Article
Chemistry, Multidisciplinary
Yusuke Asami, Takaaki Yoshimura, Keisuke Manabe, Tomonari Yamada, Hiroyuki Sugimori
Summary: In this study, a deep learning technique was used to analyze the triceps surae muscle and determine muscle volume using an interpolation method. The results showed that the model had high accuracy in predicting the three muscle types and the interpolation method was effective in finding volume. The DSC values of the interpolated images were all above 0.8, indicating good performance of the model in segmentation.
APPLIED SCIENCES-BASEL
(2021)
Article
Medicine, General & Internal
Marlena Rohm, Marius Markmann, Johannes Forsting, Robert Rehmann, Martijn Froeling, Lara Schlaffke
Summary: The study aimed to evaluate the feasibility of automatic segmentation of a heterogeneous database, showing that convolutional neural networks trained on heterogeneous data are able to predict labels for lower leg muscles more accurately.
Article
Clinical Neurology
Lotte Huysmans, Bram De Wel, Kristl G. Claeys, Frederik Maes
Summary: Muscular dystrophies (MD) result in progressive muscle weakness, and fat-sensitive MRI can be used to assess disease progression by quantifying the fat fraction percentage (FF%) per muscle. Precise 3D muscle segmentation is necessary for accurate fat replacement quantification, but manual segmentation is time-consuming. To address this, the researchers used deep learning to train AI models to segment muscles in leg MRI images. The results showed high segmentation accuracy, regardless of fat infiltration levels and MRI field of view.
FRONTIERS IN NEUROLOGY
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Marc-Adrien Hostin, Augustin C. Ogier, Constance P. Michel, Yann Le Fur, Maxime Guye, Shahram Attarian, Etienne Fortanier, Marc-Emmanuel Bellemare, David Bendahan
Summary: Deep learning methods have shown promising results in segmenting lower limb muscle MRIs of healthy individuals, but their performance on patients with neuromuscular disease (NMD) has not been adequately evaluated. This retrospective study aimed to assess the impact of fat infiltration on CNN segmentation of MRIs from NMD patients.
JOURNAL OF MAGNETIC RESONANCE IMAGING
(2023)
Review
Clinical Neurology
Augustin C. Ogier, Marc-Adrien Hostin, Marc-Emmanuel Bellemare, David Bendahan
Summary: This study discusses the limited therapeutic strategies for neuromuscular disorders due to a lack of sensitive biomarkers, and the use of MRI in NMD research. Automatic or semi-automatic segmentation methods are proposed, with a focus on their reliability, reproducibility, and limitations, particularly highlighting deep learning methods.
FRONTIERS IN NEUROLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Shunjie Dong, Zixuan Pan, Yu Fu, Qianqian Yang, Yuanxue Gao, Tianbai Yu, Yiyu Shi, Cheng Zhuo
Summary: In this paper, an enhanced Deformable U-Net (DeU-Net) is proposed for cardiac MRI segmentation. The DeU-Net consists of three modules: Temporal Deformable Aggregation Module (TDAM), Enhanced Deformable Attention Network (EDAN), and Probabilistic Noise Correction Module (PNCM). Experimental results demonstrate the state-of-the-art performance of DeU-Net on the Extended ACDC dataset.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Neurosciences
Meta N. Eek, Jesper Augustsson, Roland Zugner, Roy Tranberg
Summary: This study aimed to investigate the feasibility and usefulness of measuring single-leg vertical jumping in young adults with cerebral palsy (CP). The results showed that this test was able to accurately measure jump height and power generation in individuals with CP, providing additional information about motor function.
Article
Biotechnology & Applied Microbiology
Rula Amer, Jannette Nassar, Amira Trabelsi, David Bendahan, Hayit Greenspan, Noam Ben-Eliezer
Summary: This study presents a new quantitative biomarker for disease severity using quantitative MRI and a deep learning network. By automatically segmenting lower limb anatomy, the severity and spatial distribution of intra-muscular adipose tissue can be accurately measured. The method demonstrates high accuracy and correlation, making it valuable for evaluating muscle pathologies.
BIOENGINEERING-BASEL
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Andrew S. Boehringer, Amirhossein Sanaat, Hossein Arabi, Habib Zaidi
Summary: This study aims to assess the performance of active learning techniques in training a brain MRI glioma segmentation model. The publicly available training dataset for the 2021 RSNA-ASNR-MICCAI Brain Tumor Segmentation Challenge consisting of 1251 multi-institutional, multi-parametric MR images was used. It was demonstrated that active learning can lead to comparable model performance while reducing the amount of manually annotated data required for training.
INSIGHTS INTO IMAGING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Jakob Meglic, Mohammed R. S. Sunoqrot, Tone Frost Bathen, Mattijs Elschot
Summary: Label-set selection significantly impacts the performance of a deep learning-based prostate segmentation model, and using different label-sets has a measurable impact on the model's performance. Deep learning segmentation appears to have higher inter-reader agreement than manual segmentation.
INSIGHTS INTO IMAGING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Joshua R. Astley, Alberto M. Biancardi, Helen Marshall, Paul J. C. Hughes, Guilhem J. Collier, Laurie J. Smith, James A. Eaden, Rod Hughes, Jim M. Wild, Bilal A. Tahir
Summary: This study investigates a dual-channel approach based on deep learning, using H-1-MRI and Xe-129-MRI, to generate lung cavity estimations (LCEs) more accurately. The results show that the dual-channel approach outperforms single-channel alternatives in pulmonary image segmentation and the DL-generated VDPs are statistically indistinguishable from the manually generated ones.
JOURNAL OF MAGNETIC RESONANCE IMAGING
(2023)
Article
Computer Science, Information Systems
Saddam Hussain Khan, Asifullah Khan, Yeon Soo Lee, Mehdi Hassan, Woong Kyo Jeong
Summary: This paper proposes a Region and Edge-based Deep Auto-Encoder (RE-DAE) for the segmentation of shoulder muscle MRI. By combining average and max-pooling operations, this method can extract homogeneous and anatomical information of the muscle and tear regions. Experimental results show that the proposed method achieves high accuracy and feasibility in the segmentation of shoulder muscle MRI.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Neurosciences
Livia Rodrigues, Thiago Junqueira Ribeiro Rezende, Guilherme Wertheimer, Yves Santos, Marcondes Franca, Leticia Rittner
Summary: The hypothalamus is important for sleep regulation, body temperature control, and metabolic homeostasis. Structural abnormalities in the hypothalamus are associated with neuropsychiatric disorders. This study introduces an automated segmentation method for the hypothalamus using a teacher-student-based model. The method shows good performance on different datasets.
Article
Radiology, Nuclear Medicine & Medical Imaging
Marie Franziska Thomas, Florian Kofler, Lioba Grundl, Tom Finck, Hongwei Li, Claus Zimmer, Bjorn Menze, Benedikt Wiestler
Summary: This study developed a generative adversarial network for synthesizing missing magnetic resonance sequences to enable automated segmentation of gliomas. The synthesized images outperformed conventional methods in segmentation performance and qualitative image appearance metrics, particularly when FLAIR sequences were missing.
INVESTIGATIVE RADIOLOGY
(2022)
Article
Medicine, General & Internal
Hao Chen, Na Zhao, Tao Tan, Yan Kang, Chuanqi Sun, Guoxi Xie, Nico Verdonschot, Andre Sprengers
Summary: This study proposes a method for automatic segmentation of knee bone and cartilage using a three-dimensional deep neural network. By adding adversarial loss and restoration network, the contextual information and resolution of the segmentation results can be improved while ensuring accuracy. Through evaluation and comparison, this method achieves good results in knee image segmentation tasks.
FRONTIERS IN MEDICINE
(2022)
Article
Cardiac & Cardiovascular Systems
Jian An, Ye Du, Xiaohong Li, Qingbo Bao, Yanqing Guo, Yang Song, Yongping Jia
Summary: The study demonstrates that early application of sacubitril-valsartan has a protective effect on rats with acute myocardial infarction. The sacubitril-valsartan group showed a reduction in left ventricular volume and collagen fiber range, as well as an increase in ejection fraction compared to the valsartan group.
Article
Computer Science, Information Systems
Dongnan Liu, Chaoyi Zhang, Yang Song, Heng Huang, Chenyu Wang, Michael Barnett, Weidong Cai
Summary: Recent advances in unsupervised domain adaptation (UDA) techniques have successfully improved the generalization ability of data-driven deep learning architectures in cross-domain computer vision tasks. However, existing methods still face challenges in eliminating domain-specific factors from extracted features. To address this issue, we propose a Domain Disentanglement Faster-RCNN (DDF) method that disentangles both global and local features using GTD and ISD modules, respectively. Our DDF method outperforms state-of-the-art methods on four benchmark UDA object detection tasks, demonstrating its effectiveness and wide applicability.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Physiology
Bart Bolsterlee
Summary: A new framework is proposed for comprehensive analysis of the three-dimensional shape and architecture of human skeletal muscles using magnetic resonance and diffusion tensor imaging data. Three key features of the framework include identifying corresponding points inside and on the surface of different muscles, reconstructing average muscle shape and fiber orientations, and utilizing data on between-muscle variation for statistical inferences. The framework can be applied to explore various aspects of muscle mechanics, adaptations, and comparisons between species.
JOURNAL OF APPLIED PHYSIOLOGY
(2022)
Article
Biochemistry & Molecular Biology
Jiefan Gao, Li Wang, Lei Bu, Yangyang Song, Xiao Huang, Jing Zhao
Summary: This research investigates the impact of 1,25VitD3 on the regulatory T cells/Th17 axis and the expressions and concentrations of related cytokines in women with URSA. The results suggest that 1,25VitD3 plays a role in modulating immune responses by strengthening Tregs function and inhibiting inflammatory responses of Th17 cells, which may be beneficial for successful pregnancy outcomes.
CURRENT MOLECULAR PHARMACOLOGY
(2022)
Article
Cardiac & Cardiovascular Systems
Nino Mihatov, Eric A. Secemsky, Dean J. Kereiakes, P. Gabriel Steg, Donald E. Cutlip, Ajay J. Kirtane, Roxana Mehran, Bokai Zhao, Yang Song, C. Michael Gibson, Robert W. Yeh
Summary: In a low-bleeding risk population, stratifying patients based on predicted ischemic risk and the DAPT score best discerned benefit versus harm of extended DAPT duration on ischemic events, bleeding events, and NACE.
CARDIOVASCULAR REVASCULARIZATION MEDICINE
(2022)
Article
Environmental Sciences
Haidong Li, Xiaoqi Tao, Erqun Song, Yang Song
Summary: Iron oxide nanoparticles (IONPs) are an important component of airborne particulate matter and may have adverse effects on respiratory health. This study found that IONPs interact with pulmonary surfactant, leading to reduced cellular uptake and increased cytotoxicity. Mechanistic studies showed that IONPs react with intracellular hydrogen peroxide, generating hydroxyl radicals and causing oxidative stress, lipid accumulation, and inflammation.
Article
Environmental Sciences
Jia Gu, Wei Zhang, Sai Xu, Yang Song, Jun Ma
Summary: This study systematically investigates the role of hydrogen sulfide (H2S) in the transformation of tetrachloro-p-benzoquinone (TCBQ) under UV irradiation. The findings show that the first triplet state of TCBQ can react with bisulfide anion (HS-) to form primary dechlorination product (HS-TriCBQ). The addition of nucleophiles (OH- or HS-) to the hydroxyl or thiol group is the most efficient pathway for dechlorination, and electron transfer from HS- can generate hydrosulfide radical for the dechlorination of TCBQ.
Article
Computer Science, Interdisciplinary Applications
Lei Fan, Arcot Sowmya, Erik Meijering, Yang Song
Summary: In this paper, a novel framework for cancer prediction is proposed, which utilizes self-supervised learning methods to extract features from histopathological whole slide images and considers the overall survival of multiple patients. Experimental results demonstrate the excellent predictive accuracy of this framework.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Anatomy & Morphology
Brian V. Y. Chow, Catherine Morgan, Caroline Rae, David I. Warton, Iona Novak, Suzanne Davies, Ann Lancaster, Gordana C. Popovic, Rodrigo R. N. Rizzo, Claudia Y. Rizzo, Maria Kyriagis, Robert D. Herbert, Bart Bolsterlee
Summary: This study used MRI and artificial intelligence methods to investigate the synchronous growth of human lower leg muscles. The findings showed that the muscles in the lower leg do not grow synchronously, with faster growth in infancy and before the age of 5. This finding is important for early detection of abnormal growth and targeted interventions for muscle-related conditions.
JOURNAL OF ANATOMY
(2023)
Review
Radiology, Nuclear Medicine & Medical Imaging
Melissa T. Hooijmans, Lara Schlaffke, Bart Bolsterlee, Sarah Schlaeger, Benjamin Marty, Valentina Mazzoli
Summary: Due to its sensitivity to soft tissues, MRI is widely used to assess muscle anatomical parameters. qMRI enhances the capabilities of MRI by providing information on muscle composition and function. This review aims to provide an updated overview of qMRI techniques for evaluating muscle structure and composition and their relation to muscle function.
JOURNAL OF MAGNETIC RESONANCE IMAGING
(2023)
Article
Engineering, Biomedical
Alice Hatt, Robert Lloyd, Bart Bolsterlee, Lynne E. Bilston
Summary: Human adipose tissue deforms significantly under physiological loading and impact, making accurate data on strain-dependent stiffness of fat essential for biomechanical models. Using magnetic resonance elastography, the shear modulus of fat was quantified in vivo and found to increase exponentially with compressive strain. This information can be used to develop realistic computational models for impact, injury, and biomechanics.
JOURNAL OF THE MECHANICAL BEHAVIOR OF BIOMEDICAL MATERIALS
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Yiwen Xu, Maurice Pagnucco, Yang Song
Summary: This paper proposes a Decoupled High-frequency semantic Guidance-based GAN (DHG-GAN) for diverse image outpainting, which aims to restore large missing regions surrounding a known region while generating multiple plausible results. Experimental results demonstrate that the proposed method outperforms existing approaches on CelebA-HQ, Place2, and Oxford Flower102 datasets.
COMPUTER VISION - ACCV 2022, PT VII
(2023)
Proceedings Paper
Engineering, Biomedical
Jiayi Zhu, Bart Bolsterlee, Brian V. Y. Chow, Yang Song, Erik Meijering
Summary: Musculoskeletal research often requires muscle segmentation from MRI scans, which can now be automated using deep neural networks. However, these networks generally perform worse than human raters and struggle to generalize to scans of children with cerebral palsy (CP). To address these issues, a new attention-based hybrid network has been proposed, which learns to segment musculoskeletal structures by combining inter- and intra-slice features. This network outperforms other methods and shows robust generalization capabilities on both CP and non-CP pediatric scans.
2022 IEEE INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (IEEE ISBI 2022)
(2022)
Article
Spectroscopy
Tian Peng, Xiao Xue-song, Su Gui-tian, Duan Han-feng, Jin Yao-dong, Song Yang-yang, Huang Tao, Zhang Hang
Summary: Ionic liquids have unique properties that make them attractive in the chemical industry and related fields. However, their classification as green solvents requires further study on their toxicity, degradation, and environmental impact. Therefore, the development of detection methods for ionic liquids in different solvents is of great importance.
SPECTROSCOPY AND SPECTRAL ANALYSIS
(2022)