Review
Computer Science, Artificial Intelligence
Lei Li, Veronika A. Zimmer, Julia A. Schnabel, Xiahai Zhuang
Summary: This paper provides a systematic review of computing methods for the segmentation and quantification of left atrial (LA) cavity, wall, scar, and ablation gap from late gadolinium enhancement magnetic resonance imaging (LGE MRI), as well as the related literature for atrial fibrillation (AF) studies. The review suggests that there is still a large scope for further algorithmic developments in this field due to performance issues related to the high variability of enhancement appearance and differences in image acquisition.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Yashu Liu, Wei Wang, Gongning Luo, Kuanquan Wang, Dong Liang, Shuo Li
Summary: This study proposes an automatic and efficient left atrium (LA) segmentation model based on a convolutional neural network for late gadolinium-enhanced magnetic resonance imaging (LGE MRI). The model utilizes an uncertainty-guided symmetric multilevel supervision (SML) structure to learn multiscale representations of the LA and improve the segmentation accuracy on the surface. Experimental results show that the proposed model achieves better or comparable performance compared to state-of-the-art models.
Article
Engineering, Biomedical
Michail Mamalakis, Pankaj Garg, Tom Nelson, Justin Lee, Andrew J. Swift, James M. Wild, Richard H. Clayton
Summary: Patients with myocardial infarction are at elevated risk of sudden cardiac death, and scar tissue arising from infarction is known to play a role. The accurate identification of scars therefore is crucial for risk assessment, quantification and guiding interventions. We propose a fully automatic framework that develops 3D anatomical models of the left ventricle with border zone and core scar regions that are free from bias effect.
COMPUTERIZED MEDICAL IMAGING AND GRAPHICS
(2023)
Article
Engineering, Biomedical
Feiyan Li, Weisheng Li, Xinbo Gao, Rui Liu, Bin Xiao
Summary: A comprehensive information integration network (CII-Net) is proposed to segment the left atrium (LA) from late gadolinium-enhanced (LGE) cardiac magnetic resonance (CMR) images, which can more completely capture the comprehensive image characteristics. Compared with other methods, CII-Net achieves an average DICE score of 91.9%, average Jaccard score of 85.1%, average Hausdorff distance of 5.924 mm, and average symmetric surface distance of 0.993 mm without post-processing, demonstrating the potential for clinical application.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Cardiac & Cardiovascular Systems
Yaacoub Chahine, Bahareh Askari-Atapour, Kirsten T. Kwan, Carter A. Anderson, Fima Macheret, Tanzina Afroze, Savannah F. Bifulco, Matthew D. Cham, Karen Ordovas, Patrick M. Boyle, Nazem Akoum
Summary: This study found an association between epicardial adipose tissue (EAT) and obesity, as well as left atrial (LA) volume and fibrosis. However, there was no clear spatial overlap between EAT and fibrotic areas, suggesting a systemic or paracrine mechanism.
FRONTIERS IN CARDIOVASCULAR MEDICINE
(2022)
Article
Computer Science, Information Systems
Asma Kausar, Imran Razzak, Mohammad Ibrahim Shapiai, Amin Beheshti
Summary: In this paper, a novel deep learning architecture is proposed for the segmentation of the left atrium from MRI volumes. The method incorporates a residual learning based encoder-decoder network, along with a loss function and parameter adjustments to address class imbalance and unavailability of large medical imaging dataset. Experimental results demonstrate significant improvement in segmentation performance compared to state-of-the-art approaches, with fewer parameters used, potentially supporting cardiac diagnosis and surgery.
MULTIMEDIA SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Lei Li, Veronika A. Zimmer, Julia A. Schnabel, Xiahai Zhuang
Summary: This study introduces a new framework named AtrialJSQnet that can conduct left atrial segmentation, scar projection onto the left atrial surface, and scar quantification simultaneously, achieving competitive performance on public datasets. Additionally, by incorporating a shape attention mechanism and spatial encoding loss, the method demonstrates significant effects in reducing noisy patches and improving task performance.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Biology
Mattia Corti, Alberto Zingaro, Luca Dede', Alfio Maria Quarteroni
Summary: This study analyzes the hemodynamics of the left atrium, comparing differences between healthy individuals and patients with atrial fibrillation. Using patient-specific geometries, a computational simulation of blood flow dynamics in the left atria is conducted. A novel procedure for computing boundary data for 3D hemodynamic simulations is introduced, which is particularly helpful in the absence of clinical measurements. Various fluid dynamics indicators are evaluated for atrial hemodynamics and validated against clinical measurements. The impact of geometric and clinical characteristics on the risk of thrombosis is investigated, and a new indicator called "age stasis" is proposed to highlight the correlation between thrombus formation and atrial fibrillation.
COMPUTERS IN BIOLOGY AND MEDICINE
(2022)
Article
Physiology
Lisa A. Gottlieb, Fanny Vaillant, Emma Abell, Charly Belterman, Virginie Loyer, Dounia El Hamrani, Jerome Naulin, Marion Constantin, Bruno Quesson, Bastiaan J. Boukens, Ruben Coronel, Lukas R. C. Dekker
Summary: Localized pulmonary vein scar only facilitates the inducibility of atrial fibrillation in combination with increased left atrial pressure. This is associated with changes in tissue excitability remote from the pulmonary vein scar.
FRONTIERS IN PHYSIOLOGY
(2021)
Article
Cardiac & Cardiovascular Systems
Vanessa Sciacca, Thomas Fink, Hermann Koerperich, Leonard Bergau, Denise Guckel, Flemming Nischik, Jan Eckstein, Martin Braun, Mustapha El Hamriti, Guram Imnadze, Misagh Piran, Philipp Sommer, Christian Sohns
Summary: This study evaluated the procedural data and the scar formation in atrial fibrillation (AF) patients who underwent ablation using high-power and short-duration (vHPSD) energy delivery. The results showed that vHPSD ablation resulted in shorter procedure duration, high efficacy, and durable pulmonary vein (PV) lesions. Cardiac magnetic resonance imaging (MRI) demonstrated homogeneous and contiguous scar formation. The advantages of vHPSD ablation are the quick and effective procedure, while the limitation is the small sample size.
Article
Radiology, Nuclear Medicine & Medical Imaging
Xin Tian, Cen Wang, Duo Gao, Bu-Lang Gao, Cai-Ying Li
Summary: This study aimed to assess the morphological and functional features of the left atrium (LA) and the left atrial appendage (LAA) in patients with atrial fibrillation (AF) using computed tomography angiography (CTA) images. The results showed that AF patients had a larger minor axis of the LAA orifice and a more circular LAA orifice compared to the control group. The LAA orifice area and perimeter were positively correlated with LAA volume change. Female patients had larger LAA orifice major and minor axes, area, perimeter, and LAA depth compared to male patients in the AF group.
QUANTITATIVE IMAGING IN MEDICINE AND SURGERY
(2022)
Article
Engineering, Biomedical
Michail Mamalakis, Pankaj Garg, Tom Nelson, Justin Lee, Jim M. Wild, Richard H. Clayton
Summary: MA-SOCRATIS is an unsupervised automatic pipeline for segmenting left ventricular myocardium and scar regions from LGE-MRI images. It employs two different pipelines for myocardial and scar segmentation, achieving robust and accurate performance without the need for training or tuning.
COMPUTERIZED MEDICAL IMAGING AND GRAPHICS
(2021)
Article
Medicine, General & Internal
Mireia Ble, Begona Benito, Elisa Cuadrado-Godia, Silvia Perez-Fernandez, Miquel Gomez, Aleksandra Mas-Stachurska, Helena Tizon-Marcos, Lluis Molina, Julio Marti-Almor, Merce Cladellas
Summary: This study aimed to detect atrial disease in CrS patients through analyzing atrial size and function. The results showed that patients with AF had larger atrial volume and worse atrial function, with an independent association between detection of AF and atrial volume, LAEF, and strain.
JOURNAL OF CLINICAL MEDICINE
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Chia-Hung Yang, Hao-Tien Liu, Hui-Ling Lee, Fen-Chiung Lin, Chung-Chuan Chou
Summary: This study found that left atrial function plays an important role in the absence of atrial fibrillation (AF) in patients with left atrial dimension greater than or equal to 50 mm, and the late diastolic component of left atrial strain rate is the only independent variable associated with AF.
QUANTITATIVE IMAGING IN MEDICINE AND SURGERY
(2022)
Article
Cardiac & Cardiovascular Systems
Chan Soon Park, Eue-Keun Choi, So-Ryoung Lee, Hyo-Jeong Ahn, Soonil Kwon, Sunhwa Kim, Suk Ho Sohn, Jae Woong Choi, Ho Young Hwang, Seil Oh
Summary: In patients with persistent AF and a large LA, there was no significant difference in prognosis between RFCA, CBA, and thoracoscopic maze procedures. Early recurrence during the blanking period predicted late recurrence in catheter ablation, but not in thoracoscopic maze surgery.
FRONTIERS IN CARDIOVASCULAR MEDICINE
(2022)
Review
Biology
Md. Kamrul Hasan, Md. Asif Ahamad, Choon Hwai Yap, Guang Yang
Summary: The Computer-aided Diagnosis or Detection (CAD) approach for skin lesion analysis is a promising field of research that can help reduce the burden and cost of skin cancer screening. This article provides a comprehensive literature survey and review of 594 publications on skin lesion analysis from 2011 to 2022, covering segmentation and classification methods. The article analyzes and summarizes different aspects of these publications to provide valuable information for the development of CAD systems.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Biology
Sanjay Kumar, Abhishek Mallik, Akshi Kumar, Javier Del Ser, Guang Yang
Summary: Electrocardiogram (ECG) is a widely used non-invasive technique to diagnose cardiovascular diseases. In this work, a deep learning and fuzzy clustering based approach (Fuzz-ClustNet) is proposed for Arrhythmia detection from ECG signals. The collected ECG signals are denoised and segmented, followed by data augmentation and feature extraction using a CNN. Fuzzy clustering algorithm is then used to classify the ECG signals for their respective cardio diseases. The proposed approach shows better performance compared to other contemporary algorithms.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Cardiac & Cardiovascular Systems
Matthew O'Connor, Omar Riad, Rui Shi, Dan Hunnybun, Wei Li, Julian W. E. Jarman, John Foran, Christopher A. Rinaldi, Vias Markides, Michael A. Gatzoulis, Tom Wong
Summary: This study aims to investigate the feasibility and safety of left bundle branch area pacing (LBBAP) in congenital heart disease (CHD) patients compared to non-CHD patients. The results showed that LBBAP in CHD patients had similar success rates and lead parameters, as well as comparable procedural and fluoroscopy times, with non-CHD patients.
Article
Computer Science, Interdisciplinary Applications
Zhifan Gao, Yifeng Guo, Jiajing Zhang, Tieyong Zeng, Guang Yang
Summary: The long acquisition time of magnetic resonance imaging (MRI) has been a limitation in terms of patient comfort and motion artifacts. Compressed sensing in MRI (CS-MRI) has enabled fast acquisition without compromising SNR and resolution, but current CS-MRI methods struggle with aliasing artifacts, leading to unsatisfactory reconstruction performance. In order to address this challenge, a hierarchical perception adversarial learning framework (HP-ALF) is proposed. HP-ALF utilizes a hierarchical mechanism to perceive image information and effectively removes aliasing artifacts while recovering fine details.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Michael Yeung, Leonardo Rundo, Yang Nan, Evis Sala, Carola-Bibiane Schonlieb, Guang Yang
Summary: The study introduces a method called DSC++ loss to address the calibration issue of DSC loss in biomedical image segmentation. Experimental results demonstrate that DSC++ loss significantly improves calibration performance across different datasets and tasks. Additionally, the DSC++ loss allows the adjustment of recall-precision bias in model predictions based on specific tasks.
JOURNAL OF DIGITAL IMAGING
(2023)
Article
Cardiac & Cardiovascular Systems
Habib Rehman Khan, Haci Yakup Yakupoglu, Ines Kralj-Hans, Shouvik Haldar, Toufan Bahrami, Jonathan Clague, Anthony De Souza, Wajid Hussain, Julian Jarman, David Gareth Jones, Tushar Salukhe, Vias Markides, Dhiraj Gupta, Rajdeep Khattar, Tom Wong, CASA AF Investigators
Summary: This study examined the post-ablation left atrial (LA) function and its impact on AF recurrence in patients with de-novo long-standing persistent AF. The results showed that patients who maintained sinus rhythm had improved LA function and better left ventricular diastolic function compared to those with AF recurrence after ablation.
CIRCULATION-CARDIOVASCULAR IMAGING
(2023)
Article
Cardiac & Cardiovascular Systems
Timothy R. Betts, Wilson W. Good, Lea Melki, Andreas Metzner, Andrew Grace, Atul Verma, Stephen Murray, Simon James, Tom Wong, Lucas V. A. Boersma, Daniel Steven, Arian Sultan, Sonia Busch, Petr Neuzil, Carlo de Asmundis, Justin Lee, Tamas Szili-Torok
Summary: The RECOVER AF study evaluated the performance of non-contact charge-density mapping to guide the ablation of non-PV targets in persistent AF patients. The study showed that 76% of patients were AF-free at 12 months. For patients who had only received PVI, the AF-free rate was 91%.
Article
Computer Science, Artificial Intelligence
Hao Li, Yang Nan, Javier Del Ser, Guang Yang
Summary: This paper proposes a novel 3D large-kernel (LK) attention module to address the challenges of overlapping organs and difficult tumor segmentation in medical images. The LK attention module combines biologically inspired self-attention and convolution to achieve accurate multi-organ and tumor segmentation.
COGNITIVE COMPUTATION
(2023)
Article
Computer Science, Interdisciplinary Applications
Zheyao Gao, Fuping Wu, Weiguo Gao, Xiahai Zhuang
Summary: Large training datasets are crucial for deep learning-based methods, but obtaining a large number of labeled training images solely from one center can be challenging in medical image segmentation. Distributed learning, such as swarm learning, has the potential to utilize multi-center data while preserving data privacy. However, data distributions across centers may vary significantly due to diverse imaging protocols and vendors, leading to degraded models. In this work, a novel swarm learning approach is proposed to address this issue by leveraging label skew-aware loss and aligning local feature distributions. The approach was validated on four public datasets, and the results demonstrate superior performance compared to existing methods. Code will be released once the paper is accepted.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Clinical Neurology
Yongkai Liu, Yannan Yu, Jiahong Ouyang, Bin Jiang, Guang Yang, Sophie Ostmeier, Max Wintermark, Patrik Michel, David S. Liebeskind, Maarten G. Lansberg, Gregory W. Albers, Greg Zaharchuk
Summary: This study developed a deep learning model fused with clinical variables to predict the 90-day stroke outcome in acute ischemic stroke patients. The fused model showed superior performance compared to clinical and imaging models, with reduced subjectivity and user burden.
Article
Engineering, Civil
Gang Wang, Mingliang Zhou, Xuekai Wei, Guang Yang
Summary: This study proposes an abandoned object detection approach based on vehicular ad-hoc networks (VANETs) and edge artificial intelligence (AI) in road scenes. A detection algorithm is proposed that combines a deep learning network and a deduplication module for improved detection accuracy and reduced repeated detection rates in mobile computing. Furthermore, a location estimation approach based on the World Geodetic System 1984 (WGS84) coordinate system and an affine projection model is proposed for accurate computation of abandoned object positions.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Yang Nan, Javier Del Ser, Zeyu Tang, Peng Tang, Xiaodan Xing, Yingying Fang, Francisco Herrera, Witold Pedrycz, Simon Walsh, Guang Yang
Summary: Airway segmentation plays a crucial role in the examination, diagnosis, and prognosis of lung diseases. However, manual delineation of airways is time-consuming and subjective. To address this issue, researchers have proposed automatic segmentation methods using CT images. Small-sized airway branches greatly affect the accuracy of segmentation, especially due to variability in voxel values and severe data imbalance. This article presents an efficient method for airway segmentation using a novel fuzzy attention neural network and a comprehensive loss function. The proposed method demonstrates efficiency, generalization, and robustness through testing on datasets of different lung diseases.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Fuping Wu, Xiahai Zhuang
Summary: Supervised segmentation is expensive in biomedical image analysis, and semi-supervised segmentation is proposed as an efficient alternative that utilizes both labeled and unlabeled images. This work introduces a new formulation based on risk minimization that considers the risks on both labeled and unlabeled images. Experimental results on medical image segmentation tasks demonstrate the effectiveness and superior performance of the proposed method.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Engineering, Electrical & Electronic
Xin Zhao, Tongming Wang, Jingsong Chen, Bingrun Jiang, Haotian Li, Nan Zhang, Guang Yang, Senchun Chai
Summary: This study proposes a novel contrastive learning strategy that leverages the relative position differences between image slices, combined with global and local features, to address the challenge of generating positive and negative data pairs for medical image segmentation. By employing a two-dimensional fully connected conditional random field for iterative optimization, segmentation accuracy is enhanced and isolated mis-segmented regions are reduced. Experimental results demonstrate that this method outperforms existing semi-supervised and self-supervised techniques in medical segmentation tasks with limited annotated samples.
INTERNATIONAL JOURNAL OF IMAGING SYSTEMS AND TECHNOLOGY
(2023)
Article
Medicine, General & Internal
Yongwon Cho, Soojung Park, Sung Ho Hwang, Minseok Ko, Do-Sun Lim, Cheol Woong Yu, Seong-Mi Park, Mi-Na Kim, Yu-Whan Oh, Guang Yang
Summary: This study proposes a deep learning architecture that can automatically detect the complex structure of the aortic annulus plane for transcatheter aortic valve replacement (TAVR) using cardiac computed tomography (CT). The performance of the proposed model, ADPANet, was evaluated and found to outperform other convolutional neural networks in terms of detecting the aortic annulus plane.
JOURNAL OF KOREAN MEDICAL SCIENCE
(2023)
Article
Computer Science, Artificial Intelligence
Hong Liu, Dong Wei, Donghuan Lu, Xiaoying Tang, Liansheng Wang, Yefeng Zheng
Summary: This study proposes a framework based on hybrid 2D-3D convolutional neural networks for obtaining continuous 3D retinal layer surfaces from OCT volumes. The framework works well with both full and sparse annotations and utilizes alignment displacement vectors and layer segmentation to align the B-scans and segment the layers. Experimental results show that the framework outperforms state-of-the-art 2D deep learning methods in terms of layer segmentation accuracy and cross-B-scan 3D continuity.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Simon Oxenford, Ana Sofia Rios, Barbara Hollunder, Clemens Neudorfer, Alexandre Boutet, Gavin J. B. Elias, Jurgen Germann, Aaron Loh, Wissam Deeb, Bryan Salvato, Leonardo Almeida, Kelly D. Foote, Robert Amaral, Paul B. Rosenberg, David F. Tang-Wai, David A. Wolk, Anna D. Burke, Marwan N. Sabbagh, Stephen Salloway, M. Mallar Chakravarty, Gwenn S. Smith, Constantine G. Lyketsos, Michael S. Okun, William S., Zoltan Mari, Francisco A. Ponce, Andres Lozano, Wolf-Julian Neumann, Bassam Al-Fatly, Andreas Horn
Summary: Spatial normalization is a method to map subject brain images to an average template brain, allowing comparison of brain imaging results. We introduce a novel tool called WarpDrive, which enables manual refinements of image alignment after automated registration. The tool improves accuracy of data representation and aids in understanding patient outcomes.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Ricards Marcinkevics, Patricia Reis Wolfertstetter, Ugne Klimiene, Kieran Chin-Cheong, Alyssia Paschke, Julia Zerres, Markus Denzinger, David Niederberger, Sven Wellmann, Ece Ozkan, Christian Knorr, Julia E. Vogt
Summary: This study presents interpretable machine learning models for predicting the diagnosis, management, and severity of suspected appendicitis using ultrasound images. The proposed models utilize concept bottleneck models (CBM) that facilitate interpretation and intervention by clinicians, without compromising performance or requiring time-consuming image annotation.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Jian-Qing Zheng, Ziyang Wang, Baoru Huang, Ngee Han Lim, Bartlomiej W. Papiez
Summary: This article introduces a new method for medical image registration, which utilizes a separable motion backbone and a residual aligner module to better handle the discontinuous motion of multiple neighboring objects. The proposed method achieves excellent registration results on abdominal CT scans and lung CT scans.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Xiangqiong Wu, Guanghua Tan, Hongxia Luo, Zhilun Chen, Bin Pu, Shengli Li, Kenli Li
Summary: This study develops a user-friendly framework for the automated diagnosis of thyroid nodules in ultrasound videos, simulating the diagnostic workflow of radiologists. By interpreting image characteristics and modeling temporal contextual information, the efficiency and generalizability of the diagnosis can be improved.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Riddhish Bhalodia, Shireen Elhabian, Jadie Adams, Wenzheng Tao, Ladislav Kavan, Ross Whitaker
Summary: This paper introduces DeepSSM, a deep learning-based framework for image-to-shape modeling. By learning the functional mapping from images to low-dimensional shape descriptors, DeepSSM can directly infer statistical representation of anatomy from 3D images. Compared to traditional methods, DeepSSM eliminates the need for heavy manual preprocessing and segmentation, and significantly improves computational time.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Florentin Liebmann, Marco von Atzigen, Dominik Stutz, Julian Wolf, Lukas Zingg, Daniel Suter, Nicola A. Cavalcanti, Laura Leoty, Hooman Esfandiari, Jess G. Snedeker, Martin R. Oswald, Marc Pollefeys, Mazda Farshad, Philipp Furnstahl
Summary: This study presents a marker-less approach for automatic registration and real-time navigation of lumbar spinal fusion surgery using a deep neural network, avoiding radiation exposure and surgical errors. The method was validated on an ex-vivo surgery and a public dataset.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Piyush Tiwary, Kinjawl Bhattacharyya, A. P. Prathosh
Summary: Domain shift refers to the change of distributional characteristics between training and testing datasets, leading to performance drop. For medical image tasks, domain shift can be caused by changes in imaging modalities, devices, and staining mechanisms. Existing approaches based on generative models suffer from training difficulties and lack of diversity. In this paper, the authors propose the use of energy-based models (EBMs) for unpaired image-to-image translation in medical images. The proposed method, called Cycle Consistent Twin EBMs (CCT-EBM), employs a pair of EBMs in the latent space of an Auto-Encoder to ensure translation symmetry and coupling between domains.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Yutong Xie, Jianpeng Zhang, Lingqiao Liu, Hu Wang, Yiwen Ye, Johan Verjans, Yong Xia
Summary: This paper proposes a hybrid pre-training paradigm that combines self-supervised learning and supervised learning to improve the representation quality for medical image segmentation tasks. It introduces a reference task in self-supervised learning and optimizes the model using a gradient matching method. The experimental results demonstrate the effectiveness of this approach on multiple medical image segmentation benchmarks.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Youyi Song, Jing Zou, Kup-Sze Choi, Baiying Lei, Jing Qin
Summary: Cell classification is crucial for intelligent cervical cancer screening, but the variation in cells' appearance and shape poses challenges. A new learning algorithm, worse-case boosting, is proposed to improve classification accuracy for under-represented data. Experimental results demonstrate the effectiveness of this algorithm in two publicly available datasets, achieving a 4% improvement in accuracy.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Sangjoon Park, Eun Sun Lee, Kyung Sook Shin, Jeong Eun Lee, Jong Chul Ye
Summary: The increasing demand for AI systems to monitor human errors and abnormalities in healthcare presents challenges. This study presents a model called Medical X-VL, which is tailored for the medical domain and outperformed current state-of-the-art models in two medical image datasets. The model enables various zero-shot tasks for monitoring AI in the medical domain.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Anna Klimovskaia Susmelj, Berkan Lafci, Firat Ozdemir, Neda Davoudi, Xose Luis Dean-Ben, Fernando Perez-Cruz, Daniel Razansky
Summary: Optoacoustic imaging is a technique that uses optical excitation and ultrasound detection for biological tissue imaging. The quality of the images depends on the extent of tomographic coverage provided by the ultrasound detector arrays. However, full coverage is not always possible due to experimental constraints. The proposed signal domain adaptation network aims to reduce limited-view artifacts in the images.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Srijay Deshpande, Muhammad Dawood, Fayyaz Minhas, Nasir Rajpoot
Summary: In this work, a novel framework called SynCLay is proposed for automated synthesis of histology images based on user-defined cellular layouts. The framework can generate realistic and high-quality histology images with different cellular arrangements, which is helpful for studying the role of cells in the tumor microenvironment. The framework integrates a nuclear segmentation and classification model to refine nuclear structures and generate nuclear masks. Evaluation using quantitative metrics and feedback from pathologists shows that the synthetic images generated by SynCLay have high realism scores and can accurately differentiate between benign and malignant tumors.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Ahmed H. Shahin, An Zhao, Alexander C. Whitehead, Daniel C. Alexander, Joseph Jacob, David Barber
Summary: Survival analysis is a valuable tool in healthcare for predicting the time to specific events. This paper introduces CenTime, a novel approach that directly estimates the time to event. The method performs well with censored data and can be easily integrated with deep learning models. Compared to standard methods, CenTime offers superior performance in predicting event time while maintaining comparable ranking performance.
MEDICAL IMAGE ANALYSIS
(2024)
Article
Computer Science, Artificial Intelligence
Bingyuan Liu, Jose Dolz, Adrian Galdran, Riadh Kobbi, Ismail Ben Ayed
Summary: Most segmentation losses, such as CE and Dice, are variants of the Cross-Entropy or Dice losses. This work provides a theoretical analysis that shows a deeper connection between CE and Dice than previously thought. From a constrained-optimization perspective, both CE and Dice decompose into similar ground-truth matching terms and region-size penalty terms. The analysis uncovers hidden region-size biases: Dice has an intrinsic bias towards extremely imbalanced solutions, while CE implicitly encourages the ground-truth region proportions. Based on this analysis, a principled and simple solution is proposed to explicitly control the region-size bias.
MEDICAL IMAGE ANALYSIS
(2024)