRI Magnetic resonance imaging FOV Field of view OIQ Overall image quality DC Diagnostic confidence IQR Interquartile range Declarations Ethical Statement: Informed consent was obtained from all individual participants included in the study.All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.This study was approved by the Ethics Committee/Institutional Review Board of Guiqian International General Hospital. Clinical Trial Registration: This study is a non-interventional diagnostic accuracy study evaluating an accelerated MRI technique. No clinical trial registration is required as it does not involve patient interventions, treatment allocation, or alterations to clinical care. As requested. Clinical trial number: not applicable. Acknowledgements Not applicable. Author contributions Author A: Conceptualized and designed the study; performed data curation, measurement, and formal analysis; wrote the original manuscript draft; reviewed and edited the manuscript; created figures and tables. Author B: Contributed to study design; performed data curation and measurement; reviewed and edited the manuscript; optimized figures and tables. Author C: Provided scientific oversight for study feasibility; critically reviewed the manuscript for intellectual content and provided revision suggestions; provided technical support. Author D: Critically reviewed the manuscript for intellectual content and provided revision suggestions; provided technical support. Author E: Performed data collection and measurement. Author F Performed data collection and measurement. Author G: Performed data collection and measurement. All authors reviewed and approved the final manuscript. All authors contributed significantly to this work. Funding Declaration: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data availability The data is available from the autors upon reasonable request. Ethical approval and consent to participate Informed consent and consent for publication was obtained from all participants. This study was conducted in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its’ later amendments. Consent for publication Consent for publication was given by the local ethic’s review board. All patients included in this study consented in the publication of the data. Competing interests The authors declare no competing interests. References Nabizadeh, Fardin et al. “T1 and T2 weighted lesions and cognition in multiple Sclerosis: A systematic review and meta-analysis.” Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia vol. 119 (2024): 1-7. doi:10.1016/j.jocn.2023.11.014 Deshmane A, Gulani V, Griswold MA, et al. Journal of magnetic resonance imaging: JMRI 2012 Jul;36 (1):55-72 doi:10.1002/jmri.23639 Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imagingMagnetic resonance in medicine 2007 Dec;58 (6):1182-95 doi:10.1002/mrm.21391 Wenqi Huang, Ziwen Ke, Zhuo-Xu Cui, et al. Deep low-Rank plus sparse network for dynamic MR imagingMedical image analysis 2021 10;73:102190 doi:10.1016/j.media.2021.102190 Shekhar S Chandra, Marlon Bran Lorenzana, Xinwen Liu, et al. Deep learning in magnetic resonance image reconstructionJournal of medical imaging and radiation oncology 2021 Aug;65 (5):564-577 doi:10.1111/1754-9485.13276 Jing Liu, Wei Li, Ziyuan Li, et al. European radiology 2023 Jul;33 (7):4864-4874 doi:10.1007/s00330-023-09470-x Ming Ni, Miao He, Yuxin Yang, Xiaoyi Wen, et al. Application research of AI-assisted compressed sensing technology in MRI scanning of the knee joint: 3D-MRI perspectiveEuropean radiology 2024 May;34 (5):3046-3058 doi:10.1007/s00330-023-10368-x Qiang Zhao, Jiajia Xu, Yu Xin Yang, et al. AI-assisted accelerated MRI of the ankle: clinical practice assessmentEuropean radiology experimental 2023 10 20;7 (1):62 doi:10.1186/s41747-023-00374-5 Sven S Walter, Jan Vosshenrich, Tatiane Cantarelli Rodrigues, et al. Deep Learning Superresolution for Simultaneous Multislice Parallel Imaging-Accelerated Knee MRI Using Arthroscopy ValidationRadiology 2025 Jan;314 (1): e241249 doi:10.1148/radiol.241249 Ruo-Fan Sheng, Li-Yun Zheng, Kai-Pu Jin, et al. Single-breath-hold T2WI liver MRI with deep learning-based reconstruction: A clinical feasibility study in comparison to conventional multi-breath-hold T2WI liver MRIMagnetic resonance imaging 2021 09;81:75-81 doi:10.1016/j.mri.2021.06.014 Yanjie Zhao, Chengdong Peng, Shaofang Wang, et al. The feasibility investigation of AI -assisted compressed sensing in kidney MR imaging: an ultra-fast T2WI imaging technologyBMC medical imaging 2022 07 04;22 (1):119 doi:10.1186/s12880-022-00842-1 Xianghu Yan, Yi Luo, Xiao Chen, et al. From Compressed-Sensing to Deep Learning MR: Comparative Biventricular Cardiac Function Analysis in a Patient CohortJournal of magnetic resonance imaging: JMRI 2024 04;59 (4):1231-1241 doi:10.1002/jmri.28899 Kang-Lung Lee, Dimitri A Kessler, Simon Dezonie, et al. Assessment of deep learning-based reconstruction on T2-weighted and diffusion-weighted prostate MRI image qualityEuropean journal of radiology 2023 Sep;166:111017 doi:10.1016/j.ejrad.2023.111017 Sung Hwan Bae, Jiyoung Hwang, Seong Sook Hong, et al. Clinical feasibility of accelerated diffusion weighted imaging of the abdomen with deep learning reconstruction: Comparison with conventional diffusion weighted imagingEuropean journal of radiology 2022 Sep;154:110428 doi:10.1016/j.ejrad.2022.110428 Daniel Wessling, Sebastian Gassenmaier, Susann-Cathrin Olthof, et al. Novel deep-learning-based diffusion weighted imaging sequence in 1.5 T breast MRIEuropean journal of radiology 2023 Sep;166:110948 doi:10.1016/j.ejrad.2023.110948 Jeong Woo Kim, Bit Na Park, Dominik Nickel, Mun Young Paek, Chang Hee Lee; Clinical feasibility of deep learning-accelerated single-shot turbo spin echo sequence with enhanced denoising for pancreas MRI at 3 TeslaEuropean journal of radiology 2024 Dec;181:111737 doi:10.1016/j.ejrad.2024.111737 Kai Liu, Bin Xi, Haitao Sun, et al. The clinical feasibility of artificial intelligence-assisted compressed sensing single-shot fluid-attenuated inversion recovery (ACS-SS-FLAIR) for evaluation of uncooperative patients with brain diseases: comparison with the conventional T2-FLAIR with parallel imagingActa radiologica (Stockholm, Sweden: 1987) 2023 May;64 (5):1943-1949 doi:10.1177/02841851221139125 Qizheng Wang, Weili Zhao, Xiaoying Xing, et al. Feasibility of AI-assisted compressed sensing protocols in knee MR imaging: a prospective multi-reader studyEuropean radiology 2023 Dec;33 (12):8585-8596 doi:10.1007/s00330-023-09823-6 Ming Ni, Miao He, Yuxin Yang, et al. Application research of AI-assisted compressed sensing technology in MRI scanning of the knee joint: 3D-MRI perspectiveEuropean radiology 2024 May;34 (5):3046-3058 doi:10.1007/s00330-023-10368-x Yangsean Choi, Ji Su Ko, Ji Eun Park, et al. Beyond the Conventional Structural MRI: Clinical Application of Deep Learning Image Reconstruction and Synthetic MRI of the BrainInvestigative radiology 2025 Jan 01;60 (1):27-42 doi:10.1097/RLI.0000000000001114 Kai Liu, Haitao Sun, Xingxing Wang, et al. Feasibility of the application of deep learning-reconstructed ultra-fast respiratory-triggered T2-weighted imaging at 3 T in liver imagingMagnetic resonance imaging 2024 Jun;109:27-33 doi:10.1016/j.mri.2024.03.001 Kai Liu, Haitao Sun, Xingxing Wang, et al. Feasibility of the application of deep learning-reconstructed ultra-fast respiratory-triggered T2-weighted imaging at 3 T in liver imagingMagnetic resonance imaging 2024 Jun;109:27-33 doi:10.1016/j.mri.2024.03.001 Giovanni Foti, Chiara Longo; Deep learning and AI in reducing magnetic resonance imaging scanning time: advantages and pitfalls in clinical practicePolish journal of radiology 2024;89: e443-e451 doi:10.5114/pjr/192822 Jan Vosshenrich, Mary Bruno, Tatiane Cantarelli Rodrigues, et al. Arthroscopy-validated Diagnostic Performance of 7-Minute Five-Sequence Deep Learning Super-Resolution 3-T Shoulder MRIRadiology 2025 Feb;314 (2): e241351 doi:10.1148/radiol.241351 Sven S Walter, Jan Vosshenrich, Tatiane Cantarelli Rodrigues, et al. Deep Learning Superresolution for Simultaneous Multislice Parallel Imaging-Accelerated Knee MRI Using Arthroscopy ValidationRadiology 2025 Jan;314 (1): e241249 doi:10.1148/radiol.241249 Jan M Brendel., Reza Dehdab., Judith Herrmann., et al. Deep learning reconstruction for accelerated 3-D magnetic resonance cholangiopancreatographyRadiol Med 2025; doi:10.1007/s11547-025-01987-z Ann-Christin Klemenz, Linda Reichardt, Margarita Gorodezky, et al. Accelerated Cardiac MRI with Deep Learning-based Image Reconstruction for Cine ImagingRadiology. Cardiothoracic imaging 2024 Dec;6 (6): e230419 doi:10.1148/ryct.230419 H Akai, K Yasaka, H Sugawara, et al. Faster acquisition of magnetic resonance imaging sequences of the knee via deep learning reconstruction: a volunteer studyClinical radiology 2024 06;79 (6):453-459 doi:10.1016/j.crad.2024.03.002 T Tajima, H Akai, K Yasaka, et al. Usefulness of deep learning-based noise reduction for 1.5 T MRI brain imagesClinical radiology 2023 01;78 (1): e13-e21 doi:10.1016/j.crad.2022.08.127 Tables Table 1. Acquisition parameters of the brain MRI sequences. Parameters DR-ACS-T2WI ACS-T2WI Routine T2WI TR (ms) 5100 5100 5100 TE (ms) 115 115 115 Slice thickness (mm) 5 5 5 Slice gap (mm) 1.5 1.5 1.5 FOV (mm 2 ) 240×200 240×200 240×200 Matrix 400×320 400×320 400×320 Slice 21 21 21 AI-assisted compressed Yes Yes No Deep-learning reconstructed Yes No No Scan time(s) 25.7 25.7 61 Table 2. Comparison of the rating results of images acquired with DR-ACS-T2WI, ACS-T2WI and routine T2WI. Parameters Reader Reader 1 Reader 2 DR-ACS-T2WI ACS-T2WI Routine T2WI DR-ACS-T2WI ACS-T2WI Routine T2WI Overall image quality 4.90±0.30（4,5） 4.83±0.39（4,5） 4.65±0.48（3,5） 4.90±0.30（4,5） 4.82±0.39（4,5） 4.64±0.49（3,5） Motion artifact 4.91±0.29（4,5） 4.81±0.40（4,5） 4.50±0.50（3,5） 4.91±0.28（4,5） 4.80±0.40（4,5） 4.50±0.50（3,5） Diagnostic confidence 4.92±0.28（4,5） 4.80±0.41（4,5） 4.56±0.50（3,5） 4.91±0.30（4,5） 4.79±0.42（4,5） 4.56±0.51（3,5） Table 3. Inter-rater Consistency Analysis (Kappa Values) of multi-sequence images by two radiologists. DR-ACS-T2WI ACS-T2WI Routine T2WI kappa 95%CI P kappa 95%CI P kappa 95%CI P Overall image quality 0.85 0.783-0.917 ＜0.001 0.72 0.636-0.804 ＜0.001 0.63 0.540-0.720 ＜0.001 Motion artifact 0.78 0.703-0.857 ＜0.001 0.65 0.561-0.739 ＜0.001 0.65 0.561, 0.739 ＜0.001 Diagnostic confidence 0.82 0.748-0.892 ＜0.001 0.71 0.625-0.795 ＜0.001 0.62 0.529, 0.711 ＜0.001 Table 4. The gray matter SNR(SNR GM ), white matter SNR(SNR WM ), and gray-white matter CNR(CNR WM/GM ) in Group DR-ACS-T2WI, ACS-T2WI and routine T2WI. N SNR WM SNR GM CNR WM/GM DR-ACS-T2WI 110 65.06±12.1 97.25±18.52 32.93±12.35 ACS-T2WI 110 47.62±8.65 71.54±12.05 24.29±9.08 Routine T2WI 110 34.32±6.51 51.92±8.62 17.31±6.01 t DR-ACS-T2WI vs. ACS-T2WI 12.289 12.202 5.909 P ＜0.001 ＜0.001 ＜0.001 t DR-ACS-T2WI vs. Routine T2WI 23.452 23.27 11.924 P ＜0.001 ＜0.001 ＜0.001 t ACS-T2WI vs. Routine T2WI 12.879 13.894 6.727 P ＜0.001 ＜0.001 ＜0.001 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 06 Mar, 2026 Reviews received at journal 08 Dec, 2025 Reviews received at journal 30 Nov, 2025 Reviewers agreed at journal 26 Nov, 2025 Reviewers agreed at journal 22 Nov, 2025 Reviewers invited by journal 08 Sep, 2025 Editor invited by journal 05 Sep, 2025 Editor assigned by journal 11 Aug, 2025 Submission checks completed at journal 08 Aug, 2025 First submitted to journal 08 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7239920","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":513227232,"identity":"89ded48a-e299-4f79-8859-02a55148284e","order_by":0,"name":"Shiwei Lai","email":"","orcid":"","institution":"Guiqian International Hospital","correspondingAuthor":false,"prefix":"","firstName":"Shiwei","middleName":"","lastName":"Lai","suffix":""},{"id":513227233,"identity":"699bcb84-a8cc-41b1-a06a-5f7289fa5e93","order_by":1,"name":"Yong Xiang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxElEQVRIiWNgGAWjYBACNv7mgw8+VNgwyzMzHyBOC5/EsWTDGWfS2A3b2xKI0yLHkKMmzNt2mJ/hzBkDIh3GcIaNccYZZmnGGTkfb7xhsJPTbSCkhbn3GNAvbMbsErmbLecwJBubHSBoy7l0oF94khln5G6T5mE4kLiNsJYcM2neNon6hhs5z0jSYsAM9D4bkVoggZzADAxkY8s5BkT4Rb4fHJX/QVH58MabCjs5glpQgAQPkVGDrIVUHaNgFIyCUTAiAABwLEK8G+euwgAAAABJRU5ErkJggg==","orcid":"","institution":"Guiqian International Hospital","correspondingAuthor":true,"prefix":"","firstName":"Yong","middleName":"","lastName":"Xiang","suffix":""},{"id":513227234,"identity":"a2fa07cb-a8a9-475e-8c35-04d119cd1166","order_by":2,"name":"Shaoxin Xiang","email":"","orcid":"","institution":"MR Scientific Collaboration，United Imaging Healthcare","correspondingAuthor":false,"prefix":"","firstName":"Shaoxin","middleName":"","lastName":"Xiang","suffix":""},{"id":513227235,"identity":"c6dbb13a-8ba1-4464-adab-66ca08d8901a","order_by":3,"name":"Jiahuan Dai","email":"","orcid":"","institution":"MR Scientific Collaboration，Beijing United Imaging Research Institute of Intelligent Imaging","correspondingAuthor":false,"prefix":"","firstName":"Jiahuan","middleName":"","lastName":"Dai","suffix":""},{"id":513227236,"identity":"4eaebf07-235c-404e-94fe-24d5c99a5b86","order_by":4,"name":"Zuonian Hu","email":"","orcid":"","institution":"Guiqian International Hospital","correspondingAuthor":false,"prefix":"","firstName":"Zuonian","middleName":"","lastName":"Hu","suffix":""},{"id":513227237,"identity":"522ba2a7-dc24-495e-9b68-5002cab964a4","order_by":5,"name":"Xiaquan Ning","email":"","orcid":"","institution":"Guiqian International Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xiaquan","middleName":"","lastName":"Ning","suffix":""},{"id":513227238,"identity":"607ef326-1b23-4dd3-aae5-061095d26473","order_by":6,"name":"Ju Yin","email":"","orcid":"","institution":"Guiqian International Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ju","middleName":"","lastName":"Yin","suffix":""}],"badges":[],"createdAt":"2025-07-29 06:54:25","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7239920/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7239920/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91817284,"identity":"c949f953-1e65-4bcc-9048-c6604cd7ff77","added_by":"auto","created_at":"2025-09-22 06:54:41","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1378271,"visible":true,"origin":"","legend":"","description":"","filename":"EfficacyEvaluationofAIAssistedCompressedSensingCombinedwithDeepLearningReconstructioninAcceleratingBrainT2WeightedImaging1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/83c84d0217f3e7e4e2a346e6.docx"},{"id":91817053,"identity":"1c285dea-8274-4d1d-a16d-081a0f3e807e","added_by":"auto","created_at":"2025-09-22 06:53:24","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9198,"visible":true,"origin":"","legend":"","description":"","filename":"36d7d89ede0c46c38d1b689efa1c3664.json","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/448cf5fbc1b83fa655f3ec3c.json"},{"id":91816953,"identity":"a536cd8c-473a-48c1-9f54-51f7ffef94e9","added_by":"auto","created_at":"2025-09-22 06:53:05","extension":"jpeg","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":139119,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/f8c703147c121fd50d26a65b.jpeg"},{"id":91816945,"identity":"bdb9e968-3f9d-4010-b9b4-0faebb38e993","added_by":"auto","created_at":"2025-09-22 06:53:04","extension":"jpeg","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":161476,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/21ac8ae3ab337f7d8b3c06ed.jpeg"},{"id":91816622,"identity":"dafaba49-bc2a-4291-9c48-d85ab51e785f","added_by":"auto","created_at":"2025-09-22 06:52:15","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":174414,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/a5786efb06430a2706161258.png"},{"id":91816975,"identity":"99b7730b-3446-4fb4-8a16-7fa779c7f327","added_by":"auto","created_at":"2025-09-22 06:53:09","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":104189,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/ac661512c0c0316866af52cb.png"},{"id":91816916,"identity":"533cdbf1-eb02-4c6f-8e53-f7d99fdd75f9","added_by":"auto","created_at":"2025-09-22 06:52:58","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":109281,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/44e085b419c0cd91bd3ce968.png"},{"id":91489257,"identity":"7bb26d8d-f28a-40b7-b021-1491996790ec","added_by":"auto","created_at":"2025-09-17 05:13:08","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":494858,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of images obtained by routine T2WI (a1), ACS-T2WI (b1) and DR-ACS-T2WI (c1) sequences. Magnified images of the right prefrontal lobe, including routine T2WI (a2), ACS-T2WI (b2), and DR-ACS-T2WI (c2). Gibbs artifacts (indicated by green arrows).\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/1bbc5207447d7eb695ad4b11.png"},{"id":91489264,"identity":"06cd77fe-11c2-46df-b6e2-6f653b37170b","added_by":"auto","created_at":"2025-09-17 05:13:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":485254,"visible":true,"origin":"","legend":"\u003cp\u003eAxial images at the pontine level in a case of cerebral infarction demonstrating routine T2WI (a1), ACS-T2WI (b1), and DR-ACS-T2WI (c1), with corresponding magnified views of abnormal signals (a2, b2, c2), show a left pontine lesion (indicated by green arrows).\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/789cf0acbe3226e9036c5ace.png"},{"id":91489261,"identity":"a03bfb6a-28f4-4c06-b472-028fd29e315f","added_by":"auto","created_at":"2025-09-17 05:13:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":508147,"visible":true,"origin":"","legend":"\u003cp\u003eTemporal lobe imaging in a tumor patient demonstrating routine T2WI (a1), ACS-T2WI (b1), and DR-ACS-T2WI (c1), with corresponding magnified views of the tumor region (a2, b2, c2) (indicated by green arrows).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/73cf985d9081b587a92190fa.png"},{"id":91816907,"identity":"1e1ed05d-5fba-4429-b2ef-fd60299a2a0b","added_by":"auto","created_at":"2025-09-22 06:52:57","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":95622,"visible":true,"origin":"","legend":"\u003cp\u003eSNR, and CNR measurements of different anatomical structures under DR-ACS, ACS, and routine T2WI. Statistically different pairs are marked with star, *** means p\u0026lt;0.01. ACS, artificial intelligence–assisted compressed sensing; DR, DeepRecon; SNR, signal-to-noise ratio; CNR, contrast-to-noise ratio.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/c5d1875e1d557cbe75690b73.png"},{"id":91818022,"identity":"0adae33b-c0bd-46ea-85ba-363f5fe808bc","added_by":"auto","created_at":"2025-09-22 07:01:19","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2348566,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7239920/v1/dd631aae-93c2-4267-8c4e-7fd7a333a1b1.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eEfficacy Evaluation of AI-Assisted Compressed Sensing Combined with\u003c/strong\u003e \u003cstrong\u003eDeep-Learning Reconstruction in Accelerating Brain T2-Weighted Imaging: A Clinical Feasibility Study\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMagnetic resonance imaging (MRI) is indispensable for neurological diagnostics. T2-weighted sequences remain the clinical cornerstone for detecting fluid-associated pathologies, derived from their enhanced tissue contrast resolution in hydrophilic environments and heightened sensitivity to proton mobility alterations. By highlighting differences in tissue relaxation times, T2WI enables radiologists to identify various pathological changes, such as cerebral infarcts, tumors, and demyelinating lesions, with high sensitivity.\u003csup\u003e1\u003c/sup\u003e Yet the relatively long scan times remain a significant challenge, particularly for uncooperative patients. Traditional acceleration techniques, such as parallel imaging\u003csup\u003e2\u003c/sup\u003e and compressed sensing (CS)\u003csup\u003e3\u003c/sup\u003e, have been developed to address this issue. However, these methods often entail a compromise between acceleration efficiency and image quality. Specifically, the primary drawback of parallel imaging techniques is the reduction in the acquisition of phase encoding lines, which results in decreased signal-to-noise ratio (SNR) and increased noise in the image. Similarly, simply applying compressed sensing techniques for image acquisition also leads to degradation in image quality. The application of these traditional acceleration techniques frequently results in a decrement of SNR or spatial resolution, limiting diagnostic confidence.\u003csup\u003e4\u003c/sup\u003e However, the brain harbors highly intricate structures, including neural nuclei, cerebral vasculature, and cranial nerves. Given the minimal signal contrast between various brain tissues and pathological entities, high - resolution MRI with an elevated SNR and contrast-to-noise ratio (CNR) is essential for clear visualization.\u003c/p\u003e\n\u003cp\u003eThe integration of artificial intelligence (AI) into MRI workflows has catalyzed paradigm shifts across acquisition and reconstruction pipelines, redefining clinical imaging capabilities\u003csup\u003e5\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e6\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e7\u003c/sup\u003e.\u0026nbsp;In the field of image acquisition, AI-driven compressed sensing (ACS) emerges from multidimensional integration of parallel imaging, compressed sensing, half-Fourier sampling, and deep neural architectures, achieving scan time reductions without compromising diagnostic precision.\u003csup\u003e8\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e9\u003c/sup\u003e ACS have achieved substantial scan time reductions of over 50% in liver and cardiac, thereby optimizing clinical workflow efficiency and reduce the impact of breathing movements on image quality.\u003csup\u003e10\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e11\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e12\u003c/sup\u003eConcurrently, deep learning reconstruction (DR)\u0026nbsp;technique\u0026nbsp;synchronize k-space manipulation with image enhancement through adaptive noise-signal discrimination, optimizing both signal-to-noise ratio and contrast-to-noise ratio metrics. Deep learning reconstruction can be effectively applied to diffusion-weighted imaging (DWI) across multiple anatomical regions, demonstrating capabilities in mitigating geometric distortion and reducing scan time.\u003csup\u003e13\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e14\u003c/sup\u003eFurthermore, this technique enables optimal image quality enhancement in MRI systems with different magnetic field strengths, including 1.5T, 3T, etc.\u0026nbsp;\u003csup\u003e15\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e16\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eThese parallel advancements demonstrate how intelligent algorithm design can overcome traditional MRI limitations through data-driven information retrieval and computational signal processing. While extant literature substantiates the neuroimaging efficacy of ACS and DR as standalone modalities, both parallel imaging and compressed sensing techniques result in the loss of k-space information while accelerating the scanning process, leading to degradation of image quality.\u003csup\u003e17\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e18\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e19\u003c/sup\u003e Critical gaps persist in delineating their synergistic effects when implemented within dual-modality AI frameworks, highlighting the need for a systematic evaluation of integrating ACS and DR in brain MRI.\u003csup\u003e20\u003c/sup\u003e Therefore, integrating ACS-accelerated and deep learning technology enables not only accelerated imaging but also enhances overall image quality.\u003csup\u003e21\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eConsequently, this study investigates to assess the efficacy of integrating ACS and DR techniques in transcending the inherent limitations of conventional T2-weighted imaging (T2WI). Specifically, the study aims to determine if this integrated approach can achieve significant reductions in scan duration, concomitant enhancements in SNR and CNR, and substantial minimization of motion-induced diagnostic inaccuracies.\u0026nbsp;\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003e\u003cstrong\u003eStudy Design and Participants\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA prospective, single-center study enrolled 110 adults (mean age: 52 \u0026plusmn; 14 years; 58 male, comprising 27 cases of intracranial space-occupying lesions) referred for brain MRI at GUIQIAN International General Hospital. The exclusion criteria were as follows: (1) contraindications to MRI; (2) history of prior neurosurgery. Written informed consent was obtained from all participants. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Acquisition\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll scans were performed on a 3T scanner (uMR880, United Imaging Healthcare, Shanghai, PR China) using a 48-channel head coil. The scanning sequences included brain ACS-T2WI sequence and the routine T2WI sequence without any acceleration technology, respectively. Following T2WI-ACS acquisition, the DR technique (uAIFI-DeepRecon, United Imaging Healthcare) was applied to generate DR-ACS-T2WI sequences through offline post-processing of the raw ACS data on the magnetic resonance host.\u0026nbsp;The following parameters were used for all sequences: field of view (FOV) = 240\u0026times;220 mm\u003csup\u003e2\u003c/sup\u003e; resolution = 400\u0026times;320; TR/TE=5100/115.64 ms. The other detailed scanning parameters of different sequences are shown in Table 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQualitative image analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTwo experienced radiologists (radiologists 1, 6 years of experience, and radiologists 2, 7 years of experience) independently assessed the overall image quality (OIQ), motion artifact, and diagnostic confidence (DC) of the three groups\u0026nbsp;at a post-processing workstation (uOmnispace.MR, United Imaging Healthcare Co., Ltd., Shanghai, China).\u0026nbsp;These qualitative indicators were assessed based on a five-point Likert scale as follows: 1 = severe, 2 = moderate, 3 = mild, 4 = fine, 5 = excellent.\u0026nbsp;The objective evaluation included the SNR for gray matter and white matter, the CNR between gray and white matter, and the scan time.\u0026nbsp;Finally, the formulas used\u0026nbsp;to calculate the SNR and CNR were as follows:\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u003cimg width=\"80\" height=\"49\" src=\"data:image/wmf;base64,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\" alt=\"image\"\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cimg width=\"257\" height=\"31\" src=\"data:image/wmf;base64,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\" alt=\"image\"\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003ewhere \u003cimg width=\"41\" height=\"17\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e\u0026nbsp;is the meaning of signal Intensity of white matter, gray matter or cerebrospinal fluid , and \u003cimg width=\"37\" height=\"17\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e\u0026nbsp;is the variance of background.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analysis was performed using SPSS 26.0 software (IBM SPSS Statistics). The Shapiro-Wilk test was employed to determine if the data followed a normal distribution, for continuous data with a normal distribution are reported as the means \u0026plusmn; standard deviation (SD), whereas non-normally distributed data are expressed as the median and interquartile range (IQR). For subjective scores and objective indicators among the three groups, multiple comparisons were performed using the analysis of variance (ANOVA). Interobserver agreement was assessed using Cohen\u0026apos;s kappa test. The interpretation criteria for kappa coefficients followed these thresholds: \u0026lt;0.20 (poor); 0.20-0.39 (fair); 0.40-059 (moderate); 0.60-0.79 (substantial), and 0.80-1.00 (almost perfect agreement), with P\u0026lt;0.05 considered statistically significant.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eThe scan time of Groups DR-ACS-T2WI and ACS-T2WI (both 25.7s) was 57.87% shorter than that of Group the routine T2WI (61s).\u003c/p\u003e\n\u003cp\u003eIn the qualitative evaluation for image quality, Table 2 displays the rating results from the two radiologists. DR-ACS-T2WI achieved the highest ratings across all categories (OIQ, motion artifact, and DC), with scores of 4.90±0.30 , 4.91±0.29\u0026nbsp;, 4.92±0.28\u0026nbsp;(Reader 1) and 4.90±0.30\u0026nbsp;, 4.91±0.28\u0026nbsp;, 4.91±0.30\u0026nbsp;(Reader 2). ACS-T2WI ranked second, with scores of 4.83±0.39, 4.81±0.40\u0026nbsp;, 4.80±0.41\u0026nbsp;(Reader 1) and 4.82±0.39, 4.80±0.40\u0026nbsp;, 4.79±0.42\u0026nbsp; \u0026nbsp;(Reader 2). Routine T2WI showed comparatively lower scores: 4.65±0.48, 4.50±0.50, 4.56±0.50\u0026nbsp;(Reader 1) and 4.64±0.49, 4.50±0.50, 4.56±0.51\u0026nbsp;(Reader 2). The two radiologists demonstrated substantial inter-rater agreement in subjective evaluations, with Kappa coefficients all greater than 0.6 (Table 3 and Table 4).\u003c/p\u003e\n\u003cp\u003eBesides, DR-ACS-T2WI demonstrated superior SNRs in both white matter (65.06±12.1) and gray matter (97.25±18.52) compared to ACS-T2WI (47.62±8.65 and 71.54±12.05, respectively) and routine T2WI (34.32±6.51 and 51.92±8.62, respectively; P \u0026lt; 0.001 for all). Similarly, the gray-white matter CNR of DR-ACS-T2WI (32.93±12.35) was significantly higher than that of ACS-T2WI (24.29±9.08) and routine T2WI (17.31±6.01; P \u0026lt; 0.001) (Table 4 and figure 4).\u003c/p\u003e\n\u003cp\u003eFigs. 1-3 show comparisons of images obtained by\u0026nbsp;routine T2WI, ACS-T2WI, and DR-ACS-T2WI. In figure 1, routine T2WI sequences exhibit prominent Gibbs artifacts (green arrows) around the cerebral cortex. In contrast, images utilizing AI-assisted compressed sensing techniques and deep-learning reconstruction algorithms show better artifact suppression while maintaining anatomical integrity and spatial resolution.\u003c/p\u003e\n\u003cp\u003eFigure 2 shows images from a 77-year-old female with cerebral infarction.\u0026nbsp;Routine T2WI shows partial signal loss in the left pontine lesion due to motion artifacts from prolonged scanning, which may obscure diagnostic accuracy. Conversely, both ACS-T2WI and DR-ACS-T2WI preserve homogeneous and intact signal intensity within the lesion. Notably, DR-ACS-T2WI further enhances delineation with sharper lesion margins.\u003c/p\u003e\n\u003cp\u003eFigure 3 presents brain images from a tumor patient, illustrating the limitations of routine T2WI and ACS-T2WI in tumor margin definition, where both sequences exhibit ill-defined boundaries and noise. In the pontine lesion of routine T2WI, partial signal loss is observed, while the signal loss in ACS-T2WI is moderately improved. However, the tumor margins remain relatively blurred. By contrast, DR-ACS-T2WI demonstrates complete tumor signal and clear margins, achieving the best performance.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis present study evaluated the clinical feasibility of DR-ACS for brain T2WI, focusing on subjective image quality (overall image quality, motion artifact, and diagnostic confidence, with a 5-point scale) and objective metrics (SNR, CNR, and scan time), compared with the conventional PI-accelerated T2WI. The results revealed that the scan time of DR-ACS-T2WI and ACS-T2WI (25.7s) was 57.87% shorter than that of the routine T2WI (61s). The overall image quality, motion artifact, and diagnostic confidence scores of DR-ACS-T2WI significantly higher than those of ACS-T2WI and routine T2WI. DR-ACS-T2WI demonstrated superior SNRs in both white matter and gray matter compared to ACS-T2WIand routine T2WI. Similarly, the gray-white matter CNR of DR-ACS-T2WI was significantly higher than that of ACS-T2WI and routine T2WI.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn clinical diagnosis, T2WI is an indispensable component of MRI examinations. It plays a crucial and pivotal role in both visualizing the structural manifestations of brain diseases and facilitating subsequent functional diagnosis. By highlighting differences in tissue relaxation times, T2WI enables radiologists to identify various pathological changes, such as cerebral infarcts, tumors, and demyelinating lesions, with high sensitivity.\u0026nbsp;However, the relatively long scan durations associated with T2WI remain a persistent and significant challenge in brain MRI. Prolonged scanning not only increases patient discomfort and the risk of motion artifacts but also reduces the overall efficiency of the imaging workflow.\u0026nbsp;In this study, DR-ACS-T2WI achieved a 57.87% reduction in scan time compared with routine T2WI, enabling the acquisition of T2 structural images in only 25.7 seconds. The 57.87% reduction in scan time aligns with prior musculoskeletal and abdominal studies.\u003csup\u003e23\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e24\u003c/sup\u003e However, what sets this study apart is the significant improvement in both the SNR and CNR compared to the routine T2WI. These enhancements surpasses earlier reports,\u003csup\u003e25\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e26\u003c/sup\u003e indicating that DR-ACS not only speeds up the scanning process but also elevates the overall quality of the acquired images. This is of paramount importance in brain imaging, where subtle differences in tissue contrast can be\u0026nbsp;crucial for accurate diagnosis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor the qualitative assessment of T2WI imaging, noteworthy improvements on the mean gray matter SNR values and white matter SNR values when employing DR-ACS as compared to the ACS and routine T2WI were observed(Figure 4). When compared to conventional T2WI, the gray matter SNR, white matter SNR, and gray-white matter CNR of ACS-T2WI demonstrated approximately 40% improvements, whereas those of DR-ACS-T2WI showed roughly 90% enhancements across all metrics. These findings indicate that both DR and ACS can significantly improve image quality to varying degrees. Notably, the combined application of DR and ACS proposed in this study achieves an additional improvement in image quality beyond the baseline performance of ACS alone. The notable enhancement in DR-ACS-T2WI is primarily driven by the residual U-Net architecture embedded within the framework. This architecture effectively mitigated the noise amplification inherent to CS, preserving fine anatomical details such as cortical ribbon integrity and perivascular spaces (Figure 3). Maintaining these details is essential for a comprehensive understanding of brain anatomy and the accurate identification of potential pathologies. For instance, a recent study demonstrated that combining compressed sensing with deep learning reconstruction\u0026nbsp;(DR) denoising improves knee image quality compared to compressed sensing alone\u003csup\u003e28\u003c/sup\u003e, which aligns with our findings.\u003c/p\u003e\n\u003cp\u003eIn the subjective evaluations conducted by two radiologists, DR-ACS-T2WI demonstrated superior performance across all metrics, with quantitative scores reinforcing its diagnostic advantage. Specifically, its image quality score (4.90±0.30 for both readers) surpassed both ACS-T2WI and conventional T2WI, aligning with objective assessment results. Meanwhile, DR-ACS-T2WI achieved superior artifact suppression scores (4.91±0.29 for Reader1 and 4.91±0.28 for Reader2), a critical advantage in neuroimaging where motion artifacts disproportionately degrade visualization of delicate brain structures such as the cortical ribbon and perivascular spaces\u003csup\u003e27\u003c/sup\u003e. This performance is visually corroborated by Figure 1 and Figure 2, which demonstrate marked reduction in susceptibility-induced artifacts within DR-ACS-T2WI compared to baseline modalities. The enhanced artifact suppression directly contributes to improved visualization of anatomical details, which is essential for accurate neurological diagnosis. The diagnostic confidence score of DR-ACS-T2WI reached 4.92±0.28 (Reader 1) and 4.91±0.30 (Reader 2), significantly higher than comparator sequences. This superiority is visually corroborated by Figure 2 and Figure 3, which demonstrate marked reduction in perilesional artifacts and enhanced boundary definition in DR-ACS-T2WI. Such improvements enable clear visualization of\u0026nbsp;subcentimeter lesions previously obscured by motion artifacts\u0026nbsp;in routine T2WI, a finding consistent with\u0026nbsp;liver MRI studies, where DR-based sequences improved lesion detectability in uncooperative patients.\u003csup\u003e26\u003c/sup\u003e The enhanced anatomical clarity directly supports more reliable diagnostic confidence, underscoring the modality’s clinical utility in neurological applications.\u003c/p\u003e\n\u003cp\u003eDespite these promising results, the study has several limitations. The\u0026nbsp;data in this study were derived from a single-center source, which may not be representative of the broader population. Therefore, future research should involve multi-center trials, allowing for the validation of these findings across diverse populations, different MRI platforms, more different MRI sequences, and various clinical settings.\u003csup\u003e29\u003c/sup\u003e Secondly, the current investigation omitted a granular examination of individual intracranial lesion subtypes. Given that pathological diversity across lesion categories could introduce confounding variables affecting result accuracy, subsequent studies should prioritize particular neurological disorders to establish validated protocols for targeted clinical applications. Future investigations should focus on certain specific diseases to explore more refined application scenarios. This will help to ensure the generalizability and reliability of the DR-ACS technique in routine clinical practice.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eOur study demonstrated that brain T2WI using the ACS combined with DR technology could achieve ultra-fast, high-quality imaging with enhanced diagnostic accuracy. By reducing motion artifacts and scan time, this technology holds promise for improving patient throughput and clinical outcomes. \u0026nbsp;\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eACS AI-assisted compressed sensing \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDR Deep-learning reconstruction \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCS Compressed sensing\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eT2WI T2-weighted imaging\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSD Standard deviation\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAI Artificial intelligence\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCNR Contrast-to-noise-ratio\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSNR Signal-to-noise-ratio\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eROI Region of interest\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMRI Magnetic resonance imaging\u003c/p\u003e\n\u003cp\u003eFOV Field of view\u003c/p\u003e\n\u003cp\u003eOIQ Overall image quality\u003c/p\u003e\n\u003cp\u003eDC Diagnostic confidence\u003c/p\u003e\n\u003cp\u003eIQR Interquartile range\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical Statement:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent was obtained from all individual participants included in the study.All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.This study was approved by the Ethics Committee/Institutional Review Board of\u0026nbsp;Guiqian International General Hospital.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical Trial Registration:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study is a non-interventional diagnostic accuracy study evaluating an accelerated MRI technique. No clinical trial registration is required as it does not involve patient interventions, treatment allocation, or alterations to clinical care. As requested.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number:\u0026nbsp;\u003c/strong\u003enot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthor A: Conceptualized and designed the study; performed data curation, measurement, and formal analysis; wrote the original manuscript draft; reviewed and edited the manuscript; created figures and tables.\u003c/p\u003e\n\u003cp\u003eAuthor B: Contributed to study design; performed data curation and measurement; reviewed and edited the manuscript; optimized figures and tables.\u003c/p\u003e\n\u003cp\u003eAuthor C: Provided scientific oversight for study feasibility; critically reviewed the manuscript for intellectual content and provided revision suggestions; provided technical support.\u003c/p\u003e\n\u003cp\u003eAuthor D: Critically reviewed the manuscript for intellectual content and provided revision suggestions; provided technical support.\u003c/p\u003e\n\u003cp\u003eAuthor E: Performed data collection and measurement.\u003c/p\u003e\n\u003cp\u003eAuthor F Performed data collection and measurement.\u003c/p\u003e\n\u003cp\u003eAuthor G: Performed data collection and measurement.\u003c/p\u003e\n\u003cp\u003eAll authors reviewed and approved the final manuscript. All authors contributed significantly to this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Declaration:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data is available from the autors upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent and consent for publication was obtained from all participants. This study was conducted in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its\u0026rsquo; later amendments.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConsent for publication was given by the local ethic\u0026rsquo;s review board. All patients included in this study consented in the publication of the data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNabizadeh, Fardin et al. \u0026ldquo;T1 and T2 weighted lesions and cognition in multiple Sclerosis: A systematic review and meta-analysis.\u0026rdquo; Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia vol. 119 (2024): 1-7. doi:10.1016/j.jocn.2023.11.014\u003c/li\u003e\n\u003cli\u003eDeshmane A, Gulani V, Griswold MA, et al. Journal of magnetic resonance imaging: JMRI 2012 Jul;36 (1):55-72 doi:10.1002/jmri.23639\u003c/li\u003e\n\u003cli\u003eLustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imagingMagnetic resonance in medicine 2007 Dec;58 (6):1182-95 doi:10.1002/mrm.21391\u003c/li\u003e\n\u003cli\u003eWenqi Huang, Ziwen Ke, Zhuo-Xu Cui, et al. Deep low-Rank plus sparse network for dynamic MR imagingMedical image analysis 2021 10;73:102190 doi:10.1016/j.media.2021.102190 \u003c/li\u003e\n\u003cli\u003eShekhar S Chandra, Marlon Bran Lorenzana, Xinwen Liu, et al. Deep learning in magnetic resonance image reconstructionJournal of medical imaging and radiation oncology 2021 Aug;65 (5):564-577 doi:10.1111/1754-9485.13276 \u003c/li\u003e\n\u003cli\u003eJing Liu, Wei Li, Ziyuan Li, et al. European radiology 2023 Jul;33 (7):4864-4874 doi:10.1007/s00330-023-09470-x\u003c/li\u003e\n\u003cli\u003eMing Ni, Miao He, Yuxin Yang, Xiaoyi Wen, et al. Application research of AI-assisted compressed sensing technology in MRI scanning of the knee joint: 3D-MRI perspectiveEuropean radiology 2024 May;34 (5):3046-3058 doi:10.1007/s00330-023-10368-x\u003c/li\u003e\n\u003cli\u003eQiang Zhao, Jiajia Xu, Yu Xin Yang, et al. AI-assisted accelerated MRI of the ankle: clinical practice assessmentEuropean radiology experimental 2023 10 20;7 (1):62 doi:10.1186/s41747-023-00374-5\u003c/li\u003e\n\u003cli\u003eSven S Walter, Jan Vosshenrich, Tatiane Cantarelli Rodrigues, et al. Deep Learning Superresolution for Simultaneous Multislice Parallel Imaging-Accelerated Knee MRI Using Arthroscopy ValidationRadiology 2025 Jan;314 (1): e241249 doi:10.1148/radiol.241249\u003c/li\u003e\n\u003cli\u003eRuo-Fan Sheng, Li-Yun Zheng, Kai-Pu Jin, et al. Single-breath-hold T2WI liver MRI with deep learning-based reconstruction: A clinical feasibility study in comparison to conventional multi-breath-hold T2WI liver MRIMagnetic resonance imaging 2021 09;81:75-81 doi:10.1016/j.mri.2021.06.014\u003c/li\u003e\n\u003cli\u003eYanjie Zhao, Chengdong Peng, Shaofang Wang, et al. The feasibility investigation of AI -assisted compressed sensing in kidney MR imaging: an ultra-fast T2WI imaging technologyBMC medical imaging 2022 07 04;22 (1):119 doi:10.1186/s12880-022-00842-1\u003c/li\u003e\n\u003cli\u003eXianghu Yan, Yi Luo, Xiao Chen, et al. From Compressed-Sensing to Deep Learning MR: Comparative Biventricular Cardiac Function Analysis in a Patient CohortJournal of magnetic resonance imaging: JMRI 2024 04;59 (4):1231-1241 doi:10.1002/jmri.28899\u003c/li\u003e\n\u003cli\u003eKang-Lung Lee, Dimitri A Kessler, Simon Dezonie, et al. Assessment of deep learning-based reconstruction on T2-weighted and diffusion-weighted prostate MRI image qualityEuropean journal of radiology 2023 Sep;166:111017 doi:10.1016/j.ejrad.2023.111017\u003c/li\u003e\n\u003cli\u003eSung Hwan Bae, Jiyoung Hwang, Seong Sook Hong, et al. Clinical feasibility of accelerated diffusion weighted imaging of the abdomen with deep learning reconstruction: Comparison with conventional diffusion weighted imagingEuropean journal of radiology 2022 Sep;154:110428 doi:10.1016/j.ejrad.2022.110428\u003c/li\u003e\n\u003cli\u003eDaniel Wessling, Sebastian Gassenmaier, Susann-Cathrin Olthof, et al. Novel deep-learning-based diffusion weighted imaging sequence in 1.5 T breast MRIEuropean journal of radiology 2023 Sep;166:110948 doi:10.1016/j.ejrad.2023.110948\u003c/li\u003e\n\u003cli\u003eJeong Woo Kim, Bit Na Park, Dominik Nickel, Mun Young Paek, Chang Hee Lee; Clinical feasibility of deep learning-accelerated single-shot turbo spin echo sequence with enhanced denoising for pancreas MRI at 3 TeslaEuropean journal of radiology 2024 Dec;181:111737 doi:10.1016/j.ejrad.2024.111737\u003c/li\u003e\n\u003cli\u003eKai Liu, Bin Xi, Haitao Sun, et al. The clinical feasibility of artificial intelligence-assisted compressed sensing single-shot fluid-attenuated inversion recovery (ACS-SS-FLAIR) for evaluation of uncooperative patients with brain diseases: comparison with the conventional T2-FLAIR with parallel imagingActa radiologica (Stockholm, Sweden: 1987) 2023 May;64 (5):1943-1949 doi:10.1177/02841851221139125\u003c/li\u003e\n\u003cli\u003eQizheng Wang, Weili Zhao, Xiaoying Xing, et al. Feasibility of AI-assisted compressed sensing protocols in knee MR imaging: a prospective multi-reader studyEuropean radiology 2023 Dec;33 (12):8585-8596 doi:10.1007/s00330-023-09823-6\u003c/li\u003e\n\u003cli\u003eMing Ni, Miao He, Yuxin Yang, et al. Application research of AI-assisted compressed sensing technology in MRI scanning of the knee joint: 3D-MRI perspectiveEuropean radiology 2024 May;34 (5):3046-3058 doi:10.1007/s00330-023-10368-x\u003c/li\u003e\n\u003cli\u003eYangsean Choi, Ji Su Ko, Ji Eun Park, et al. Beyond the Conventional Structural MRI: Clinical Application of Deep Learning Image Reconstruction and Synthetic MRI of the BrainInvestigative radiology 2025 Jan 01;60 (1):27-42 doi:10.1097/RLI.0000000000001114\u003c/li\u003e\n\u003cli\u003eKai Liu, Haitao Sun, Xingxing Wang, et al. Feasibility of the application of deep learning-reconstructed ultra-fast respiratory-triggered T2-weighted imaging at 3 T in liver imagingMagnetic resonance imaging 2024 Jun;109:27-33 doi:10.1016/j.mri.2024.03.001\u003c/li\u003e\n\u003cli\u003eKai Liu, Haitao Sun, Xingxing Wang, et al. Feasibility of the application of deep learning-reconstructed ultra-fast respiratory-triggered T2-weighted imaging at 3 T in liver imagingMagnetic resonance imaging 2024 Jun;109:27-33 doi:10.1016/j.mri.2024.03.001\u003c/li\u003e\n\u003cli\u003eGiovanni Foti, Chiara Longo; Deep learning and AI in reducing magnetic resonance imaging scanning time: advantages and pitfalls in clinical practicePolish journal of radiology 2024;89: e443-e451 doi:10.5114/pjr/192822\u003c/li\u003e\n\u003cli\u003eJan Vosshenrich, Mary Bruno, Tatiane Cantarelli Rodrigues, et al. Arthroscopy-validated Diagnostic Performance of 7-Minute Five-Sequence Deep Learning Super-Resolution 3-T Shoulder MRIRadiology 2025 Feb;314 (2): e241351 doi:10.1148/radiol.241351\u003c/li\u003e\n\u003cli\u003eSven S Walter, Jan Vosshenrich, Tatiane Cantarelli Rodrigues, et al. Deep Learning Superresolution for Simultaneous Multislice Parallel Imaging-Accelerated Knee MRI Using Arthroscopy ValidationRadiology 2025 Jan;314 (1): e241249 doi:10.1148/radiol.241249\u003c/li\u003e\n\u003cli\u003eJan M Brendel., Reza Dehdab., Judith Herrmann., et al. Deep learning reconstruction for accelerated 3-D magnetic resonance cholangiopancreatographyRadiol Med 2025; doi:10.1007/s11547-025-01987-z\u003c/li\u003e\n\u003cli\u003eAnn-Christin Klemenz, Linda Reichardt, Margarita Gorodezky, et al. Accelerated Cardiac MRI with Deep Learning-based Image Reconstruction for Cine ImagingRadiology. Cardiothoracic imaging 2024 Dec;6 (6): e230419 doi:10.1148/ryct.230419\u003c/li\u003e\n\u003cli\u003eH Akai, K Yasaka, H Sugawara, et al. Faster acquisition of magnetic resonance imaging sequences of the knee via deep learning reconstruction: a volunteer studyClinical radiology 2024 06;79 (6):453-459 doi:10.1016/j.crad.2024.03.002\u003c/li\u003e\n\u003cli\u003eT Tajima, H Akai, K Yasaka, et al. Usefulness of deep learning-based noise reduction for 1.5 T MRI brain imagesClinical radiology 2023 01;78 (1): e13-e21 doi:10.1016/j.crad.2022.08.127\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1. Acquisition parameters of the brain MRI sequences.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"99%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eParameters\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eDR-ACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003eRoutine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eTR (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e5100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e5100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eTE (ms)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e115\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eSlice thickness (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eSlice gap (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eFOV (mm\u003csup\u003e2\u003c/sup\u003e )\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e240\u0026times;200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e240\u0026times;200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e240\u0026times;200\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eMatrix\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e400\u0026times;320\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e400\u0026times;320\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e400\u0026times;320\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eSlice\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eAI-assisted compressed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eDeep-learning reconstructed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003eScan time(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003e25.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e25.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e61\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003eTable 2. Comparison of the\u0026nbsp;rating results\u0026nbsp;of images acquired with DR-ACS-T2WI, ACS-T2WI and routine T2WI.\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 18px;\"\u003e\n \u003cp\u003eParameters Reader\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 41px;\"\u003e\n \u003cp\u003eReader 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 39px;\"\u003e\n \u003cp\u003eReader 2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003eDR-ACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003eACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003eRoutine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003eDR-ACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003eACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003eRoutine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003eOverall image quality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.90\u0026plusmn;0.30（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e4.83\u0026plusmn;0.39（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.65\u0026plusmn;0.48（3,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.90\u0026plusmn;0.30（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.82\u0026plusmn;0.39（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.64\u0026plusmn;0.49（3,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003eMotion artifact\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.91\u0026plusmn;0.29（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e4.81\u0026plusmn;0.40（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.50\u0026plusmn;0.50（3,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.91\u0026plusmn;0.28（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.80\u0026plusmn;0.40（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.50\u0026plusmn;0.50（3,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003eDiagnostic confidence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.92\u0026plusmn;0.28（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e4.80\u0026plusmn;0.41（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.56\u0026plusmn;0.50（3,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.91\u0026plusmn;0.30（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.79\u0026plusmn;0.42（4,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e4.56\u0026plusmn;0.51（3,5）\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTable 3. Inter-rater Consistency Analysis (Kappa Values) of multi-sequence images by two radiologists.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\"\u003e\n \u003cp\u003eDR-ACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\"\u003e\n \u003cp\u003eACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\"\u003e\n \u003cp\u003eRoutine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ekappa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95%CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ekappa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95%CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ekappa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95%CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eOverall image quality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.783-0.917\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.72\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.636-0.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.540-0.720\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eMotion artifact\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.703-0.857\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.561-0.739\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.561, 0.739\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eDiagnostic confidence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.748-0.892\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.625-0.795\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.529, 0.711\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTable 4. The gray matter SNR(SNR\u003csub\u003eGM\u003c/sub\u003e), white matter SNR(SNR\u003csub\u003eWM\u003c/sub\u003e), and gray-white matter CNR(CNR\u003csub\u003eWM/GM\u003c/sub\u003e) in Group DR-ACS-T2WI, ACS-T2WI and routine T2WI.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"105%\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 25px;\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eSNR\u003csub\u003eWM\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eSNR\u003csub\u003eGM\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003eCNR\u003csub\u003eWM/GM\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eDR-ACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 25px;\"\u003e\n \u003cp\u003e110\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e65.06\u0026plusmn;12.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e97.25\u0026plusmn;18.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e32.93\u0026plusmn;12.35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 25px;\"\u003e\n \u003cp\u003e110\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e47.62\u0026plusmn;8.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e71.54\u0026plusmn;12.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e24.29\u0026plusmn;9.08\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eRoutine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 25px;\"\u003e\n \u003cp\u003e110\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e34.32\u0026plusmn;6.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e51.92\u0026plusmn;8.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e17.31\u0026plusmn;6.01\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 25px;\"\u003e\n \u003cp\u003eDR-ACS-T2WI vs. ACS-T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e12.289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e12.202\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e5.909\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 25px;\"\u003e\n \u003cp\u003eDR-ACS-T2WI vs. Routine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e23.452\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e23.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e11.924\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 25px;\"\u003e\n \u003cp\u003eACS-T2WI vs. Routine T2WI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e12.879\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e13.894\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e6.727\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"AI-assisted compressed sensing, deep-learning reconstruction, brain MRI","lastPublishedDoi":"10.21203/rs.3.rs-7239920/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7239920/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWhile using AI-assisted compressed sensing (ACS) combined with deep-learning reconstruction (DR) techniques has the potential to Shorten the acquisition time of brain T2-weighted imaging (T2WI), its imaging improvements and clinical application in brain imaging remains under explored.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePurpose\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate the clinical feasibility of DR-ACS for brain T2-weighted imaging (T2WI), focusing on image quality, acquisition efficiency, and diagnostic accuracy, compared with the routine T2WI.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterial and Methods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA prospective cohort of 110 participants underwent brain MRI using three protocols at a 3.0-T MR scanner: routine T2WI, ACS-T2WI (without DR), and DR-ACS-T2WI. Subjective image quality (overall image quality, motion artifact, and diagnostic confidence, with a 5-point scale) and objective metrics (Signal-to-Noise Ratio (SNR), Contrast-to-Noise Ratio (CNR), and scan time) were compared. Statistical analysis included ANOVA and kappa.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe overall image quality, motion artifact, and diagnostic confidence scores of DR-ACS-T2WI, assessed by two radiologists, were 4.90±0.30, 4.91±0.29 , 4.92±0.28(Reader 1) and 4.90±0.30 , 4.91±0.28, 4.91±0.30 (Reader 2), higher than those of ACS-T2WI and routine T2WI. DR-ACS-T2WI demonstrated superior SNRs in both white matter (65.06±12.1) and gray matter (97.25±18.52) compared to ACS-T2WI (47.62±8.65 and 71.54±12.05, respectively) and routine T2WI (34.32±6.51 and 51.92±8.62, respectively; P \u0026lt; 0.001 for all). Similarly, the gray-white matter CNR of DR-ACS-T2WI (32.93±12.35) was significantly higher than that of ACS-T2WI (24.29±9.08) and routine T2WI (17.31±6.01; P \u0026lt; 0.001). Additionally, the scan time of DR-ACS-T2WI and ACS-T2WI (both 25.7s) was 57.87% shorter than that of the routine T2WI (61s).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe ACS combined with DR is clinically feasible for MRI examinations of brain diseases, offering significantly shorter image acquisition time and higher image quality compared with the routine T2WI.\u003c/p\u003e","manuscriptTitle":"Efficacy Evaluation of AI-Assisted Compressed Sensing Combined with Deep-Learning Reconstruction in Accelerating Brain T2-Weighted Imaging: A Clinical Feasibility Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-17 05:13:03","doi":"10.21203/rs.3.rs-7239920/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-06T13:05:34+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-08T20:15:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-30T13:43:34+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"174152050688614595193741398298552218784","date":"2025-11-26T17:44:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"261273389537732932948824114608899836934","date":"2025-11-23T01:56:12+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-08T15:55:59+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-05T13:34:35+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-11T05:13:29+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-08T13:43:22+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Imaging","date":"2025-08-08T13:40:25+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"901aed3d-9059-4074-84b6-9de037f97b74","owner":[],"postedDate":"September 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-03-25T06:38:21+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-17 05:13:03","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7239920","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7239920","identity":"rs-7239920","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}