Physics in medicine and biology最新文献

{"title":"Quantification of the tumour microvascular response to high dose-per-fraction radiotherapy.","authors":"W Jeffrey Zabel, Hector A Contreras-Sanchez, Nader Allam, Muhammad Mohsin Qureshi, Costel Flueraru, Edward Taylor, I Alex Vitkin","doi":"10.1088/1361-6560/ade048","DOIUrl":"10.1088/1361-6560/ade048","url":null,"abstract":"<p><p><i>Objective</i>. Microvascular ablation during high dose-per-fraction radiotherapy (HDFRT) is disparately reported in the literature. This study was conducted to quantify the tumour microvascular response to different HDFRT schedules.<i>Approach</i>. A high single-dose irradiation of 20 Gy and two multifraction schedules (three fractions of 10 Gy and 15 Gy each) were studied. Patient-derived BxPC-3 pancreatic tumours in a mouse dorsal skinfold window chamber were treated and their 3D microvascular networks were longitudinally imaged with speckle variance optical coherence tomography for up to 7 weeks post irradiation. The overall vascular volume density (VVD), VVD for small vessels (diameters between 15-25<i>μ</i>m and 25-35<i>μ</i>m), and the vascular convexity index<i>λ</i>(a measure of vessel organization and space filling at short distances) were quantified.<i>Main results</i>. There were no significant differences in overall VVD for treated vs. control tumours at all timepoints. Examination of small-diameter vessels revealed some transient reductions in VVD_{15-25<i>μ</i>m}and VVD_{25-35<i>μ</i>m}compared to controls at<i>t</i>∼ 3 weeks for larger dose-per-fraction regimens (3 × 15 Gy and 1 × 20 Gy); ablated vasculature regrew back to baseline values by 7 weeks. Convexity indices for these larger-dose-per-fraction tumours were ∼55% larger than unirradiated controls by the end of monitoring period; no such effects were seen in the 3 × 10 Gy cohort.<i>Significance</i>. The results of this study reveal the complex role of small vessels in microvascular ablation caused by HDFRT, with a dependence on the dose per fraction and total delivered dose. After small vessel ablation, regrown vessels had more uniform and regular spacing than non-ablated vessels as quantified by<i>λ</i>, potentially suggesting improved tumour response if subsequent retreatments are attempted.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144216557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dosimetrically coupled multiscale tetrahedral mesh models of human liver vasculature: implications for radiopharmaceutical dosimetry of both organ blood and parenchyma.","authors":"Robert J Dawson, Carlos Huesa-Berral, Lukas M Carter, Chris Beekman, Chansoo Choi, Bangho Shin, Mislav Bobić, Nicolò Cogno, Julia D Withrow, Derek W Jokisch, Alejandro Bertolet, Harald Paganetti, Wesley E Bolch","doi":"10.1088/1361-6560/addfa6","DOIUrl":"10.1088/1361-6560/addfa6","url":null,"abstract":"<p><p><i>Objective.</i>To develop a computational framework coupling multiscale vascular models of the human liver for improved radiation dosimetry calculations that clearly distinguish the absorbed dose to tissue parenchyma and that to its blood content at all spatial scales. This framework thus addresses limitations of homogeneous blood/tissue organ models in present use in radiopharmaceutical therapy.<i>Approach.</i>High-fidelity tetrahedral mesh models of liver vasculature were constructed at two spatial scales. At the macroscale, detailed hepatic arterial, venous, and portal venous networks were generated within reference adult male and female computational phantoms. At the microscale, a classical hexagonal liver lobule model incorporating sinusoids, bile compartments, and cellular components was developed. A mathematical framework was further developed to couple Monte Carlo radiation transport results across these spatial scales, enabling comprehensive dosimetric calculations for radiation dose to both blood and parenchymal tissues.<i>Main Results.</i>The coupled model system successfully accounted for the entire blood content of the liver, with approximately 31% represented in macroscale vessels (⩾100<i>μ</i>m diameter) and 69% within microscale structures. Specific absorbed fractions were computed for monoenergetic photons, electrons, and alpha particles, demonstrating reciprocity between blood-to-parenchyma and parenchyma-to-blood crossfire. Reference<i>S</i>-values were computed for 22 therapeutic and 11 diagnostic radionuclides, providing the first comprehensive dataset for blood-specific and parenchyma-specific internal dosimetry calculations in the liver.<i>Significance.</i>This work establishes a novel framework for multi-scale radiation transport calculations in vascularized organs, enabling separate tracking of blood and parenchymal tissue doses. The methodology has immediate applications in improving dose calculations for radiopharmaceutical therapies, Y-90 microsphere radioembolization treatment, and analysis of blood dose during external beam radiotherapy. The approach can be readily adapted for other vascularized organs, representing a significant advancement in radiation dosimetry accuracy for both therapeutic and diagnostic applications by fully and independently accounting for organ activity localized within two tissue compartments-organ blood and organ parenchyma.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144209174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of an ultrasensitive small animal PET with 4-layer DOI detectors for sub-second dynamic rodent imaging.","authors":"Han Gyu Kang, Hideaki Tashima, Hidekatsu Wakizaka, Go Akamatsu, Yuma Iwao, Chie Toramatsu, Taiga Yamaya","doi":"10.1088/1361-6560/ade112","DOIUrl":"10.1088/1361-6560/ade112","url":null,"abstract":"<p><p><i>Objective.</i>Dynamic positron emission tomography (PET) imaging is important for preclinical research since it can visualize the functional information of rodent models as a function of time. However, the temporal resolution of small animal PET imaging has been limited to a scale of seconds due to low sensitivity, and it is not sufficient to capture cardiac or brain function accurately. Here, we present an ultrasensitive small-animal PET scanner with total-body coverage for sub-second dynamic imaging of a rat.<i>Methods.</i>The ultrasensitive small animal PET scanner has a 155 mm inner diameter and 325.6 mm axial coverage. The PET scanner has six rings, each of which has 10 depth-of-interaction (DOI) detectors. Each DOI detector consists of a four-layer Zr-doped gadolinium oxyorthosilicate crystal array (2.85 mm pitch, 30 mm total thickness) and 8 × 8 multi-anode photomultiplier tubes. The physical PET performance was evaluated based on the National Electrical Manufacturers Association NU4 protocol. Sub-second dynamic rat imaging was performed with¹⁸F-FDG tracer.<i>Main results.</i>The peak absolute sensitivity was 20.2% and spatial resolution was 2.6 mm at the center of the field of view with an energy window of 400-600 keV. Total-body images of a rat were obtained with a single bed position. The cardiac function of a rat was visualized with 0.25 s temporal resolution, which was hardly possible with conventional small animal PET scanners.<i>Significance</i>. The developed ultrasensitive animal PET enabled sub-second dynamic PET imaging in rodent models with total-body coverage. In conclusion, the ultrasensitive small animal PET scanner can serve as a useful molecular imaging tool for preclinical research with its long axial coverage sub-second temporal resolution.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144226270","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hydrated electron yield dependence on instantaneous dose rates with electron ultra-high dose rate (UHDR) irradiation.","authors":"Xu Cao, Aubrey Parks, William Thomas, Matthew S Reed, Wesley S Culberson, Brian W Pogue","doi":"10.1088/1361-6560/ade223","DOIUrl":"10.1088/1361-6560/ade223","url":null,"abstract":"<p><p><i>Objective.</i>The aim of this study was to quantify the characteristic transient production yield of hydrated electrons (<i>G</i>-value) in water under ultra-high dose rates (UHDR) with electron irradiation. Changes in this yield with UHDR irradiation may provide insights into the radiation chemistry origins of the normal tissue-sparing effect observed in FLASH therapy.<i>Approach.</i>A multi-pass transmission measurement technique was used to detect and quantify hydrated electrons based on its near-infrared absorption. Transient absorbance profiles of hydrated electrons were used to estimate the<i>G</i>-value as a function of variation in pulse width, source-to-surface distance and pulse frequency, to assess dependence upon instantaneous and average dose rates.<i>Main results.</i>The study confirmed a linear relationship between total radiation dose and production of hydrated electrons, giving a stable<i>G</i>-value for a fixed dose rate. However, the measured<i>G</i>-value increased with increasing instantaneous dose rates (0.18-0.33 MGy s^-1), ranging from 35 up to 48 nM Gy^-1. However, the<i>G</i>-value did not change appreciably when varying the pulse frequency and pulse width, which varies the average dose rate.<i>Significance.</i>The instantaneous dose rate of UHDR directly influenced the generation yield of hydrated electrons during UHDR water radiolysis. Further research is needed to solidify this connection and to better understand the role of hydrated electrons in the observed sparing effect of FLASH radiotherapy.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144249180","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automatic contour quality assurance using deep-learning based contours.","authors":"Barbara Marquez, David Fuentes, Christine B Peterson, Dong Joo Rhee, Raphael J Douglas, Raymond P Mumme, Anuja Jhingran, Julianne M Pollard, Surendra Prajapati, Thomas Whitaker, Laurence E Court","doi":"10.1088/1361-6560/ade5e6","DOIUrl":"https://doi.org/10.1088/1361-6560/ade5e6","url":null,"abstract":"<p><strong>Objective: </strong>Safe deployment of auto-contouring models requires the inclusion of automated QA. One such approach is to use two independent auto-contouring models and compare them geometrically for acceptability. This is not effective because geometric differences may not correlate with clinically significant errors. Herein, we investigated whether a two-contour QA system is improved by including dose in this comparison.&#xD;Approach. VMAT plans were generated for 86 head and neck (H&N) and 50 cervical (GYN) cancer patients, using clinically-approved PTVs and auto-contour OARs from a primary auto-contouring model. Doses to the primary OARs were compared with doses to manually drawn and approved OARs (\"the truth\"). A difference in Dmean or Dmax ≥ 2 Gy was identified as a reporting error (Derror). A second, independent auto-contouring model was then used to contour the OARs (verification). The primary and verification auto-contouring models were compared geometrically (DSC, sDSC, HD95, MSD) and dosimetrically (Dmean, Dmax). The ability of comparison metrics between the two auto-contouring models to flag actual dosimetric errors (i.e. primary model compared with the truth) was investigated. A logistic regression model was used to predict Derror. The data was divided by disease site and into 50/50 stratified training and testing sets; k-fold cross validation was employed during training to avoid overfitting. H&N structures were further divided into size-specific groups to improve model performance and generalizability.&#xD;Main Results. Including dose metrics in the logistic regression model to predict Derror, mean increased the performance in terms of ROC-AUC and AU-PRC in the test set for H&N small structures. For Derror, max, including dose metrics increased performance for H&N small structures, H&N medium structures, and GYN structures. &#xD;Significance. In many instances, utilizing dose with geometric comparisons can improve the ability of a verification model to flag potential errors from a primary auto-contouring model.&#xD.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144326650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simulation and one-ring prototyping of 1 mm-rod-resolution hemispherical brain PET with TOF-DOI detectors.","authors":"Kurumi Narita, Go Akamatsu, Eiji Yoshida, Hideaki Tashima, Yuma Iwao, Miwako Takahashi, Taiga Yamaya","doi":"10.1088/1361-6560/ade2b4","DOIUrl":"10.1088/1361-6560/ade2b4","url":null,"abstract":"<p><p><i>Objective.</i>Brain positron emission tomography (PET) imaging plays crucial roles in research and diagnosis of various brain diseases. To achieve high spatial resolution and high sensitivity, we proposed a hemispherical geometry which offers higher sensitivity with fewer detectors than a conventional cylindrical geometry. Our developed hemispherical brain PET system, Vrain, has indeed achieved a rod resolution of 2.2 mm with a 229 ps time-of-flight (TOF) resolution. To further improve the spatial resolution, we will use TOF and depth-of-interaction (DOI) detectors with our original crosshair light-sharing (CLS) configuration. This study aimed at estimating the performance of the hemispherical brain PET with TOF-DOI detectors and at developing a one-ring PET prototype with 1.6 mm scintillator pitch CLS-based TOF-DOI detectors.<i>Approach.</i>The sensitivity, rod resolution, and image quality of the TOF-DOI hemispherical brain PET (TDHBP-sim) and Vrain (Vrain-sim) were estimated using Geant4 simulation. A one-ring prototype with a 30 cm diameter was developed using the CLS-based TOF-DOI detectors. The energy resolution, TOF timing resolution, rod resolution, and the Hoffman brain phantom image quality of the prototype were evaluated.<i>Main results.</i>In the simulation study, TDHBP-sim achieved 1.4 times better sensitivity than Vrain-sim. TDHBP-sim visualized 1.0 mm rods and gyri and sulci structures in the brain phantom. In the one-ring experiment, the energy resolution was 11.6% at 511 keV, the TOF timing resolution was 294.6 ps, and 1.0 mm rods were resolved at the central 10 cm-diameter field-of-view. The 0.8 mm-thick radioactivity distribution could be identified in the Hoffman phantom.<i>Significance.</i>The study findings suggested that a hemispherical brain PET with 1.6 mm scintillator pitch TOF-DOI detectors should offer excellent performance including 1 mm rod resolution.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144258718","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quasi-supervised MR-CT image conversion based on unpaired data.","authors":"Ruiming Zhu, Yuhui Ruan, Mingrui Li, Wei Qian, Yudong Yao, Yueyang Teng","doi":"10.1088/1361-6560/ade220","DOIUrl":"10.1088/1361-6560/ade220","url":null,"abstract":"<p><p><i>Objective</i>. In radiotherapy planning, acquiring both magnetic resonance (MR) and computed tomography (CT) images is crucial for comprehensive evaluation and treatment. However, simultaneous acquisition of MR and CT images is time-consuming, economically expensive, and involves ionizing radiation, which poses health risks to patients. The objective of this study is to generate CT images from radiation-free MR images using a novel quasi-supervised learning framework.<i>Approach</i>. In this work, we propose a quasi-supervised framework to explore the underlying relationship between unpaired MR and CT images. Normalized mutual information (NMI) is employed as a similarity metric to evaluate the correspondence between MR and CT scans. To establish optimal pairings, we compute an NMI matrix across the training set and apply the Hungarian algorithm for global matching. The resulting MR-CT pairs, along with their NMI scores, are treated as prior knowledge and integrated into the training process to guide the MR-to-CT image translation model.<i>Main results</i>. Experimental results indicate that the proposed method significantly outperforms existing unsupervised image synthesis methods in terms of both image quality and consistency of image features during the MR to CT image conversion process. The generated CT images show a higher degree of accuracy and fidelity to the original MR images, ensuring better preservation of anatomical details and structural integrity.<i>Significance</i>. This study proposes a quasi-supervised framework that converts unpaired MR and CT images into structurally consistent pseudo-pairs, providing informative priors to enhance cross-modality image synthesis. This strategy not only improves the accuracy and reliability of MR-CT conversion, but also reduces reliance on costly and scarce paired datasets. The proposed framework offers a practical and scalable solution for real-world medical imaging applications, where paired annotations are often unavailable.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144249181","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing image quality in fast neutron-based range verification of proton therapy using a deep learning-based prior in LM-MAP-EM reconstruction.","authors":"Lena M Setterdahl, Kyrre Skjerdal, Hunter N Ratliff, Kristian Smeland Ytre-Hauge, William R B Lionheart, Sean Holman, Helge E S Pettersen, Francesco Blangiardi, Danny Lathouwers, Ilker Meric","doi":"10.1088/1361-6560/ade198","DOIUrl":"10.1088/1361-6560/ade198","url":null,"abstract":"<p><p><i>Objective.</i>This study investigates the use of list-mode (LM) maximum<i>a posteriori</i>(MAP) expectation maximization (EM) incorporating prior information predicted by a convolutional neural network for image reconstruction in fast neutron (FN)-based proton therapy range verification.<i>Approach</i>. A conditional generative adversarial network (pix2pix) was trained on progressively noisier data, where detector resolution effects were introduced gradually to simulate realistic conditions. FN data were generated using Monte Carlo simulations of an 85 MeV proton pencil beam in a computed tomography-based lung cancer patient model, with range shifts emulating weight gain and loss. The network was trained to estimate the expected two-dimensional ground truth FN production distribution from simple back-projection images. Performance was evaluated using mean squared error, structural similarity index (SSIM), and the correlation between shifts in predicted distributions and true range shifts.<i>Main results</i>. Our results show that pix2pix performs well on noise-free data but suffers from significant degradation when detector resolution effects are introduced. Among the LM-MAP-EM approaches tested, incorporating a mean prior estimate into the reconstruction process improved performance, with LM-MAP-EM using a mean prior estimate outperforming naïve LM maximum likelihood EM (LM-MLEM) and conventional LM-MAP-EM with a smoothing quadratic energy function in terms of SSIM.<i>Significance</i>. Findings suggest that deep learning techniques can enhance iterative reconstruction for range verification in proton therapy. However, the effectiveness of the model is highly dependent on data quality, limiting its robustness in high-noise scenarios.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144234736","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"High-definition motion-resolved MRI using 3D radial kooshball acquisition and deep learning spatial-temporal 4D reconstruction.","authors":"Victor Murray, Can Wu, Ricardo Otazo","doi":"10.1088/1361-6560/ade195","DOIUrl":"10.1088/1361-6560/ade195","url":null,"abstract":"<p><p><i>Objective.</i>To develop motion-resolved volumetric MRI with 1.1 mm isotropic resolution and scan times <5 min using a combination of 3D radial kooshball acquisition and spatial-temporal deep learning 4D reconstruction for free-breathing high-definition (HD) lung MRI.<i>Approach.</i>Free-breathing lung MRI was conducted on eight healthy volunteers and ten patients with lung tumors on a 3 T MRI scanner using a 3D radial kooshball sequence with half-spoke (ultrashort echo time, UTE, TE = 0.12 ms) and full-spoke (T1-weighted, TE = 1.55 ms) acquisitions. Data were motion-sorted using amplitude-binning on a respiratory motion signal. Two high-definition Movienet (HD-Movienet) deep learning models were proposed to reconstruct 3D radial kooshball data: slice-by-slice reconstruction in the coronal orientation using 2D convolutional kernels (2D-based HD-Movienet) and reconstruction on blocks of eight coronal slices using 3D convolutional kernels (3D-based HD-Movienet). Two applications were considered: (a) anatomical imaging at expiration and inspiration with four motion states and a scan time of 2 min, and (b) dynamic motion imaging with 10 motion states and a scan time of 4 min. The training was performed using XD-GRASP 4D images reconstructed from 4.5 min and 6.5 min acquisitions as references.<i>Main Results.</i>2D-based HD-Movienet achieved a reconstruction time of <6 s, significantly faster than the iterative XD-GRASP reconstruction (>10 min with GPU optimization) while maintaining comparable image quality to XD-GRASP with two extra minutes of scan time. The 3D-based HD-Movienet improved reconstruction quality at the expense of longer reconstruction times (<11 s).<i>Significance.</i>HD-Movienet demonstrates the feasibility of motion-resolved 4D MRI with isotropic 1.1 mm resolution and scan times of only 2 min for four motion states and 4 min for 10 motion states, marking a significant advancement in clinical free-breathing lung MRI.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144234737","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MRI super-resolution reconstruction using efficient diffusion probabilistic model with residual shifting.","authors":"Mojtaba Safari, Shansong Wang, Zach Eidex, Qiang Li, Richard L J Qiu, Erik H Middlebrooks, David S Yu, Xiaofeng Yang","doi":"10.1088/1361-6560/ade049","DOIUrl":"10.1088/1361-6560/ade049","url":null,"abstract":"<p><p><i>Objective.</i>Magnetic resonance imaging (MRI) is essential in clinical and research contexts, providing exceptional soft-tissue contrast. However, prolonged acquisition times often lead to patient discomfort and motion artifacts. Diffusion-based deep learning super-resolution (SR) techniques reconstruct high-resolution (HR) images from low-resolution (LR) pairs, but they involve extensive sampling steps, limiting real-time application. To overcome these issues, this study introduces a residual error-shifting mechanism markedly reducing sampling steps while maintaining vital anatomical details, thereby accelerating MRI reconstruction.<i>Approach.</i>We developed Res-SRDiff, a novel diffusion-based SR framework incorporating residual error shifting into the forward diffusion process. This integration aligns the degraded HR and LR distributions, enabling efficient HR image reconstruction. We evaluated Res-SRDiff using ultra-high-field brain T1 MP2RAGE maps and T2-weighted prostate images, benchmarking it against Bicubic, Pix2pix, CycleGAN, SPSR, I²SB, and TM-DDPM methods. Quantitative assessments employed peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), gradient magnitude similarity deviation (GMSD), and learned perceptual image patch similarity. Additionally, we qualitatively and quantitatively assessed the proposed framework's individual components through an ablation study and conducted a Likert-based image quality evaluation.<i>Main results.</i>Res-SRDiff significantly surpassed most comparison methods regarding PSNR, SSIM, and GMSD for both datasets, with statistically significant improvements (p-values≪0.05). The model achieved high-fidelity image reconstruction using only four sampling steps, drastically reducing computation time to under one second per slice. In contrast, traditional methods like TM-DDPM and I²SB required approximately 20 and 38 s per slice, respectively. Qualitative analysis showed Res-SRDiff effectively preserved fine anatomical details and lesion morphologies. The Likert study indicated that our method received the highest scores,4.14±0.77(brain) and4.80±0.40(prostate).<i>Significance.</i>Res-SRDiff demonstrates efficiency and accuracy, markedly improving computational speed and image quality. Incorporating residual error shifting into diffusion-based SR facilitates rapid, robust HR image reconstruction, enhancing clinical MRI workflow and advancing medical imaging research. Code available athttps://github.com/mosaf/Res-SRDiff.</p>","PeriodicalId":20185,"journal":{"name":"Physics in medicine and biology","volume":" ","pages":""},"PeriodicalIF":3.3,"publicationDate":"2025-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12167925/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144216556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}