관심 연구 주제별 최신 논문 요약 — PubMed 최근 2년, 매일 오전 9시 자동 업데이트Latest paper summaries by research topic — PubMed last 2 years, auto-updated daily at 9 AM
의료영상 AI2편
Evaluating machine learning algorithms based on thermal imaging and milk-based parameters to identify subclinical mastitis in dairy cows.
👤 Paudyal Sushil, Neupane Rajesh, Shrestha Bhuwan et al.📰 The veterinary quarterly📅 2026
📝 초록 요약
The objective was to evaluate the potential of machine learning algorithms utilizing thermal imaging and/or milk-based parameters for the identification of subclinical mastitis(SCM) in dairy cows. Holstein dairy cows (n = 194) were enrolled, and milk samples were collected from each quarter separately (n = 776 quarter-level samples) and analyzed for milk-based parameters. Using only thermal imaging features, the ABC model yielded accuracy and precision of 0.82 and 0.67, respectively, with 0.56 AUC. We conclude that SCM prediction using machine learning algorithms is most promising when combined with thermal images and milk-based parameters.
The objective was to evaluate the potential of machine learning algorithms utilizing thermal imaging and/or milk-based parameters for the identification of subclinical mastitis(SCM) in dairy cows. Holstein dairy cows (n = 194) were enrolled, and milk samples were collected from each quarter separately (n = 776 quarter-level samples) and analyzed for milk-based parameters. Four images per cow, representing all four quarters, were captured using a handheld infrared camera. Skin temperatures were extracted using Fluke Connect software, yielding 776 images for analysis. SCM was defined at the quarter level as SCC > 200,000 cells/mL, identifying 139 infected and 637 healthy quarters. The data were analyzed to test Random Forest Classifier(RFC), Logistic Regression Classifier, AdaBoost Classifier(ABC), and Naïve Bayes Classifier algorithms. The RFC models using milk features and temperature features achieved accuracy and precision scores of 0.88 and 0.88, respectively, yielding 0.90 AUC values, with high feature importance for milk lactose and SNF. Using only thermal imaging features, the ABC model yielded accuracy and precision of 0.82 and 0.67, respectively, with 0.56 AUC. We conclude that SCM prediction using machine learning algorithms is most promising when combined with thermal images and milk-based parameters. Future studies with larger datasets and refined methods for thermal image capture and analysis are warranted to validate these findings.
Automated Diagnosis of Breast Cancer Using Deep Learning Techniques Applied to Digital Mammography and Magnetic Resonance Images.
👤 Alakhras Maram, Tayseer Inas, Alquran Hiam et al.📰 Topics in magnetic resonance imaging : TMRI📅 2026
📝 초록 요약
Introduction:Breast cancer is a global issue impacting women's well-being, highlighting the importance of early detection to improve treatment outcomes and decrease mortality rates. This study aimed to assess various AI methodologies to classify breast images into normal, benign, and malignant. Res-Net model achieved the highest accuracy, sensitivity, and specificity of 90.89%, 90.93%, and 95.39%, respectively. ResNet-18 showed the most optimal model for extracting features from mammograms compared with other CNN networks (Mobile-Net, Shuffle-Net, and Inception-V3) while Mobile-Net model was the most suitable in MRI.
Introduction:Breast cancer is a global issue impacting women's well-being, highlighting the importance of early detection to improve treatment outcomes and decrease mortality rates. This study aimed to assess various AI methodologies to classify breast images into normal, benign, and malignant. A hybrid of 4 CNN-pertained networks-Res-Net18, Mobile-Net, Shuffle-Net, and Inception-V3-were applied on 269 mammograms and 267 dynamic contrast-enhanced MRI examinations. Transfer learning was used, adapting the last fully connected layers to classify 3 classes (normal, benign, and malignant) resulting in 12 features. Support vector machine was employed to categorize images. Classifier performances were evaluated using the confusion matrix, accuracy, precision, sensitivity, specificity, F1 score, and area under the receiver operating curve (AUC-ROC). Res-Net model achieved the highest accuracy, sensitivity, and specificity of 90.89%, 90.93%, and 95.39%, respectively. Whereas Shuffle-Net displayed the lowest accuracy of 84.76%. The AUC ranged between 0.95 and 0.97 among pretrained networks while it was higher (0.99) for the hybrid model. For MRI image classification, the Mobile-net network recorded the highest accuracy, sensitivity, and specificity of 88.55%, 88.49%, and 94.22%, respectively, while the Res-Net exhibited the lowest accuracy of 84.35%. The AUC ranged between 0.94 and 0.96 among pretrained networks while it was higher (0.98) for the Hybrid model. ResNet-18 showed the most optimal model for extracting features from mammograms compared with other CNN networks (Mobile-Net, Shuffle-Net, and Inception-V3) while Mobile-Net model was the most suitable in MRI. The effectiveness of deep learning in accurately classifying mammograms and MRI images can be improved by using a hybrid model.
DS-MTNet: A dual-stream multi-task network for histopathology localization and histologic risk stratification from small biopsy specimens of bladder cancer.
👤 Fu Langyuan, Huang Bo, Shi Lingling et al.📰 Computer methods and programs in biomedicine📅 2026
📝 초록 요약
Accurate histologic risk stratification of bladder cancer (BC) from whole-slide images (WSIs) remains challenging due to complex cellular spatial organization and the limited capability of conventional convolution-based methods to capture topological relationships. To address these challenges, this study presents a dual-stream multi-task network, termed DS-MTNet, for knowledge-driven risk stratification and tumor localization in computational pathology. In addition, a multi-task learning framework with a hard weight control mechanism is employed to jointly optimize cancer region detection and risk stratification, thereby reducing task interference and enhancing knowledge sharing across tasks. Extensive experiments conducted on a dataset of 115 BCE WSIs showed that DS-MTNet outperformed the selected comparative methods for histologic risk stratification under the current experimental setting.
Accurate histologic risk stratification of bladder cancer (BC) from whole-slide images (WSIs) remains challenging due to complex cellular spatial organization and the limited capability of conventional convolution-based methods to capture topological relationships. To address these challenges, this study presents a dual-stream multi-task network, termed DS-MTNet, for knowledge-driven risk stratification and tumor localization in computational pathology. DS-MTNet integrates two complementary feature learning streams: a local stream that captures fine-grained morphological representations from histopathological images, and a generalized stream that models intercellular spatial topology by transforming nuclei into graph-structured representations. The graph stream explicitly encodes cellular interactions using a GraphSAGE-based architecture, enabling effective characterization of the tumor microenvironment. In addition, a multi-task learning framework with a hard weight control mechanism is employed to jointly optimize cancer region detection and risk stratification, thereby reducing task interference and enhancing knowledge sharing across tasks. Extensive experiments conducted on a dataset of 115 BCE WSIs showed that DS-MTNet outperformed the selected comparative methods for histologic risk stratification under the current experimental setting. Ablation studies further confirm the effectiveness of the dual-stream design and the multi-task learning strategy. Moreover, DS-MTNet provides interpretable decision evidence through tumor localization maps, Grad-CAM visualizations, and graph-based explanations, facilitating transparent knowledge extraction from histopathological data. These results indicate that DS-MTNet offers an effective and interpretable knowledge-based framework for histologic risk stratification, highlighting its potential applicability in intelligent decision-support systems for computational pathology.
Development and Clinical Validation of a Protocol-Agnostic Machine Learning Platform for Automated Treatment Planning in External Beam Radiation Therapy.
👤 Shade Julie K, Lakshminarayanan Pranav, Hoban Peter et al.📰 Advances in radiation oncology📅 2026
📝 초록 요약
To develop and validate a protocol-agnostic machine learning platform ("Predictive Planning") for knowledge-based planning (KBP) in external beam radiation therapy. Five thousand three hundred thirty-four retrospective photon beam treatment plans (1145 Head and Neck [H&N], 1623 Thoracic, 781 Abdominal, 1785 Pelvis) from 2 institutions were used to develop and internally test general-purpose models for organ at risk (OAR) dose-volume histogram prediction. Models were deployed in a commercial system (Plan AI, Sun Nuclear Corporation), and 72 retrospective clinical plans (18 H&N, 20 Thoracic, 17 Abdominal, 17 Pelvis) were replanned using treatment planning system optimization objectives predicted by the models ("Predictive Plans"; PPs), without objective value changes. Dose metrics were compared for 25, 11, 8, and 7 OARs for H&N, Thoracic, Abdominal, and Pelvis plans, respectively, and for planning target volumes.
To develop and validate a protocol-agnostic machine learning platform ("Predictive Planning") for knowledge-based planning (KBP) in external beam radiation therapy. Five thousand three hundred thirty-four retrospective photon beam treatment plans (1145 Head and Neck [H&N], 1623 Thoracic, 781 Abdominal, 1785 Pelvis) from 2 institutions were used to develop and internally test general-purpose models for organ at risk (OAR) dose-volume histogram prediction. Models were deployed in a commercial system (Plan AI, Sun Nuclear Corporation), and 72 retrospective clinical plans (18 H&N, 20 Thoracic, 17 Abdominal, 17 Pelvis) were replanned using treatment planning system optimization objectives predicted by the models ("Predictive Plans"; PPs), without objective value changes. Dose metrics were compared for 25, 11, 8, and 7 OARs for H&N, Thoracic, Abdominal, and Pelvis plans, respectively, and for planning target volumes. In the plan comparison study, there were no OAR dose metrics for which PPs were statistically significantly greater than clinical plans. For H&N, PP mean dose (Gy) was significantly lower for Brain (5.8 vs 7.3, P = .0003), Brainstem (9.6 vs 13.8, P < .0001), Glottis (26.9 vs 31.9, P = .0017), OpticChiasm (7.3 vs 12.2, P < .0001), Parotid_L/R (19.0/20.8 vs 22.3/25.7, P < .0001/<.0001) and SpinalCord (10.1 vs 15.7, P < .0001). For Abdominal, PP mean dose was significantly lower for Heart (2.2 vs 3.4, P = .0039), Kidney_R (6.2 vs 8.6, P = .002), SpinalCanal (4.8 vs 6.7, P = .0004), and Stomach (9.1 vs 10.9, P = .002). For Pelvis, PP mean dose was significantly lower for Bladder (30.7 vs 33.7, P < .0001), Femur_Head_L/R (13.8/13.9 vs 17.0/16.5, P = .0026/.0046), PenileBulb (15.8 vs 21.9, P = .0039), and Rectum (28.1 vs 34.2, P = .0002). Mean planning target volume coverage was significantly higher for H&N and Thoracic, and equivalent for Abdominal and Pelvis. Predictive planning represents a shift in KBP from protocol-specific models trained with small, uniform data sets to protocol-agnostic, disease-site-specific models trained with large, heterogeneous data sets. Plans generated using optimization objectives predicted by the models had equivalent or superior dosimetric outcomes.
An automatic detection model for spread through air spaces in postoperative pathological sections based on deep learning in NSCLC.
👤 Zhang Yu, Deng Shuzhe, Guo Mingyuan et al.📰 Computer assisted surgery (Abingdon, England)📅 2026
📝 초록 요약
Spread through air spaces (STAS) is recognized as an aggressive pattern of invasion in lung cancer and has been associated with poorer survival outcomes. An artificial intelligence framework consisting of a tumor region segmentation algorithm and an object detection algorithm was developed. Additionally, the AI-derived STAS counts showed substantial agreement with pathologists (interclass correlation coefficient = 0.703). We propose a deep learning-based artificial intelligence framework for automated STAS detection and quantification in digital whole-slide images of NSCLC.
Spread through air spaces (STAS) is recognized as an aggressive pattern of invasion in lung cancer and has been associated with poorer survival outcomes. However, STAS is frequently overlooked or misdiagnosed during routine pathological diagnosis. We analyzed 129 pathological slides from 91 STAS-positive patients with non-small cell lung cancer (NSCLC). An artificial intelligence framework consisting of a tumor region segmentation algorithm and an object detection algorithm was developed. The segmentation algorithm was first used to isolate the main tumor region, followed by STAS detection using an improved object detection model. The segmentation module achieved a Jaccard similarity of 0.846. The proposed object detection algorithm demonstrated superior performance with a precision of 0.738, recall of 0.747 and average precision (AP) of 0.784 and F1 score of 0.742, outperforming other object detection methods. Both segmentation and detection results met diagnostic requirements, with higher accuracy observed in lung adenocarcinoma (LUAD) than in lung squamous cell carcinoma (LUSC). In external validation, the model achieved a precision of 0.670, recall of 0.705 and AP of 0.732 and F1 score of 0.687. Additionally, the AI-derived STAS counts showed substantial agreement with pathologists (interclass correlation coefficient = 0.703). Kaplan-Meier survival analysis revealed that the number of STAS events was significantly associated with disease-free survival (DFS) in stage I LUAD (p < .05). We propose a deep learning-based artificial intelligence framework for automated STAS detection and quantification in digital whole-slide images of NSCLC. This model holds promising potential to assist pathologists in achieving more comprehensive and accurate diagnosis of STAS.
Recent advances in biosensing platforms utilizing exosomal biomarker profiling for cancer diagnosis.
👤 Song Sojin, Hyung Sujin, Lee Jeeyun et al.📰 Biosensors & bioelectronics📅 2026
📝 초록 요약
Tumor-derived exosomes carry multidimensional molecular cargo, including surface proteins, microRNAs, and lipids that encode tumor identity and disease dynamics. These features support their application as biomarkers for liquid biopsy-based cancer diagnostics. We integrate engineering principles, clinical performance metrics, and AI-assisted analysis across complementary biosensing modalities to establish a cross-platform analytical framework. We compare optical platforms based on surface plasmon resonance, localized surface plasmon resonance, and surface-enhanced Raman scattering with photoluminescence- and electrochemical-based platforms in terms of sensitivity, clinical compatibility, and translational potential.
Tumor-derived exosomes carry multidimensional molecular cargo, including surface proteins, microRNAs, and lipids that encode tumor identity and disease dynamics. These features support their application as biomarkers for liquid biopsy-based cancer diagnostics. Circulating tumor DNA undergoes rapid nuclease-mediated degradation, whereas exosomes retain stable molecular information that reflects the proteomic, transcriptomic, and metabolic states of parent tumor cells. However, clinical translation of exosome-based sensing remains limited by variability in isolation, biological heterogeneity, and the analytical difficulty of detecting low-abundance biomarkers in clinical samples. In this review, we examine cancer-specific exosomal signatures across breast, lung, colorectal, and gastric cancers and evaluate biosensing platforms for exosomal biomarker profiling. We integrate engineering principles, clinical performance metrics, and AI-assisted analysis across complementary biosensing modalities to establish a cross-platform analytical framework. We compare optical platforms based on surface plasmon resonance, localized surface plasmon resonance, and surface-enhanced Raman scattering with photoluminescence- and electrochemical-based platforms in terms of sensitivity, clinical compatibility, and translational potential. Furthermore, we examine artificial intelligence (AI)-assisted biosensing frameworks, including classical machine learning classifiers, deep convolutional networks, ensemble models, explainable AI methods, and large language model interfaces. We evaluate how each framework addresses high-dimensional spectral complexity, nonlinear relationships among signals, and inter-patient variability in exosomal data. Finally, we identify remaining challenges, such as the lack of standardized isolation protocols and the absence of large-scale clinical validation. We further highlight minimal residual disease monitoring and early-stage cancer detection as important and underexplored directions for AI-integrated exosomal biosensing in precision oncology.
The advantage of Raman spectroscopy (RS) in renal analysis lies in its capacity to non-invasively acquire label-free molecular vibrational fingerprint information from biological samples. It characterizes alterations in diverse biomolecules, including proteins, lipids, nucleic acids, urates, oxalates, and cytochromes, facilitating the detection of molecular and metabolic abnormalities prior to the manifestation of morphological changes. The review explains how RS deciphers the molecular characteristics of various kidney diseases without using dyes. Comparisons are drawn among surface-enhanced Raman spectroscopy (SERS), tip-enhanced Raman spectroscopy (TERS), confocal Raman microscopy (CRM), and coherent Raman techniques to elucidate their applications in molecular mapping of urine, blood, kidney tissue, and individual cells.
The advantage of Raman spectroscopy (RS) in renal analysis lies in its capacity to non-invasively acquire label-free molecular vibrational fingerprint information from biological samples. It characterizes alterations in diverse biomolecules, including proteins, lipids, nucleic acids, urates, oxalates, and cytochromes, facilitating the detection of molecular and metabolic abnormalities prior to the manifestation of morphological changes. The review explains how RS deciphers the molecular characteristics of various kidney diseases without using dyes. Comparisons are drawn among surface-enhanced Raman spectroscopy (SERS), tip-enhanced Raman spectroscopy (TERS), confocal Raman microscopy (CRM), and coherent Raman techniques to elucidate their applications in molecular mapping of urine, blood, kidney tissue, and individual cells. We also summarize the Raman spectral features of diverse biological samples for the molecular diagnosis of kidney diseases, with attention to the role of advanced analytical methods, including multivariate statistics, machine learning, and deep learning, in spectral interpretation and disease classification. Coupled with high-throughput instrumentation, miniaturized platforms, and artificial intelligence, RS holds potential for precision diagnostics and clinical translation in nephrology.
MMFVS-Net: A triple-symmetric cross-attention network for multimodal optical image fusion and high-accuracy virtual staining of breast cancer tissues.
👤 Shang Lin Wei, Guo Ying Xi, Li Yun Hong et al.📰 Computer methods and programs in biomedicine📅 2026
📝 초록 요약
Virtual staining is based on label-free imaging technology and deep learning, which enables the efficient and non-destructive generation of virtual stained images with effects similar to those of traditional pathological staining, thus holding great application potential in future clinical medical practice. This study took breast cancer tissue as the research object. A multi-branch input network was established to fuse the three types of optical images and generate high-accuracy virtual stained images. This work advances the application of multimodal fusion in virtual staining, address the key challenges of cross-modal interactive information mining and multi-branch model training, and further enhance the accuracy and reliability of virtual staining.
Virtual staining is based on label-free imaging technology and deep learning, which enables the efficient and non-destructive generation of virtual stained images with effects similar to those of traditional pathological staining, thus holding great application potential in future clinical medical practice. However, the information provided by a single imaging technology is limited, which is insufficient to support high-accuracy virtual staining. This study took breast cancer tissue as the research object. Bright-field, auto-fluorescence and phase contrast imaging technology were introduced to simultaneously extract the multimodal complementary information. A multi-branch input network was established to fuse the three types of optical images and generate high-accuracy virtual stained images. To enhance the model's performance, three strategies were employed, including designing a Triple-symmetric cross-attention module to handle the interactive information of multimodal image features, introducing multi-scale discriminator to ensure the model's ability to capture both global features and local details, and applying transfer-learning to address the gradient dilution issue of multi-branches. Through the evaluation of various image similarity metrics and nuclear statistical metrics, the virtual stained images based on multimodal fusion achieved high accuracy and were significantly superior to any single-modal virtual staining approach. Further ablation experiments explained the reason of the aforementioned results and quantitatively demonstrated the improvements brought by the three optimization strategies. This work advances the application of multimodal fusion in virtual staining, address the key challenges of cross-modal interactive information mining and multi-branch model training, and further enhance the accuracy and reliability of virtual staining.
BundleParc: Consistent white matter bundle parcellation without tractography.
👤 Théberge Antoine, Yamani Zineb El, Barakovic Muhamed et al.📰 Medical image analysis📅 2026
📝 초록 요약
Tractometry, also known as tract profiling, is a powerful technique for probing microstructural properties along white matter (WM) tracts. A prerequisite for tractography-based tractometry is bundle parcellation-the subdivision of WM bundles into smaller segments where microstructural measures can be computed. However, existing parcellation methods lack consistency across bundles and timepoints, which reduces reproducibility and limits their utility for both longitudinal and cross-sectional studies. In this work, we introduce BundleParc, a consistent and tractography-free bundle parcellation method.
Tractometry, also known as tract profiling, is a powerful technique for probing microstructural properties along white matter (WM) tracts. A prerequisite for tractography-based tractometry is bundle parcellation-the subdivision of WM bundles into smaller segments where microstructural measures can be computed. However, existing parcellation methods lack consistency across bundles and timepoints, which reduces reproducibility and limits their utility for both longitudinal and cross-sectional studies. Moreover, these methods typically depend on tractography and bundle segmentation, two processes that are computationally expensive and often highly variable. In this work, we introduce BundleParc, a consistent and tractography-free bundle parcellation method. Instead of relying on streamline generation, BundleParc maps fiber orientation distribution function (fODF) volumes directly to label maps. Rigorous evaluation on research and clinical cohorts show that BundleParc is not only much simpler than state-of-the-art tract-based profiling methods, it is also consistently more accurate, robust and reproducible. With these results, BundleParc is a new solution for fast, easy-to-use, and off-the-shelf bundle segmentation and parcellation.
Efficient Deep Learning Models for Predicting Individualized Task Activation From Resting-State Functional Connectivity.
👤 Madsen Soren J, Lee Young-Eun, Quah Shaun K L et al.📰 Human brain mapping📅 2026
📝 초록 요약
Deep learning models have demonstrated the potential to predict task-evoked brain activation from resting-state functional magnetic resonance imaging, offering a pathway toward individualized brain mapping without requiring task-based data. In this study, we systematically evaluate architectural strategies for improving the efficiency and scalability of such models. Using data from the Human Connectome Project, we replicate the BrainSurfCNN framework and introduce two extensions: BrainSERF, which incorporates channel-wise attention through squeeze-and-excitation modules, and BrainSurfGCN, a graph-based model that leverages cortical mesh topology for efficient message passing. Together, these findings demonstrate that incorporating topological and functional structural priors can improve the efficiency of deep learning models without sacrificing accuracy, while also emphasizing that prediction performance is fundamentally limited by the reliability of the underlying neural signals.
Deep learning models have demonstrated the potential to predict task-evoked brain activation from resting-state functional magnetic resonance imaging, offering a pathway toward individualized brain mapping without requiring task-based data. In this study, we systematically evaluate architectural strategies for improving the efficiency and scalability of such models. Using data from the Human Connectome Project, we replicate the BrainSurfCNN framework and introduce two extensions: BrainSERF, which incorporates channel-wise attention through squeeze-and-excitation modules, and BrainSurfGCN, a graph-based model that leverages cortical mesh topology for efficient message passing. Across multiple evaluation metrics, including spatial correlation, Dice score, Dice AUC, and subject identification accuracy, all models achieve comparable predictive performance. Despite similar accuracy, the proposed models offer distinct advantages. BrainSERF provides modest improvements in capturing individual-specific features, while BrainSurfGCN achieves substantial reductions in model size and training time, highlighting a favorable trade-off between performance and computational efficiency. Beyond architectural comparisons, we investigate factors driving variability in prediction accuracy. We find that behavioral task performance, resting-state data quality, and inter-subject variability in task activation jointly constrain prediction fidelity. In particular, contrasts with lower signal reliability and higher variability exhibit reduced predictability across all models. Together, these findings demonstrate that incorporating topological and functional structural priors can improve the efficiency of deep learning models without sacrificing accuracy, while also emphasizing that prediction performance is fundamentally limited by the reliability of the underlying neural signals.
Zero-shot arbitrary-scale super resolution in susceptibility-weighted imaging for cerebral microbleed analysis.
👤 Liu Fengchun, Zhang Rong, Lv Zhongyue et al.📰 Computer methods and programs in biomedicine📅 2026
📝 초록 요약
Susceptibility Weighted Imaging (SWI) plays a pivotal role in detecting cerebral microbleeds (CMBs), key biomarkers of vascular abnormalities and neurodegenerative diseases. To address these challenges, we propose MagNeRF, a zero-shot, single-subject arbitrary-scale SR framework that learns an implicit prior from a single LR volume without external paired training data. Further validation on a T1-weighted dataset and a real-world LR T2*-weighted dataset confirms its robustness across diverse MRI contrasts. By enabling HR reconstruction from LR inputs, MagNeRF reduces patient burden, enhances diagnostic accuracy, and broadens the clinical applicability of SWI.
Susceptibility Weighted Imaging (SWI) plays a pivotal role in detecting cerebral microbleeds (CMBs), key biomarkers of vascular abnormalities and neurodegenerative diseases. Clinical protocols often use larger slice spacing to reduce scan time, resulting in low-resolution SWI with compromised through-plane detail. Although deep learning-based super-resolution (SR) methods show promise, they require large paired high-resolution (HR) and low-resolution (LR) datasets that are difficult to acquire in medical imaging. To address these challenges, we propose MagNeRF, a zero-shot, single-subject arbitrary-scale SR framework that learns an implicit prior from a single LR volume without external paired training data. MagNeRF introduces three key innovations for SWI: (1) a dilated patch-based sampling strategy to improve spatial context and local detail recovery; (2) a spherical sampling strategy to capture the radial gradient decay of magnetic susceptibility signals in SWI; and (3) adaptive loss functions (adaptive multi-scale structural similarity and adaptive mean squared error) that emphasize perceptual fidelity and structural preservation. We rigorously evaluated MagNeRF on two SWI datasets targeting CMBs. Results demonstrate that MagNeRF outperforms state-of-the-art methods, producing reconstructed SWI volumes with high visual fidelity and preserved diagnostically relevant structures. Further validation on a T1-weighted dataset and a real-world LR T2*-weighted dataset confirms its robustness across diverse MRI contrasts. Notably, downstream CMB lesion segmentation using the reconstructed HR images achieves performance closely approaching that of the original HR data, underscoring the clinical utility of the proposed approach. MagNeRF shows significant potential in preserving clinically meaningful microbleed features in SWI. By enabling HR reconstruction from LR inputs, MagNeRF reduces patient burden, enhances diagnostic accuracy, and broadens the clinical applicability of SWI.
Anatomy guided truncated conditional diffusion model for super-resolution arterial spin labeling imaging.
👤 Xu Yunzhi, Zheng Jiaxin, Lin Ruoge et al.📰 NeuroImage📅 2026
📝 초록 요약
This work proposes a super-resolution method for ASL based on an anatomy guided truncated conditional diffusion model, including an anatomy guided data synthesizer, an anatomy guided truncated diffusion module, and a two-stage inference module. The proposed model was developed and validated on retrospective, prospective, and clinical datasets. The proposed model demonstrates superior performance, including 2-18% gains in SSIM and 26-50% reductions in FID on the prospectively acquired data, compared to existing methods. The results suggest that the method offers a promising approach for achieving high-resolution ASL imaging.
Arterial Spin Labeling (ASL) is a non-invasive magnetic resonance imaging technique used to measure cerebral blood flow. However, ASL images typically suffer from low resolution and long acquisition times because of the low signal-to-noise ratio. Therefore, high-resolution ASL images are challenging to acquire but highly desirable. This work proposes a super-resolution method for ASL based on an anatomy guided truncated conditional diffusion model, including an anatomy guided data synthesizer, an anatomy guided truncated diffusion module, and a two-stage inference module. The proposed model was developed and validated on retrospective, prospective, and clinical datasets. Its performance was compared against both conventional and deep learning methods. The proposed model demonstrates superior performance, including 2-18% gains in SSIM and 26-50% reductions in FID on the prospectively acquired data, compared to existing methods. The results suggest that the method offers a promising approach for achieving high-resolution ASL imaging. The proposed model is available on GitHub: https://github.com/yunzxu/ATCDM.
DMRNet: a dynamic multi-scale residual network for Shamrock view and lumbar plexus segmentation.
👤 Cui Haipo, Wang Yuxiang, Lin Liangqing et al.📰 Computer assisted surgery (Abingdon, England)📅 2026
📝 초록 요약
Lumbar plexus block (LPB) is a regional anesthesia technique widely used for hip and knee surgeries. However, despite the assistance of ultrasound guidance, the complex anatomical structure of the lumbar plexus poses significant challenges for anesthesiologists during the procedure. To accurately identify the lumbar plexus located in the posterior third of the psoas major in the Shamrock view, a deep learning-based segmentation model named DMRNet was proposed. These findings suggest that DMRNet may serve as a useful assistive tool for sonoanatomical recognition in Shamrock view ultrasound images and may provide educational support for ultrasound-guided LPB training.
Lumbar plexus block (LPB) is a regional anesthesia technique widely used for hip and knee surgeries. However, despite the assistance of ultrasound guidance, the complex anatomical structure of the lumbar plexus poses significant challenges for anesthesiologists during the procedure. To accurately identify the lumbar plexus located in the posterior third of the psoas major in the Shamrock view, a deep learning-based segmentation model named DMRNet was proposed. This model is designed to precisely delineate muscles, nerves, and bony structures in Shamrock view ultrasound images. DMRNet integrates several innovative modules, including Adaptive Multi-Scale Dilated (AMD) Module that enhances the model's ability to capture multi-scale features; Dense Attention Residual (DAR) Module that adaptively selects salient feature regions; Attention-Enhanced Hybrid (AEH) Module that emphasizes critical features while suppressing irrelevant ones; and two attention mechanisms, Boundary-Aware Spatial Attention Mechanism (BASA) and the Enhanced Residual Multi-Head Attention Mechanism (ER-MHA), that improve the model's capacity to recognize complex contextual patterns. Experimental results demonstrated that DMRNet achieved a mean Intersection over Union of 0.863 and a mean Dice coefficient of 0.926 across all target structures, outperforming other state-of-the-art models. These findings suggest that DMRNet may serve as a useful assistive tool for sonoanatomical recognition in Shamrock view ultrasound images and may provide educational support for ultrasound-guided LPB training. A dynamic multi-scale residual network is proposed for lumbar lexus (LP) segmentation.Two attention mechanisms were improved, and three modules were proposed to improve downsampling, skip connection, and upsampling.We built our own Shamrock view ultrasound database containing the lumbar plexus nerves, and the Anesthesiologist physician of the anesthesiology department annotated it using Labelme.External validation was performed using the Ultrasonic Brachial Plexus Dataset (UBPD).
A systematic review of transformer-enhanced UNet architectures for 3D medical image segmentation: Trends, challenges, and the ATD-TᵣEEv framework.
👤 Krishnaveni Tuniki, Kollem Sreedhar📰 Computational biology and chemistry📅 2026
📝 초록 요약
Accurate 3D medical image segmentation plays an important role in diagnosing and treating tumors; however, it is difficult due to biological intricacies, noise within images, and insufficient training data. UNet-based models have demonstrated impressive results but tend to lack the ability to capture long-range dependencies in volumetric data. This systematic review highlights current trends in recent UNet architectures utilizing Transformers for 3D medical imaging segmentation and tumor detection, focusing on architecture design, clinical applicability, and research challenges. In total, 87 top-tier articles were selected and analyzed according to the developed taxonomy ATD-TᵣEEv, which is a six-dimension model involving aspects such as Architecture, Token/Attention, Data and Modality, Training/Pretraining, Efficiency, and Robustness of Evaluation.
Accurate 3D medical image segmentation plays an important role in diagnosing and treating tumors; however, it is difficult due to biological intricacies, noise within images, and insufficient training data. UNet-based models have demonstrated impressive results but tend to lack the ability to capture long-range dependencies in volumetric data. This systematic review highlights current trends in recent UNet architectures utilizing Transformers for 3D medical imaging segmentation and tumor detection, focusing on architecture design, clinical applicability, and research challenges. A systematic literature review was carried out based on the PRISMA guideline, involving research articles in the period 2016-2025 published in IEEE Xplore, SpringerLink, Elsevier, PubMed, Wiley, and arXiv. In total, 87 top-tier articles were selected and analyzed according to the developed taxonomy ATD-TᵣEEv, which is a six-dimension model involving aspects such as Architecture, Token/Attention, Data and Modality, Training/Pretraining, Efficiency, and Robustness of Evaluation. Transformer-based hybrid models like TransUNet, FET-UNet, DS-UNETR+ +, VMAXL-UNet, and EfficientNet-Enhanced UNet demonstrated better Dice scores (0.81-0.96) in standard datasets such as BraTS (2019-2021), KiTS19, Synapse, and CE-MRI because of their multi-level attention and modeling of global contexts. Nevertheless, there are still some problems that need addressing, such as GPU memory usage, lack of statistical validation, data imbalance, and insufficient evaluation under noisy or incomplete clinical conditions. In addition to summarizing previous works, this review offers an analytical taxonomy (ATD-TᵣEEv) which incorporates architectural tendencies, analyzes efficiency-fairness trade-offs, and considers clinical translation problems. The presented taxonomy can be used as a basis for further research on efficient, fair, and clinically reliable transformer-based image segmentation approaches.