Magnetic resonance imaging最新文献

{"title":"Evaluation of blood supply using dynamic contrast-enhanced magnetic resonance imaging and its application in spinal tumor surgery","authors":"Hongyang Guo ,&nbsp;Xueyan Li ,&nbsp;Xiaodong Gao ,&nbsp;Ying Wang ,&nbsp;Yuanjie Fan ,&nbsp;Jiawei Miao ,&nbsp;Chen Cheng ,&nbsp;Yongqiang Jiao","doi":"10.1016/j.mri.2025.110400","DOIUrl":"10.1016/j.mri.2025.110400","url":null,"abstract":"<div><h3>Objective</h3><div>This study aimed to assess the utility of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in evaluating blood supply for spinal tumors and its predictive value for intraoperative blood loss and transfusion requirements during spinal tumor surgery.</div></div><div><h3>Methods</h3><div>We retrospectively analyzed clinical data from 20 patients with single vertebral tumors who underwent surgery at the Affiliated Hospital of Hebei University of Technology between December 2018 and December 2020. Patients were categorized based on DCE-MRI into two groups: those with tumors indicating rich blood supply (15 patients, with 12 undergoing preoperative embolization) and those with non-rich blood supply tumors (5 patients, without embolization). The primary outcomes measured were blood loss, operation time, and blood transfusion.</div></div><div><h3>Results</h3><div>The blood flow (BF) ratio from DCE-MRI showed a significant positive correlation with DSA scores (<em>r</em> = 0.569, <em>P</em> &lt; 0.05), indicating the reliability of DCE-MRI in evaluating tumor vascularity. In the group with rich blood supply tumors, the median DSA score was 3.25 (range 3–4), and the BF ratio ranged from 1.84 to 5.14, with a median value greater than 1.8. The BF ratio also correlated significantly with intraoperative bleeding (<em>r</em> = 0.537, <em>P</em> &lt; 0.001) and blood transfusion requirements (<em>r</em> = 0.579, P &lt; 0.001). The correlation with operation time was less pronounced (<em>r</em> = 0.259, P &lt; 0.001).</div></div><div><h3>Conclusion</h3><div>The DCE-MRI BF ratio is significantly correlated with intraoperative blood loss and transfusion requirements, providing valuable preoperative guidance for spinal tumor surgery. Its non-invasive predictive capabilities offer a clear advantage over traditional angiography, facilitating more informed surgical planning and patient care.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"121 ","pages":"Article 110400"},"PeriodicalIF":2.1,"publicationDate":"2025-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143990499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Age-related change of glymphatic function in normative children assessed using diffusion tensor imaging-analysis along the perivascular space","authors":"Alex Mun-Ching Wong ,&nbsp;Tiing-Yee Siow ,&nbsp;Yi-Ting Cheng ,&nbsp;Eddy Chien-Yuan Lin ,&nbsp;Shin-Nan Lin ,&nbsp;Kuang-Lin Lin ,&nbsp;Cheng-Hong Toh","doi":"10.1016/j.mri.2025.110398","DOIUrl":"10.1016/j.mri.2025.110398","url":null,"abstract":"<div><h3>Background</h3><div>The glymphatic system, crucial for brain waste removal and homeostasis, has an underexplored developmental trajectory in children. This study describes changes in glymphatic function during childhood via diffusion tensor imaging-analysis along the perivascular space (DTI-ALPS).</div></div><div><h3>Methods</h3><div>We retrospectively studied DTI MR images of 72 pediatric participants (mean age = 92.37 months; 0–19.8 years; 50 % females), all showing normal MRI findings from 2019 to 2022. Imaging utilized 3 T scanners with a DTI sequence of 21 diffusion-encoded gradients, focusing on the ALPS index to assess glymphatic function. Clinical outcomes were determined using the Pediatric Cerebral Performance Category Scale and medical records within two weeks post-MRI. Regions-of-interest on diffusion maps were selected manually, guided by fractional anisotropy maps, for automatic ALPS computation. Pearson correlation and multiple linear regression analyzed the relationship between ALPS indices, age, and clinical scores, with a paired <em>t</em>-test comparing bilateral ALPS indices. Significance was set at <em>P</em> &lt; 0.05.</div></div><div><h3>Results</h3><div>Significant positive correlations between age and both left (<em>R</em> = 0.510, <em>P</em> &lt; 0.001) and right (<em>R</em> = 0.688, P &lt; 0.001) DTI-ALPS indices were observed, indicating developmental changes in glymphatic function. Age alone significantly predicted the DTI-ALPS indices (left ALPS: adjusted R² = 0.235; right ALPS: adjusted R² = 0.460), underscoring its developmental trajectory. The study found no significant differences between left and right DTI-ALPS indices, suggesting symmetrical glymphatic function during childhood.</div></div><div><h3>Conclusion</h3><div>This study reveals developmental changes in the glymphatic system across childhood, demonstrating an age-related increase in glymphatic function and bilateral symmetry.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110398"},"PeriodicalIF":2.1,"publicationDate":"2025-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143887320","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Linear rotationally invariant kurtosis measures from double diffusion encoding MRI","authors":"Hunter G. Moss ,&nbsp;Thorsten Feiweier ,&nbsp;Andreana Benitez ,&nbsp;Jens H. Jensen","doi":"10.1016/j.mri.2025.110399","DOIUrl":"10.1016/j.mri.2025.110399","url":null,"abstract":"<div><h3>Purpose</h3><div>To characterize the complete set of linear rotationally invariant kurtosis measures provided by double diffusion encoding (DDE) MRI, show their utility in distinguishing different types of multiple Gaussian compartment (MGC) models, and demonstrate simplified acquisition and analysis schemes for their estimation.</div></div><div><h3>Theory and methods</h3><div>The lowest order novel information obtainable with DDE MRI can be encapsulated in a six-dimensional kurtosis tensor. The most basic DDE MRI kurtosis measures are rotational invariants that are linear in this tensor while depending on no other physical quantities. We identify four such invariants and show that any others must be linear combinations of these. The invariants are applied to classify MGC models according to whether they include microscopic anisotropy or intercompartmental water exchange. In addition, they are used to investigate the effect of exchange on estimates of the microscopic fractional anisotropy (μFA). Simplified acquisition and analysis schemes for the invariants are proposed and demonstrated with human brain data obtained at 3 T.</div></div><div><h3>Results</h3><div>For the considered brain regions, the kurtosis invariants are found to be largely consistent with MGC models having microscopic anisotropy. They also indicate that water exchange in gray matter may affect estimates of μFA.</div></div><div><h3>Conclusion</h3><div>The kurtosis measures can classify MGC models according to whether they have microscopic anisotropy or water exchange, and they can be estimated with simple acquisition and analysis schemes. Measurements of the invariants in brain support the validity of MGC models with microscopic anisotropy and the importance of water exchange for modeling diffusion in gray matter.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110399"},"PeriodicalIF":2.1,"publicationDate":"2025-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143891357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"End-to-End Deep Learning-Based Motion Correction and Reconstruction for Accelerated Whole-Heart Joint T1/T2 Mapping","authors":"Lina Felsner ,&nbsp;Carlos Velasco ,&nbsp;Andrew Phair ,&nbsp;Thomas J. Fletcher ,&nbsp;Haikun Qi ,&nbsp;René M. Botnar ,&nbsp;Claudia Prieto","doi":"10.1016/j.mri.2025.110396","DOIUrl":"10.1016/j.mri.2025.110396","url":null,"abstract":"<div><h3>Purpose</h3><div>To accelerate 3D whole-heart joint T₁/T₂ mapping for myocardial tissue characterization using an end-to-end deep learning algorithm for joint motion estimation and model-based motion-corrected reconstruction of multi-contrast undersampled data.</div></div><div><h3>Methods</h3><div>A free-breathing high-resolution motion-compensated 3D joint T₁/T₂ water/fat sequence is employed. The sequence consists of the acquisition of four interleaved volumes with 2-echo encoding, resulting in eight volumes with different contrasts. An end-to-end non-rigid motion-corrected reconstruction network is used to estimate high quality motion-corrected reconstructions from the eight multi-contrast undersampled data for subsequent joint T₁/T₂ mapping. Reconstruction with the proposed approach was compared against state-of-the-art motion-corrected HD-PROST reconstruction.</div></div><div><h3>Results</h3><div>The proposed approach yields images with good visual agreement compared to the reference reconstructions. The comparison of the quantitative values in the T₁ and T₂ maps showed the absence of systematic errors, and a small bias of <span><math><mo>−</mo><mn>6.35</mn></math></span> ms and <span><math><mo>−</mo><mn>1.8</mn></math></span> ms, respectively. The proposed reconstruction time was <span><math><mn>24</mn></math></span> seconds in comparison to <span><math><mn>2.5</mn></math></span> hours with motion-corrected HD-PROST, resulting in a reconstruction speed-up of over <span><math><mn>370</mn></math></span> times.</div></div><div><h3>Conclusion</h3><div>In conclusion, this study presents a promising method for efficient whole-heart myocardial tissue characterization. Specifically, the research highlights the potential of the multi-contrast end-to-end deep learning algorithm for joint motion estimation and model-based motion-corrected reconstruction of multi-contrast undersampled data. The findings underscore its ability to compute T₁ and T₂ values with good agreement when compared to the reference motion-corrected HD-PROST method, while substantially reducing reconstruction time.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"121 ","pages":"Article 110396"},"PeriodicalIF":2.1,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143921700","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Depth-dependent characterization of cartilage nanostructures using MRI signal decays","authors":"Theodore Aptekarev ,&nbsp;Gregory Furman ,&nbsp;Vladimir Sokolovsky ,&nbsp;Farid Badar ,&nbsp;Yang Xia","doi":"10.1016/j.mri.2025.110395","DOIUrl":"10.1016/j.mri.2025.110395","url":null,"abstract":"<div><h3>Objective</h3><div>The multi-exponential nature of echo decay in nuclear magnetic resonance exam of cartilage complicates the determination of relaxation times. In this study, a novel method has been developed and applied to analyze the cartilage nanostructure using multi-exponential signals. This approach eliminates the need for relaxation time determination, avoids sample rotation, and removes the requirement for multiple experiments. A key feature of this method is its ability to provide detailed insights into the nanostructures of the sample.</div></div><div><h3>Methods</h3><div>Quantitative <em>T</em>₂ imaging method was used to examine the signal delays in mature and healthy canine articular cartilage, at a transverse resolution of 35.1 μm. A modeling method was used to analyze the multi-exponential echo decay for each resolved tissue depth along the full thickness of articular cartilage.</div></div><div><h3>Results</h3><div>The developed approach provides detailed information on the nanostructure in the tissue, which varies with cartilage depth. The information contains the volumes of the water-filled nanocavities created by the fibril structure and their orientation. This information reveals that the superficial and transitional anatomic zones of cartilage contain two distinct types of nanocavities, while the radial zone contains only one type.</div></div><div><h3>Discussion</h3><div>The proposed voxel-based method of echo decay analysis enables the estimation of nanocavities, their angular distribution, and spatial variations of the nanocavity characteristics throughout the sample. This newly developed approach demonstrated that detailed structural tissue information can be obtained as a depth function, representing a significant advancement in understanding cartilage nanostructures and holds potential for future medical applications.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110395"},"PeriodicalIF":2.1,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143858988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting progression of enhancing non-measurable disease in high-grade glioma by using dynamic contrast-enhanced MR imaging","authors":"Haimei Cao ,&nbsp;Zhousan Huang ,&nbsp;Ruowei Qiu ,&nbsp;Xiang Xiao ,&nbsp;Zhiyong Li ,&nbsp;Jay J. Pillai ,&nbsp;Jun Hua ,&nbsp;Guanglong Huang ,&nbsp;Yikai Xu ,&nbsp;Wen Liang ,&nbsp;Yuankui Wu","doi":"10.1016/j.mri.2025.110394","DOIUrl":"10.1016/j.mri.2025.110394","url":null,"abstract":"<div><h3>Objectives</h3><div>To investigate the potential of histogram models derived from dynamic contrast-enhanced (DCE) MR imaging in predicting the progression of enhancing non-measurable disease (NMD) persisting after chemoradiotherapy in patients with high-grade glioma (HGG).</div></div><div><h3>Materials and methods</h3><div>A total of 97 glioma patients (mean age ± standard deviation, 46.7 years ±12.1; 73 men) who underwent temozolomide-based chemoradiation following gross total resection were enrolled retrospectively, including 55 (57 %) in the progression group and 42 (43 %) in the non-progression group. The histogram features of K^trans (volume transfer constant between the plasma and extravascular extracellular space) and Ve (extravascular volume) for enhancing NMDs were extracted and compared between the two groups. Histogram features with significant differences were included in binary logistic regression to construct models to predict progression within 2 to 3 months. The models were constructed based on K^trans and Ve alone or combined. Receiver operating characteristic curves were used to evaluate the prediction performance of the different models. The models were testified in a prospective cohort consisting of 15 patients with HGG.</div></div><div><h3>Results</h3><div>The histogram model of K^trans showed an area under the curve (AUC) of 0.900 in predicting progression. The model of Ve had an AUC of 0.879. When combining K^trans and Ve, the model achieved an AUC of 0.927. These models showed excellent predictive performance in the prospective study.</div></div><div><h3>Conclusion</h3><div>The histogram models based on DCE MRI can predict the progression of enhancing NMDs in HGG following chemoradiotherapy 2 to 3 months in advance.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110394"},"PeriodicalIF":2.1,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854908","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A new type of MCA pulses combining constant and offset-independent adiabaticity for magnetic resonance","authors":"Alexander V. Snadin ,&nbsp;Alexey S. Kiryutin ,&nbsp;Natalya N. Fishman ,&nbsp;Nikita N. Lukzen","doi":"10.1016/j.mri.2025.110385","DOIUrl":"10.1016/j.mri.2025.110385","url":null,"abstract":"<div><div>Adiabatic pulses are widely used in magnetic resonance techniques, and their development and refinement remain very relevant. Adiabatic inverting pulses are highly robust for radiofrequency or microwave magnetic field inhomogeneities and enable manipulation of spins over a large frequency range. In this work, new inverting pulses for spin 1/2 are proposed which combine the adiabaticity remaining constant for the single isochromat throughout the pulse and the same adiabaticity for all isochromats in a given bandwidth, but only at the single instant of time when the frequency of the pulse coincides with the frequency of the isochromat. The dependence of inversion performance of these pulses on peak amplitude of RF field, while preserving the pulse shape, is studied. These pulses may be useful for a number of MRI techniques where inverting pulses are an integral part. A comparison with other widely used adiabatic inverting pulses reveals performance improvements, achieving up to 30–40 % enhancement in inversion efficiency.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110385"},"PeriodicalIF":2.1,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143700931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deformable image registration with strategic integration pyramid framework for brain MRI","authors":"Yaoxin Zhang ,&nbsp;Qing Zhu ,&nbsp;Bowen Xie ,&nbsp;Tianxing Li","doi":"10.1016/j.mri.2025.110386","DOIUrl":"10.1016/j.mri.2025.110386","url":null,"abstract":"<div><div>Medical image registration plays a crucial role in medical imaging, with a wide range of clinical applications. In this context, brain MRI registration is commonly used in clinical practice for accurate diagnosis and treatment planning. In recent years, deep learning-based deformable registration methods have achieved remarkable results. However, existing methods have not been flexible and efficient in handling the feature relationships of anatomical structures at different levels when dealing with large deformations. To address this limitation, we propose a novel strategic integration registration network based on the pyramid structure. Our strategy mainly includes two aspects of integration: fusion of features at different scales, and integration of different neural network structures. Specifically, we design a CNN encoder and a Transformer decoder to efficiently extract and enhance both global and local features. Moreover, to overcome the error accumulation issue inherent in pyramid structures, we introduce progressive optimization iterations at the lowest scale for deformation field generation. This approach more efficiently handles the spatial relationships of images while improving accuracy. We conduct extensive evaluations across multiple brain MRI datasets, and experimental results show that our method outperforms other deep learning-based methods in terms of registration accuracy and robustness.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110386"},"PeriodicalIF":2.1,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143692525","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Radiomics-based MRI model to predict hypoperfusion in lacunar infarction","authors":"Chia-Peng Chang ,&nbsp;Yen-Chu Huang ,&nbsp;Yuan-Hsiung Tsai ,&nbsp;Leng-Chieh Lin ,&nbsp;Jen-Tsung Yang ,&nbsp;Kai-Hsiang Wu ,&nbsp;Po-Han Wu ,&nbsp;Syu-Jyun Peng","doi":"10.1016/j.mri.2025.110366","DOIUrl":"10.1016/j.mri.2025.110366","url":null,"abstract":"<div><h3>Background</h3><div>Approximately 20–30 % of patients with acute ischemic stroke due to lacunar infarction experience early neurological deterioration (END) within the first three days after onset, leading to disability or more severe sequelae. Hemodynamic perfusion deficits may play a crucial role in END, causing growth in the infarcted area and functional impairments, and even poor long-term prognosis. Therefore, it is vitally important to predict which patients may be at risk of perfusion deficits to initiate treatment and close monitoring early, preparing for potential reperfusion. Our goal is to utilize radiomic features from magnetic resonance imaging (MRI) and machine learning techniques to develop a predictive model for hypoperfusion.</div></div><div><h3>Method</h3><div>During January 2011 to December 2020, a retrospective collection of 92 patients with lacunar stroke was conducted, who underwent MRI within 48 h, had clinical laboratory values, follow-up prognosis records, and advanced perfusion image to confirm the presence of hypoperfusion. Using the initial MRI of these patients, radiomics features were extracted and selected from Diffusion Weighted Imaging (DWI), Apparent Diffusion Coefficient (ADC), and Fluid Attenuated Inversion Recovery (FLAIR) sequences. The data was divided into an 80 % training set and a 20 % testing set, and a hypoperfusion prediction model was developed using machine learning.</div></div><div><h3>Result</h3><div>Tthe model trained on DWI + FLAIR sequence showed superior performance with an accuracy of 84.1 %, AUC 0.92, recall 79.5 %, specificity 87.8 %, precision 83.8 %, and F1 score 81.2. Statistically significant clinical factors between patients with and without hypoperfusion included the NIHSS scores and the size of the lacunar infarction. Combining these two features with the top seven weighted radiomics features from DWI + FLAIR sequence, a total of nine features were used to develop a new prediction model through machine learning. This model in test set achieved an accuracy of 88.9 %, AUC 0.91, recall 87.5 %, specificity 90.0 %, precision 87.5 %, and F1 score 87.5.</div></div><div><h3>Conclusion</h3><div>Utilizing radiomics techniques on DWI and FLAIR sequences from MRI of patients with lacunar stroke, it is possible to predict the presence of hypoperfusion, necessitating close monitoring to prevent the deterioration of clinical symptoms. Incorporating stroke volume and NIHSS scores into the prediction model enhances its performance. Future studies of a larger scale are required to validate these findings.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"120 ","pages":"Article 110366"},"PeriodicalIF":2.1,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143692526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DCE-MRI based deep learning analysis of intratumoral subregion for predicting Ki-67 expression level in breast cancer","authors":"Zhimin Ding ,&nbsp;Chengmeng Zhang ,&nbsp;Cong Xia ,&nbsp;Qi Yao ,&nbsp;Yi Wei ,&nbsp;Xia Zhang ,&nbsp;Nannan Zhao ,&nbsp;Xiaoming Wang ,&nbsp;Suhua Shi","doi":"10.1016/j.mri.2025.110370","DOIUrl":"10.1016/j.mri.2025.110370","url":null,"abstract":"<div><h3>Objective</h3><div>To evaluate whether deep learning (DL) analysis of intratumor subregion based on dynamic contrast-enhanced MRI (DCE-MRI) can help predict Ki-67 expression level in breast cancer.</div></div><div><h3>Materials and methods</h3><div>A total of 290 breast cancer patients from two hospitals were retrospectively collected. A k-means clustering algorithm confirmed subregions of tumor. DL features of whole tumor and subregions were extracted from DCE-MRI images based on 3D ResNet18 pre-trained model. The logistic regression model was constructed after dimension reduction. Model performance was assessed using the area under the curve (AUC), and clinical value was demonstrated through decision curve analysis (DCA).</div></div><div><h3>Results</h3><div>The k-means clustering method clustered the tumor into two subregions (habitat 1 and habitat 2) based on voxel values. Both the habitat 1 model (validation set: AUC = 0.771, 95 %CI: 0.642–0.900 and external test set: AUC = 0.794, 95 %CI: 0.696–0.891) and the habitat 2 model (AUC = 0.734, 95 %CI: 0.605–0.862 and AUC = 0.756, 95 %CI: 0.646–0.866) showed better predictive capabilities for Ki-67 expression level than the whole tumor model (AUC = 0.686, 95 %CI: 0.550–0.823 and AUC = 0.680, 95 %CI: 0.555–0.804). The combined model based on the two subregions further enhanced the predictive capability (AUC = 0.808, 95 %CI: 0.696–0.921 and AUC = 0.842, 95 %CI: 0.758–0.926), and it demonstrated higher clinical value than other models in DCA.</div></div><div><h3>Conclusions</h3><div>The deep learning model derived from subregion of tumor showed better performance for predicting Ki-67 expression level in breast cancer patients. Additionally, the model that integrated two subregions further enhanced the predictive performance.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"119 ","pages":"Article 110370"},"PeriodicalIF":2.1,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143634247","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}