Article
Chemistry, Multidisciplinary
He Zhang, Cunxiao Miao, Linghao Zhang, Yunpeng Zhang, Yufeng Li, Kaiwen Fang
Summary: This paper proposes a real-time simulator for navigation in GNSS-denied environments, which can be used to test and verify navigation algorithms, reducing the time and personnel requirements during flight testing.
APPLIED SCIENCES-BASEL
(2023)
Article
Health Care Sciences & Services
Blanca Larraga-Garcia, Luis Castaneda Lopez, Francisco Javier Rubio Bolivar, Manuel Quintana-Diaz, Alvaro Gutierrez
Summary: Trauma is a leading cause of death in people under 45 years old and specific trauma training during and after medical school is crucial. Web-based learning can create realistic trauma scenarios and doctors perform better in treating trauma patients compared to medical students.
JOURNAL OF MEDICAL SYSTEMS
(2021)
Article
Mathematics
Dolors Serra, Pau Romero, Ignacio Garcia-Fernandez, Miguel Lozano, Alejandro Liberos, Miguel Rodrigo, Alfonso Bueno-Orovio, Antonio Berruezo, Rafael Sebastian
Summary: Personalized cardiac electrophysiology simulations have potential for studying cardiac arrhythmias, but their complexity limits their clinical application. This study presents a fast system based on cellular automata to simulate personalized cardiac electrophysiology, capable of reproducing patient-specific scenarios in a matter of seconds.
Article
Radiology, Nuclear Medicine & Medical Imaging
Ludger Feyen, Peter Minko, Nina Franke, Martin Voelker, Patrick Haage, Philipp Paprottka, Jonathan Nadjiri, Marcus Katoh
Summary: The purpose of this study was to test the feasibility of an online simulator-based comprehensive interventional radiology (IR) training curriculum during the COVID-19 pandemic. The results showed improvements in all areas after the training, indicating that this remote training approach is feasible.
ROFO-FORTSCHRITTE AUF DEM GEBIET DER RONTGENSTRAHLEN UND DER BILDGEBENDEN VERFAHREN
(2023)
Article
Engineering, Aerospace
William W. Jun, Kar-Ming Cheung, Edgar Glenn Lightsey, Charles Lee
Summary: This article introduces a filtered approach of joint Doppler and ranging (JDR) for accurate 3D positioning of lunar surface vehicles. Comparisons between JDR and Doppler-based positioning methods show that JDR obtains the lowest and most consistent overall positioning error.
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS
(2022)
Article
Agriculture, Multidisciplinary
Xiaoning Zhao, Yuefeng Du, Enrong Mao, Zhongxiang Zhu, Zhenghe Song
Summary: This study verified the feasibility and high accuracy of agricultural tractor dynamics simulation based on the driving simulator, laying a foundation for further research and application in real-time human-machine interaction.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Engineering, Biomedical
Reinhard Fuchs, Karel M. Van Praet, Richard Bieck, Joerg Kempfert, David Holzhey, Markus Kofler, Michael A. Borger, Stephan Jacobs, Volkmar Falk, Thomas Neumuth
Summary: This paper presents a multimodal training evaluation system that combines motion and muscle-action measurements to assess endoscopic training and track learning progress. The system demonstrates the effectiveness of its setup and feature calculation in distinguishing different endoscopic view modes.
INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY
(2022)
Article
Engineering, Marine
Rui Tao, Hongxiang Ren, Yi Zhou
Summary: In response to the lack of comprehensive understanding of ship firefighting equipment operation among marine students, we developed a ship firefighting training simulator. The simulator features multi-sensory human-computer interaction and utilizes three-dimensional modeling technology for ship models. We also proposed a new smoke simulation method to enhance the realism of training scenarios.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2022)
Article
Entomology
Douglas D. Gaffin, Safra F. Shakir
Summary: Research shows that scorpions have complex chemo-tactile organs on their pectines, allowing them to detect pheromone trails, with neurons within these organs interacting with each other. This simple feedback circuit may help maintain neurons in a sensitive range, aiding in accurate detection and processing of substrate information.
Article
Biotechnology & Applied Microbiology
Kasparas Zilinskas, Jennie H. Kwon, Katherine Bishara, Kaila Hayden, Ritchelli Quintao, Taufiek Konrad Rajab
Summary: Surgical simulation is crucial in training cardiac surgeons, and a new dynamic ventricular simulator has been developed to test the efficacy of aortic procedures under dynamic physiologic conditions. It can evaluate procedural success using echocardiography and hemodynamic measurements. Besides valve surgeries, the simulator has potential applications in other cardiac procedures.
BIOENGINEERING-BASEL
(2022)
Article
Surgery
Anishan Vamadevan, Lars Konge, Morten Stadeager, Flemming Bjerrum
Summary: The study investigated the effect of adding haptic simulators to a proficiency-based laparoscopy training program. The results showed that haptic simulators reduced the time to reach proficiency compared to non-haptic simulators. However, the acquired skills were not transferable to the conventional non-haptic setting.
SURGICAL ENDOSCOPY AND OTHER INTERVENTIONAL TECHNIQUES
(2023)
Article
Engineering, Marine
Tanaka Akiyama, Kostia Roncin, Jean-Francois Bousquet
Summary: In this study, a hardware-in-the-loop (HIL) simulator is developed to assess the behavior of an autonomous sailboat in navigation. The core of the HIL simulator is an embedded sailboat pilot. The sensor data input to the embedded system is provided by a navigation simulator, which takes into account various forces acting on the sailboat. The HIL simulator is tested using sea trials data from 2014, and the performance of the automated pilot is compared to that of a crew-operated vessel. The results demonstrate that the automated system can outperform human-operated vessels. Furthermore, the tool is used to identify weaknesses in the sailboat autopilot algorithm for future improvements.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Energy & Fuels
Hao Fu, Peng Li, Xiaopeng Fu, Jinyue Yan, Zhiying Wang, Kun Wang, Jianzhong Wu, Chengshan Wang
Summary: This paper presents a compact real-time simulator for large-scale wind farms based on field programmable gate array (FPGA). A spatial-temporal parallel design method is proposed to address the demand for huge computing resources associated with detailed modeling. The wind farm is decoupled into subsystems and the electrical system and control system of each subsystem are solved in parallel. The simulation incorporates module-level and superscalar pipeline techniques to improve hardware resource utilization. Case studies demonstrate the accuracy and effectiveness of the proposed design.
CSEE JOURNAL OF POWER AND ENERGY SYSTEMS
(2023)
Article
Computer Science, Information Systems
Washington Velasquez
Summary: This paper introduces an innovative proposal for utilizing a simulation/emulation tool for obtaining real environmental data using wireless sensor networks. It optimizes sensor node locations through a mathematical model and utilizes a distributed computational architecture, micro-services management, and data flow control components. The simulator makes a significant contribution in the WSN field by enabling real-time data acquisition and application in intelligent systems.
Article
Engineering, Electrical & Electronic
Binhan Du, Baojian Yang, Huaiguang Wang, Guoquan Ren, Zhiyong Shi
Summary: This article proposed a three-layer structure for the VDM-AINS, which corrects the vehicle dynamics model (VDM) using the inertial navigation system (INS) to improve the accuracy and robustness of the whole system. Simulation experiments and field tests demonstrated that the proposed method can significantly reduce velocity errors of VDM in low-speed and high-dynamic situations, and maintain a maximum position error of 241.8 m.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Dimitri Hamzaoui, Sarah Montagne, Benjamin Granger, Alexandre Allera, Malek Ezziane, Anna Luzurier, Raphaelle Quint, Mehdi Kalai, Nicholas Ayache, Herve Delingette, Raphaele Renard Penna
Summary: In this study, the intra- and inter-rater variability of prostate volume estimation using manual planimetry and ellipsoid formulas were compared. The results showed that both methods provided highly reproducible measurements, but manual planimetry had the best inter-rater reproducibility, while the ellipsoid formula showed slight overestimation of prostate volume compared to manual planimetry.
EUROPEAN RADIOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Zihao Wang, Thomas Demarcy, Clair Vandersteen, Dan Gnansia, Charles Raffaelli, Nicolas Guevara, Herve Delingette
Summary: This paper presents a Bayesian inference approach for parametric shape models to segment medical images, aiming to provide interpretable results. The framework defines likelihood appearance probability and prior label probability based on a generic shape function, controlling the trade-off between shape and appearance information.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Computer Science, Artificial Intelligence
Benoit Audelan, Dimitri Hamzaoui, Sarah Montagne, Raphaele Renard-Penna, Herve Delingette
Summary: This paper introduces a novel approach to jointly estimate a reliable consensus map and assess the presence of outliers and confidence in each rater. The robust approach is based on heavy-tailed distributions, allowing for local estimates of rater performance and the introduction of bias and spatial priors.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Oncology
Marius Ilie, Jonathan Benzaquen, Paul Tourniaire, Simon Heeke, Nicholas Ayache, Herve Delingette, Elodie Long-Mira, Sandra Lassalle, Marame Hamila, Julien Fayada, Josiane Otto, Charlotte Cohen, Abel Gomez-Caro, Jean-Philippe Berthet, Charles-Hugo Marquette, Veronique Hofman, Christophe Bontoux, Paul Hofman
Summary: The study demonstrates that deep learning combined with convolutional neural networks has the potential to assist in the diagnosis of pulmonary neuroendocrine tumors. By using a trained deep learning classifier, the method can accurately differentiate between different histologic subtypes of lung neuroendocrine carcinoma.
Article
Radiology, Nuclear Medicine & Medical Imaging
Dimitri Hamzaoui, Sarah Montagne, Raphaele Renard-Penna, Nicholas Ayache, Herve Delingette
Summary: This study proposes a deep learning-based method for automatic zonal segmentation of the prostate, achieving accurate localization of lesions. The method takes into account the image anisotropy and uses loss functions to enforce prostate partition. Experimental results on private and public datasets demonstrate that the method outperforms other network methods and remains consistent with the results obtained by radiologists.
JOURNAL OF MEDICAL IMAGING
(2022)
Article
Computer Science, Software Engineering
Antonin Bernardin, Eulalie Coevoet, Paul Kry, Sheldon Andrews, Christian Duriez, Maud Marchal
Summary: In this paper, a physics-based model of suction phenomenon is proposed for simulating deformable objects like suction cups. The model utilizes a constraint-based formulation to simulate pressure variations within suction cups, represented as pressure constraints coupled with anti-interpenetration and friction constraints. The method is capable of detecting multiple air cavities using collision detection information. The pressure constraints are solved based on the ideal gas law, considering various cavity states. The model is tested in different scenarios, demonstrating its potential applications in soft robotics and computer animation.
ACM TRANSACTIONS ON GRAPHICS
(2023)
Article
Robotics
Felix Vanneste, Olivier Goury, Christian Duriez
Summary: This letter proposes an automatic finite element model calibration method based on real data, which optimizes the mechanical properties of a given structure to achieve a specific configuration goal. The method is evaluated using anisotropic materials and shows interest in calibration for addressing errors introduced by manufacturing, imperfect models, or mechanical fatigue/plasticity.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Cardiac & Cardiovascular Systems
Sofia Monaci, Shuang Qian, Karli Gillette, Esther Puyol-Anton, Rahul Mukherjee, Mark K. Elliott, John Whitaker, Ronak Rajani, Mark O'Neill, Christopher A. Rinaldi, Gernot Plank, Andrew P. King, Martin J. Bishop
Summary: This study presents a non-invasive computational-deep learning platform for localizing post-infarct ventricular tachycardia. The platform utilizes surface electrocardiograms (ECGs) and intracardiac electrograms (EGMs) recorded from implanted devices to accurately locate the exit sites of the tachycardia. The results demonstrate the potential of this platform in clinical settings for enhancing the safety and speed of ablation planning.
Editorial Material
Cardiac & Cardiovascular Systems
Aurelien Bustin, Matthias Stuber, Maxime Sermesant, Hubert Cochet
EUROPEAN HEART JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Paul Tourniaire, Marius Ilie, Paul Hofman, Nicholas Ayache, Herve Delingette
Summary: Using mixed supervision, we improve the classification and localization performances of a weakly-supervised model based on attention-based deep Multiple Instance Learning. With a limited amount of patch-level labeled slides, we achieve performance close to fully-supervised models.
MEDICAL IMAGE ANALYSIS
(2023)
Article
Medicine, General & Internal
Raabid Hussain, Attila Frater, Roger Calixto, Chadlia Karoui, Jan Margeta, Zihao Wang, Michel Hoen, Herve Delingette, Francois Patou, Charles Raffaelli, Clair Vandersteen, Nicolas Guevara
Summary: Understanding cochlear anatomy is essential for developing less invasive cochlear implantation techniques. This study analyzed over 1000 clinical temporal bone CT images to determine population statistics and correlations between cochlear dimensions and morphology. The findings indicate that cochlear morphology follows a normal distribution and that certain dimensions are more correlated with duct lengths, wrapping factor, and volume. The results also highlight the variability in scala tympani size and suggest differences in size and shape between ears of the same individual. Overall, this research provides important insights for implant development and reducing trauma during insertion.
JOURNAL OF CLINICAL MEDICINE
(2023)
Article
Automation & Control Systems
Lingxiao Xun, Gang Zheng, Sofiane Ghenna, Alexandre Kruszewski, Eric Cattan, Christian Duriez, Sebastien Grondel
Summary: In this study, an electromechanic model and an optimal controller were proposed for an IEPA actuator, which can produce large deformation under low actuation voltage. The models successfully predicted the deformation of the actuator under different input voltages, and the optimal controller achieved high control performance. This research lays the foundation for subsequent biomedical applications.
IEEE-ASME TRANSACTIONS ON MECHATRONICS
(2023)
Article
Engineering, Electrical & Electronic
Mohammad Alshawabkeh, Hosam Alagi, Stefan Escaida Navarro, Christian Duriez, Bjorn Hein, Hubert Zangl, Lisa-Marie Faller
Summary: Soft robotics is a promising approach for collaborative robots. Soft robotic systems are safe and flexible due to their compliant nature. In this work, we develop stretchable capacitive sensors for proximity and tactile detection, and evaluate their performance under different conditions.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Artificial Intelligence
Hind Dadoun, Anne-Laure Rousseau, Eric de Kerviler, Jean-Michel Correas, Anne-Marie Tissier, Fanny Joujou, Sylvain Bodard, Kemel Khezzane, Constance De Margerie-Mellon, Herve Delingette, Nicholas Ayache
Summary: The purpose of this study was to train and evaluate a deep learning-based network for detecting, localizing, and characterizing liver lesions in abdominal ultrasound images. The results showed that the network performed better than caregivers in detecting and localizing the lesions, and achieved similar performance in lesion characterization compared to experts.
RADIOLOGY-ARTIFICIAL INTELLIGENCE
(2022)
Correction
Radiology, Nuclear Medicine & Medical Imaging
Dimitri Hamzaoui, Sarah Montagne, Benjamin Granger, Alexandre Allera, Malek Ezziane, Anna Luzurier, Raphaelle Quint, Mehdi Kalai, Nicholas Ayache, Herve Delingette, Raphaele Renard-Penna
EUROPEAN RADIOLOGY
(2022)
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)