Article
Computer Science, Artificial Intelligence
Han Xu, Jiteng Yuan, Jiayi Ma
Summary: This study proposes a novel method called MURF that mutually reinforces image registration and fusion. MURF consists of three modules, which progressively correct global and local offsets during the coarse-to-fine registration process and incorporate texture enhancement into image fusion. Extensive experiments validate the superiority and universality of MURF on different types of multi-modal data.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Alan Q. Wang, Evan M. Yu, Adrian V. Dalca, Mert R. Sabuncu
Summary: KeyMorph is a deep learning-based image registration framework that enhances robustness and interpretability by automatically detecting corresponding keypoints, incorporating symmetries, and handling image translations. This framework shows excellent performance in solving the registration problem of multi-modal brain MRI scans.
MEDICAL IMAGE ANALYSIS
(2023)
Article
Computer Science, Artificial Intelligence
Max Blendowski, Lasse Hansen, Mattias P. Heinrich
Summary: Deep learning methods have recently achieved good performance in medical image registration, with significant improvements possible by disentangling feature learning and deformation estimation. The proposed method shows promising results in multi-modal registration and can even handle small and weakly-labeled training datasets.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Computer Science, Artificial Intelligence
Chenyu Lian, Xiaomeng Li, Lingke Kong, Jiacheng Wang, Wei Zhang, Xiaoyang Huang, Liansheng Wang
Summary: CoCycleReg is a new method that unifies image registration and translation through collaborative cycle-consistency. By leveraging cycle-consistency, each part can benefit from the other, leading to improved performance in image registration and translation.
Article
Geochemistry & Geophysics
Dou Quan, Shuang Wang, Yu Gu, Ruiqi Lei, Bowu Yang, Shaowei Wei, Biao Hou, Licheng Jiao
Summary: This article proposes a deep feature correlation learning network (Cnet) for multi-modal remote sensing image registration. The network enhances feature representation by focusing on meaningful features and improves network training stability and decreases the risk of overfitting through a designed feature correlation loss function. Extensive experimental results demonstrate the effectiveness and robustness of the proposed method.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Biology
Jonas Cordes, Thomas Enzlein, Christian Marsching, Marven Hinze, Sandy Engelhardt, Carsten Hopf, Ivo Wolf
Summary: M(2)aia is an extensible open-source application that provides interactive and memory-efficient data access and signal processing for multiple large MSI datasets. It extends MITK and offers features such as fast visual interaction, image segmentation, 3D image reconstruction, and multi-modal registration, making it suitable for a wide range of MSI analysis tasks.
Article
Radiology, Nuclear Medicine & Medical Imaging
Rajeev Nowrangi, Laurie A. Perry, Jennifer Regan, David Hulefeld, Sarah O'Brien, Timothy J. OConnor, Alexander J. Towbin
Summary: Transferring medical imaging studies between different institutions is a common practice in today's medical field. The traditional method of using physical media and human couriers for image transfer is slow, unreliable, and outdated. To overcome these issues, various electronic, cloud-based solutions have been developed by image-sharing vendors. However, a new challenge arises in the difficulty of sending or receiving images across different image-sharing platforms. This study presents a solution that allows for image sharing across multiple vendor platforms.
JOURNAL OF DIGITAL IMAGING
(2022)
Article
Computer Science, Artificial Intelligence
Junkang Zhang, Yiqian Wang, Ji Dai, Melina Cavichini, Dirk-Uwe G. Bartsch, William R. Freeman, Truong Q. Nguyen, Cheolhong An
Summary: This article proposes a two-step method based on deep convolutional networks for multi-modal retinal image registration. By combining coarse and fine alignment steps with techniques such as vessel segmentation, feature detection and description, it improves the accuracy and robustness of registration. Additionally, it addresses the challenges of inconsistent modalities and lack of labeled training data through an unsupervised learning framework.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Chemistry, Analytical
Christoff M. Heunis, Filip Suligoj, Carlos Fambuena Santos, Sarthak Misra
Summary: This study introduces a multi-modal sensing and feedback framework for assisting clinicians during endovascular surgeries and catheterization procedures. The framework utilizes advanced imaging and sensing sub-systems to produce a radiation-free 3D visualization of endovascular catheters and surrounding vasculature. Experiments conducted in porcine limbs show reliable reconstruction of vasculature and catheters during externally-induced limb motions.
Article
Computer Science, Interdisciplinary Applications
Victor M. Campello, Polyxeni Gkontra, Cristian Izquierdo, Carlos Martin-Isla, Alireza Sojoudi, Peter M. Full, Klaus Maier-Hein, Yao Zhang, Zhiqiang He, Jun Ma, Mario Parreno, Alberto Albiol, Fanwei Kong, Shawn C. Shadden, Jorge Corral Acero, Vaanathi Sundaresan, Mina Saber, Mustafa Elattar, Hongwei Li, Bjoern Menze, Firas Khader, Christoph Haarburger, Cian M. Scannell, Mitko Veta, Adam Carscadden, Kumaradevan Punithakumar, Xiao Liu, Sotirios A. Tsaftaris, Xiaoqiong Huang, Xin Yang, Lei Li, Xiahai Zhuang, David Vilades, Martin L. Descalzo, Andrea Guala, Lucia La Mura, Matthias G. Friedrich, Ria Garg, Julie Lebel, Filipe Henriques, Mahir Karakas, Ersin Cavus, Steffen E. Petersen, Sergio Escalera, Santi Segui, Jose F. Rodriguez-Palomares, Karim Lekadir
Summary: The emergence of deep learning has advanced cardiac magnetic resonance segmentation, yet current models lack generalizability. A recent competition emphasized the importance of data augmentation in training deep learning models and provided new data resources for future research.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2021)
Article
Computer Science, Interdisciplinary Applications
Chaewoo Kim, Oguzcan Bekar, Hyunseok Seo, Sang-Min Park, Deukhee Lee
Summary: Automatic medical image segmentation plays a crucial role in computer-assisted surgery. In this study, we propose a model for standardizing computed tomography (CT) images to improve the consistency of medical images. By integrating an image-to-image translation network and a generative adversarial network, we are able to convert diverse CT images into standardized images for more accurate U-Net segmentation. The performance of our model is evaluated through visualization, numerical analysis, and comparison with other methods.
JOURNAL OF COMPUTATIONAL DESIGN AND ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Xin Deng, Enpeng Liu, Shengxi Li, Yiping Duan, Mai Xu
Summary: Multi-modal image registration aims to align two images from different modalities by separating alignment features from non-alignment features. The proposed DCSC model and InMIR-Net use deep learning to improve registration accuracy and efficiency. The accompanying guidance network further enhances feature extraction. Extensive experiments demonstrate the effectiveness of the method on various multi-modal image datasets.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Information Systems
Yanxia Liu, Wenqi Wang, Yuhong Li, Haoyu Lai, Sijuan Huang, Xin Yang
Summary: Deformable multi-modal medical image registration aligns anatomical structures across different modalities, but unsupervised methods face challenges in measuring image similarity and fusing representations. To address this, a novel unsupervised multi-modal adversarial registration framework is proposed, leveraging image-to-image translation. Two improvements are introduced: a geometry-consistent training scheme to encourage modality mapping and a semi-shared multi-scale registration network to accurately handle large deformation. Extensive experiments on brain and pelvic datasets demonstrate the superiority of the proposed method, highlighting its potential in clinical applications.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Chemistry, Multidisciplinary
Yuezhou Wu, Changjiang Liu
Summary: This paper studies the use of infrared heat-transfer imaging camera and visible-light camera to enhance the pilot's vision by obtaining dynamic hyperspectral images of flight approach scenes under low visibility. It proposes a multi-modal image registration method using virtual scene, RoI feature extraction, and contour angle orientation based on virtual-reality-fusion, which reduces the registration area, computation amount, and improves real-time registration accuracy. The contour angle orientation maintains the geometric deformation of multi-source images well, and the virtual reality fusion technology effectively deletes incorrectly matched point pairs. By integrating redundant and complementary information from multi-modal images, the visual perception of pilots during the approach process is enhanced as a whole.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Miguel Oliveira, Eurico Pedrosa, Andre Pinto de Aguiar, Daniela Ferreira Pinto Dias Rato, Filipe Neves dos Santos, Paulo Dias, Vitor Santos
Summary: This paper presents a novel calibration methodology for multi-sensor, multi-modal robotic systems. The approach formulates the calibration as an extended optimization problem and makes use of a topological representation to recalculate the sensor poses. This atomic transformations optimization method (ATOM) is applicable to different calibration problems and achieves comparable accuracy to state-of-the-art methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Clinical Neurology
Ying Wang, Ivan C. Zibrandtsen, Richard H. C. Lazeron, Johannes P. van Dijk, Xi Long, Ronald M. Aarts, Lei Wang, Johan B. A. M. Arends
Summary: This study aimed to address the reliability issues in the diagnosis of nonconvulsive status epilepticus (NCSE) through electroencephalography (EEG). It identified typical pitfalls in EEG analysis and suggested strategies to avoid them. The study found that short ictal discharges, abnormal background activity, and continuous discharges were prone to misinterpretation. It recommended a longer duration criterion for NCSE-EEGs and the use of personalized algorithms and context-dependent alarm thresholds to improve interpretations.
CLINICAL EEG AND NEUROSCIENCE
(2023)
Editorial Material
Computer Science, Artificial Intelligence
Chen Chen, Caifeng Shan, Ronald M. Aarts, Xi Long
JOURNAL OF AMBIENT INTELLIGENCE AND SMART ENVIRONMENTS
(2022)
Article
Biochemical Research Methods
Lynn-Jade S. Jong, Naomi de Kruif, Freija Geldof, Dinusha Veluponnar, Joyce Sanders, Marie-Jeanne T. F. D. Vrancken Peeters, Frederieke van Duijnhoven, Henricus J. C. M. Sterenborg, Behdad Dashtbozorg, Theo J. M. Ruers
Summary: This study evaluates the use of hyperspectral imaging to discriminate healthy tissue from tumor tissue in breast-conserving surgery. Convolutional neural networks are developed and evaluated using tissue slices, achieving high accuracy in predicting tissue percentages on the resection surface.
BIOMEDICAL OPTICS EXPRESS
(2022)
Article
Engineering, Biomedical
Shaoxiong Sun, Erik Bresch, Jens Muehlsteff, Lars Schmitt, Xi Long, Rick Bezemer, Igor Paulussen, Gerrit J. Noordergraaf, Ronald M. Aarts
Summary: This study proposes a non-invasive method to estimate systolic blood pressure (SBP) with high accuracy during surgery. By using ECG, PPG, and intermittent SBP measurements, the method is able to estimate SBP every 30 seconds, improving BP monitoring in the operating room and enhancing patient outcomes and experiences.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Computer Science, Artificial Intelligence
Wenli Liang, Yuanjian Yang, Fangyu Li, Xi Long, Caifeng Shan
Summary: By exploiting the complementary information of RGB modality and thermal modality, the proposed Mask-guided Modality Difference Reduction Network (MMDRNet) aims to minimize the modality discrepancy within foreground regions and achieve more discriminative representations for foreground pixels in RGB-thermal semantic segmentation. Additionally, the Dynamic Task Balance (DTB) method dynamically balances the modality difference reduction task and semantic segmentation task. Experimental results demonstrate the superiority of the proposed strategies on two datasets.
Article
Biochemical Research Methods
Freija Geldof, Mark Witteveen, Henricus J. C. M. Sterenborg, Theo J. M. Ruers, Behdad Dashtbozorg
Summary: A compact side-firing fiber probe was developed for tissue discrimination during colorectal cancer surgery using diffuse reflectance spectroscopy. The optical behavior of the probe was compared to flat-tipped fibers through simulations and experimental measurements. The developed probe and classification algorithm achieved an accuracy of 0.92 for discriminating tumor tissue from healthy tissue.
BIOMEDICAL OPTICS EXPRESS
(2023)
Article
Computer Science, Information Systems
Zheng Peng, Gabriele Varisco, Xi Long, Rong-Hao Liang, Deedee Kommers, Ward Cottaar, Peter Andriessen, Carola van Pul
Summary: The aim of this study is to develop a explainable late-onset sepsis (LOS) prediction algorithm using continuous multi-channel physiological signals. Features from different signal channels (HRV, respiration, and motion) were extracted and fed into machine learning classifiers. The best performance was achieved by an extreme gradient boosting classifier combining features from all signal channels, with an AUC of 0.88.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Oncology
Lynn-Jade S. Jong, Anouk L. Post, Dinusha Veluponnar, Freija Geldof, Henricus J. C. M. Sterenborg, Theo J. M. Ruers, Behdad Dashtbozorg
Summary: In order to assess the resection margins during breast-conserving surgeries, it is crucial to use hyperspectral imaging technique to accurately detect tumor tissues on the lumpectomy resection surface. We developed a classification model with a sensitivity of 94% and a specificity of 85% using this technique, indicating its potential for aiding surgeons in margin assessment.
Article
Oncology
Dinusha Veluponnar, Lisanne L. de Boer, Freija Geldof, Lynn-Jade S. Jong, Marcos Da Silva Guimaraes, Marie-Jeanne T. F. D. Vrancken Peeters, Frederieke van Duijnhoven, Theo Ruers, Behdad Dashtbozorg
Summary: During breast-conserving surgeries, accurate evaluation of tumor margins is challenging, leading to potential need for additional surgery or boost radiotherapy. The use of computer-aided delineation of tumor boundaries in ultrasound images shows promising results in predicting positive and close margins. There is a clinical need for an accurate and rapid margin assessment tool, and computer-aided ultrasound evaluation offers a potential solution.
Article
Engineering, Biomedical
Jiali Xie, Pedro Fonseca, Johannes van Dijk, Sebastiaan Overeem, Xi Long
Summary: By adding audio-based snoring features, the non-obtrusive assessment of obstructive sleep apnea (OSA) can be improved by estimating the apnea-hypopnea index (AHI) and classifying OSA severity. Novel features were proposed to quantify temporal changes between snores (snore rate variability) and trends in snore energy, based on the assessment of snore sounds from audio signals over the full night. These features, combined with age, body mass index (BMI), and features from literature, were used to train an extreme gradient boosting algorithm for AHI estimation and OSA severity classification.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Chemistry, Analytical
Emad Arasteh, Esther S. Veldhoen, Xi Long, Maartje van Poppel, Marjolein van der Linden, Thomas Alderliesten, Joppe Nijman, Robbin de Goederen, Jeroen Dudink
Summary: This paper presents a novel approach for simultaneously estimating children's respiratory rate and heart rate using ultra-wideband radar and deep transfer learning algorithm. By processing radar signals and transforming them into images, and analyzing these images using a pre-trained VGG-16 model, the authors achieved good prediction results.
Article
Medicine, General & Internal
Jiali Xie, Pedro Fonseca, Johannes P. Van Dijk, Xi Long, Sebastiaan Overeem
Summary: Sleep apnea, a common sleep disorder, can be detected using electrocardiographic (ECG) signals. However, respiratory effort is a better alternative to ECG-derived respiration (EDR) for breath signal replacement. Our study with 198 patients showed improved performance in detection, estimation, and classification of sleep apnea using respiratory effort instead of EDR.
Article
Pediatrics
Dandan Zhang, Zheng Peng, Carola Van Pul, Sebastiaan Overeem, Wei Chen, Jeroen Dudink, Peter Andriessen, Ronald M. Aarts, Xi Long
Summary: This study explores the effectiveness of incorporating video-based actigraphy analysis alongside cardiorespiratory signals for classifying the sleep states of preterm infants. The results suggest that combining video-based actigraphy with cardiorespiratory signals can potentially enhance the performance of sleep-state classification.
Article
Engineering, Electrical & Electronic
Chunjiao Dong, Tianchun Ye, Xi Long, Ronald M. Aarts, Johannes P. van Dijk, Chunheng Shang, Xiwen Liao, Wei Chen, Wanlin Lai, Lei Chen, Yunfeng Wang
Summary: Using monitoring devices is crucial for preventing injuries and deaths. Wearable sensors are currently used to detect seizures and alert caregivers, but the development of these devices is challenging due to the manual labeling of collected data. In this study, we collected data from epileptic patients using our proposed bracelet and introduced an algorithm and a machine learning model to label and classify seizure data. The results demonstrate that our bracelet efficiently captures seizures.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Engineering, Biomedical
Yao Guo, Xinyu Jiang, Linkai Tao, Long Meng, Chenyun Dai, Xi Long, Feng Wan, Yuan Zhang, Johannes van Dijk, Ronald M. Aarts, Wei Chen, Chen Chen
Summary: This study proposes a hybrid system combining unsupervised learning and supervised learning modules for automatic detection of epilepsy seizures. By utilizing the unsupervised learning module, the workload of data labelling can be significantly reduced. The proposed system is evaluated using the CHB-MIT scalp EEG dataset and achieves a competitive performance.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2022)