Thomas Thurner, Julia Maier, Martin Kaltenbrunner, Andreas Schrempf
Surgical simulators are valuable educational tools for physicians, enhancing their proficiency and improving patient safety. However, they typically still suffer from a lack of realism as they do not emulate dynamic tissue biomechanics haptically and fail to convincingly mimic real‐time physiological reactions. This study presents a dynamic tactile synthetic tissue, integrating both sensory and actuatory capabilities within a fully soft unit, as a core component for soft robotics and future hybrid surgical simulators utilizing dynamic physical phantoms. The adaptive surface of the tissue replica, actuated via hydraulics, is assessed by an embedded carbon black silicone sensor layer using electrical impedance tomography to determine internally or externally induced deformations. The integrated fluid chambers enable pressure and force measurements. The combination of these principles enables real‐time tissue feedback as well as closed loop operation, allowing optimal interaction with the environment. Based on the concepts of soft robotics, such artificial tissues find broad applicability, demonstrated via a soft gripper and surgical simulation applications including a dynamic, artificial brain phantom as well as a synthetic, beating heart. These advancements pave the way toward enhanced realism in surgical simulators including reliable performance evaluation and bear the potential to transform the future of surgical training methodologies.
{"title":"Dynamic Tactile Synthetic Tissue: from Soft Robotics to Hybrid Surgical Simulators","authors":"Thomas Thurner, Julia Maier, Martin Kaltenbrunner, Andreas Schrempf","doi":"10.1002/aisy.202400199","DOIUrl":"https://doi.org/10.1002/aisy.202400199","url":null,"abstract":"Surgical simulators are valuable educational tools for physicians, enhancing their proficiency and improving patient safety. However, they typically still suffer from a lack of realism as they do not emulate dynamic tissue biomechanics haptically and fail to convincingly mimic real‐time physiological reactions. This study presents a dynamic tactile synthetic tissue, integrating both sensory and actuatory capabilities within a fully soft unit, as a core component for soft robotics and future hybrid surgical simulators utilizing dynamic physical phantoms. The adaptive surface of the tissue replica, actuated via hydraulics, is assessed by an embedded carbon black silicone sensor layer using electrical impedance tomography to determine internally or externally induced deformations. The integrated fluid chambers enable pressure and force measurements. The combination of these principles enables real‐time tissue feedback as well as closed loop operation, allowing optimal interaction with the environment. Based on the concepts of soft robotics, such artificial tissues find broad applicability, demonstrated via a soft gripper and surgical simulation applications including a dynamic, artificial brain phantom as well as a synthetic, beating heart. These advancements pave the way toward enhanced realism in surgical simulators including reliable performance evaluation and bear the potential to transform the future of surgical training methodologies.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"40 10","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141929423","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
W. Kho, Hyun-Deog Hwang, Taewan Noh, Hoseong Kim, Ji Min Lee, Seung‐Eon Ahn
Memristors play a pivotal role in advanced computing, with memristor‐based crossbar arrays showing promise for various artificial neural networks. Among these, HfO2‐based ferroelectric tunnel junctions (FTJs) stand out as ideal synaptic devices for neuromorphic computing. Their compatibility with the complementary metal oxide semiconductor process and intrinsic energy efficiency make them particularly appealing. While an increasing number of studies adopt identical pulse programming (IPP) with short width to update the conductance of HfO2‐based FTJs synaptic devices, conventional ferroelectric switching models fall short in describing updates the conductance with the IPP scheme. Consequently, studies achieving conductance updates via IPP lack an underlying mechanism explanation, potentially limiting the application of HfO2‐based FTJs as synaptic devices. This study explores the potential of ferroelectric Zr‐doped HfO2 (HZO) FTJs to undergo learning through the IPP scheme. Synaptic characteristics, including the number of conductance states, symmetry, linearity, write energy, and latency by modulating IPP scheme conditions are optimized. Finally, the applicability of HZO FTJ as a synaptic device by assessing learning accuracy in pattern recognition through artificial neural network simulation based on the optimized synaptic characteristics is evaluated.
{"title":"Maximizing the Synaptic Efficiency of Ferroelectric Tunnel Junction Devices Using a Switching Mechanism Hidden in an Identical Pulse Programming Learning Scheme","authors":"W. Kho, Hyun-Deog Hwang, Taewan Noh, Hoseong Kim, Ji Min Lee, Seung‐Eon Ahn","doi":"10.1002/aisy.202400211","DOIUrl":"https://doi.org/10.1002/aisy.202400211","url":null,"abstract":"Memristors play a pivotal role in advanced computing, with memristor‐based crossbar arrays showing promise for various artificial neural networks. Among these, HfO2‐based ferroelectric tunnel junctions (FTJs) stand out as ideal synaptic devices for neuromorphic computing. Their compatibility with the complementary metal oxide semiconductor process and intrinsic energy efficiency make them particularly appealing. While an increasing number of studies adopt identical pulse programming (IPP) with short width to update the conductance of HfO2‐based FTJs synaptic devices, conventional ferroelectric switching models fall short in describing updates the conductance with the IPP scheme. Consequently, studies achieving conductance updates via IPP lack an underlying mechanism explanation, potentially limiting the application of HfO2‐based FTJs as synaptic devices. This study explores the potential of ferroelectric Zr‐doped HfO2 (HZO) FTJs to undergo learning through the IPP scheme. Synaptic characteristics, including the number of conductance states, symmetry, linearity, write energy, and latency by modulating IPP scheme conditions are optimized. Finally, the applicability of HZO FTJ as a synaptic device by assessing learning accuracy in pattern recognition through artificial neural network simulation based on the optimized synaptic characteristics is evaluated.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":" 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141828208","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Youngheon Yun, Dongchan Lee, Soyeon Lee, Salvador Pané, Josep Puigmartí‐Luis, Sungwoo Chun, Bumjin Jang
The research addresses the limitations inherent in conventional Hall effect‐based tactile sensors, particularly their restricted sensitivity by introducing an innovative metastructure. Through meticulous finite element analysis optimization, the Hall effect‐based auxetic tactile sensor (HEATS), featuring a rotating square plate configuration as the most effective auxetic pattern to enhance sensitivity, is developed. Experimental validation demonstrates significant sensitivity enhancements across a wide sensing range. HEATS exhibits a remarkable 20‐fold and 10‐fold improvement at tensile rates of 0.9% and 30%, respectively, compared to non‐auxetic sensors. Furthermore, comprehensive testing demonstrates HEATS’ exceptional precision in detecting various tactile stimuli, including muscle movements and joint angles. With its unparalleled accuracy and adaptability, HEATS offers vast potential applications in human–machine and human–robot interaction, where subtle tactile communication is a prerequisite.
{"title":"Enhancing Sensitivity across Scales with Highly Sensitive Hall Effect‐Based Auxetic Tactile Sensors","authors":"Youngheon Yun, Dongchan Lee, Soyeon Lee, Salvador Pané, Josep Puigmartí‐Luis, Sungwoo Chun, Bumjin Jang","doi":"10.1002/aisy.202400337","DOIUrl":"https://doi.org/10.1002/aisy.202400337","url":null,"abstract":"The research addresses the limitations inherent in conventional Hall effect‐based tactile sensors, particularly their restricted sensitivity by introducing an innovative metastructure. Through meticulous finite element analysis optimization, the Hall effect‐based auxetic tactile sensor (HEATS), featuring a rotating square plate configuration as the most effective auxetic pattern to enhance sensitivity, is developed. Experimental validation demonstrates significant sensitivity enhancements across a wide sensing range. HEATS exhibits a remarkable 20‐fold and 10‐fold improvement at tensile rates of 0.9% and 30%, respectively, compared to non‐auxetic sensors. Furthermore, comprehensive testing demonstrates HEATS’ exceptional precision in detecting various tactile stimuli, including muscle movements and joint angles. With its unparalleled accuracy and adaptability, HEATS offers vast potential applications in human–machine and human–robot interaction, where subtle tactile communication is a prerequisite.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"36 30","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141645384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Inspired by the efficient swimming capabilities of swordfish, a novel wireless soft swordfish‐like robot with programmable magnetization has been developed, integrating direct‐ink‐writing (DIW) 3D printing and assembly technology. This 20 mm long robot features a streamlined form and magnetically programmable movements, enabling biomimetic locomotion patterns such as straight‐line swimming and turning swimming. The robot includes a silicone‐based torso (body, abdomen, and pectoral fin) and a crescent‐shaped tail fin made from a magnetically programmable polymer embedded with neodymium‐iron‐boron (NdFeB) particles. The tail fin, fabricated by multi‐material alternating printing to achieve a gradient magnetism distribution, is controlled by an external magnetic field to mimic the rapid oscillation of a swordfish's tail, achieving a swimming speed of 0.51 BL/ s. The tail fin's asymmetric oscillation amplitudes, adjusted by magnetic field control, allow the robot to transition seamlessly from high‐speed straight swimming to agile turning. The robot can perform tracking swimming along specific planned paths, such as “C” and “Z” shaped trajectories. Potential applications include environmental monitoring and targeted drug release. The multi‐material 3D printing technology enhances the robot's efficiency and sensitivity in simulating natural biological movements, extending to the design and development of various flexible devices and soft robots.
{"title":"3D Printed Swordfish‐Like Wireless Millirobot","authors":"Xingcheng Ou, Yu Sheng, Jiaqi Huang, Dantong Huang, Xiaohong Li, Ran Bi, Guoliang Chen, Weijie Hu, Shuang‐Zhuang Guo","doi":"10.1002/aisy.202400206","DOIUrl":"https://doi.org/10.1002/aisy.202400206","url":null,"abstract":"Inspired by the efficient swimming capabilities of swordfish, a novel wireless soft swordfish‐like robot with programmable magnetization has been developed, integrating direct‐ink‐writing (DIW) 3D printing and assembly technology. This 20 mm long robot features a streamlined form and magnetically programmable movements, enabling biomimetic locomotion patterns such as straight‐line swimming and turning swimming. The robot includes a silicone‐based torso (body, abdomen, and pectoral fin) and a crescent‐shaped tail fin made from a magnetically programmable polymer embedded with neodymium‐iron‐boron (NdFeB) particles. The tail fin, fabricated by multi‐material alternating printing to achieve a gradient magnetism distribution, is controlled by an external magnetic field to mimic the rapid oscillation of a swordfish's tail, achieving a swimming speed of 0.51 BL/ s. The tail fin's asymmetric oscillation amplitudes, adjusted by magnetic field control, allow the robot to transition seamlessly from high‐speed straight swimming to agile turning. The robot can perform tracking swimming along specific planned paths, such as “C” and “Z” shaped trajectories. Potential applications include environmental monitoring and targeted drug release. The multi‐material 3D printing technology enhances the robot's efficiency and sensitivity in simulating natural biological movements, extending to the design and development of various flexible devices and soft robots.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"104 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141667056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Meftahul Ferdaus, Mahdi Abdelguerfi, Kendall N. Niles, Ken Pathak, Joe Tom
Onboard image analysis enables real‐time autonomous capabilities for unmanned platforms including aerial, ground, and aquatic drones. Performing classification on embedded systems, rather than transmitting data, allows rapid perception and decision‐making critical for time‐sensitive applications such as search and rescue, hazardous environment exploration, and military operations. To fully capitalize on these systems’ potential, specialized deep learning solutions are needed that balance accuracy and computational efficiency for time‐sensitive inference. This article introduces the widened attention‐enhanced atrous convolution‐based efficient network (WACEfNet), a new convolutional neural network designed specifically for real‐time visual classification challenges using resource‐constrained embedded devices. WACEfNet builds on EfficientNet and integrates innovative width‐wise feature processing, atrous convolutions, and attention modules to improve representational power without excessive overhead. Extensive benchmarking confirms state‐of‐the‐art performance from WACEfNet for aerial imaging applications while remaining suitable for embedded deployment. The improvements in accuracy and speed demonstrate the potential of customized deep learning advancements to unlock new capabilities for unmanned aerial vehicles and related embedded systems with tight size, weight, and power constraints. This research offers an optimized framework, combining widened residual learning and attention mechanisms, to meet the unique demands of high‐fidelity real‐time analytics across a variety of embedded perception paradigms.
{"title":"Widened Attention‐Enhanced Atrous Convolutional Network for Efficient Embedded Vision Applications under Resource Constraints","authors":"Meftahul Ferdaus, Mahdi Abdelguerfi, Kendall N. Niles, Ken Pathak, Joe Tom","doi":"10.1002/aisy.202300480","DOIUrl":"https://doi.org/10.1002/aisy.202300480","url":null,"abstract":"Onboard image analysis enables real‐time autonomous capabilities for unmanned platforms including aerial, ground, and aquatic drones. Performing classification on embedded systems, rather than transmitting data, allows rapid perception and decision‐making critical for time‐sensitive applications such as search and rescue, hazardous environment exploration, and military operations. To fully capitalize on these systems’ potential, specialized deep learning solutions are needed that balance accuracy and computational efficiency for time‐sensitive inference. This article introduces the widened attention‐enhanced atrous convolution‐based efficient network (WACEfNet), a new convolutional neural network designed specifically for real‐time visual classification challenges using resource‐constrained embedded devices. WACEfNet builds on EfficientNet and integrates innovative width‐wise feature processing, atrous convolutions, and attention modules to improve representational power without excessive overhead. Extensive benchmarking confirms state‐of‐the‐art performance from WACEfNet for aerial imaging applications while remaining suitable for embedded deployment. The improvements in accuracy and speed demonstrate the potential of customized deep learning advancements to unlock new capabilities for unmanned aerial vehicles and related embedded systems with tight size, weight, and power constraints. This research offers an optimized framework, combining widened residual learning and attention mechanisms, to meet the unique demands of high‐fidelity real‐time analytics across a variety of embedded perception paradigms.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"8 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141683437","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yang Liu, Ying Feng, Linlin Liu, Miao An, Huaming Yang
Micro/nanorobots (MNRs) are promising for biomedical applications due to their unconstrained nature and small enough size to pass through many tiny environments. However, the efficient movement of MNRs in liquid environments is still a challenge due to the low Reynolds number environment and the Brownian motion of particles. Herein, emerging MNRs with hydrogel‐loaded magnetic particles are designed. The proposed hydrogel MNRs (HMNRs) exhibit biocompatible and controllable characteristics. The motion controllability of HMNRs is realized by applying oscillating magnetic field and customized magnetic field. Experimentally, it is demonstrated that the HMNR swarms driven by the oscillating magnetic field exhibit a faster motion speed than the MNR swarms composed of magnetic particles. The HMNRs show precise controllability of the movement in the complex pipeline under the control of customized magnetic field. This method can offer a more benign approach to the general production of HMNRs for biological applications.
{"title":"Design and Motion Controllability of Emerging Hydrogel Micro/Nanorobots","authors":"Yang Liu, Ying Feng, Linlin Liu, Miao An, Huaming Yang","doi":"10.1002/aisy.202400339","DOIUrl":"https://doi.org/10.1002/aisy.202400339","url":null,"abstract":"Micro/nanorobots (MNRs) are promising for biomedical applications due to their unconstrained nature and small enough size to pass through many tiny environments. However, the efficient movement of MNRs in liquid environments is still a challenge due to the low Reynolds number environment and the Brownian motion of particles. Herein, emerging MNRs with hydrogel‐loaded magnetic particles are designed. The proposed hydrogel MNRs (HMNRs) exhibit biocompatible and controllable characteristics. The motion controllability of HMNRs is realized by applying oscillating magnetic field and customized magnetic field. Experimentally, it is demonstrated that the HMNR swarms driven by the oscillating magnetic field exhibit a faster motion speed than the MNR swarms composed of magnetic particles. The HMNRs show precise controllability of the movement in the complex pipeline under the control of customized magnetic field. This method can offer a more benign approach to the general production of HMNRs for biological applications.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"82 19","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141352997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Longitudinal analysis of the gut microbiota is crucial for understanding its relationship with gastrointestinal (GI) diseases and advancing diagnostics and treatments. Most ingestible sampling devices move passively within the GI tract, rely on physiological factors, and fail at multipoint sampling. This study proposes a multiple‐sampling capsule robot capable of collecting gut microbiota from various locations within the GI tract with minimal cross‐contamination. The proposed capsule comprises a body, a driving unit, six sampling tools, a central rod, and two heads. Electromagnetic field control facilitates control of the orientation and position of the capsule, particularly to align the channel of the capsule where the sample is collected facing downward. The capsule can collect six gut microbiota samples preventing contamination before and after sampling. The active locomotion and multiple sampling performance of the capsule are evaluated through basic performance tests (propulsion direction precision: 0.76 ± 0.52°, channel alignment precision: 0.84 ± 0.55°), phantom tests (average amount per sample: 10.3 ± 2.4 mg, cross‐contamination: 0.6 ± 0.4%), and ex‐vivo tests (average amount per sample: 9.9 ± 1.7 mg). The possibility of integration and clinical application of the capsule is confirmed through preclinical tests using a porcine model.
{"title":"Multiple Sampling Capsule Robot for Studying Gut Microbiome","authors":"Sanghyeon Park, M. Hoang, Jayoung Kim, Sukho Park","doi":"10.1002/aisy.202300625","DOIUrl":"https://doi.org/10.1002/aisy.202300625","url":null,"abstract":"Longitudinal analysis of the gut microbiota is crucial for understanding its relationship with gastrointestinal (GI) diseases and advancing diagnostics and treatments. Most ingestible sampling devices move passively within the GI tract, rely on physiological factors, and fail at multipoint sampling. This study proposes a multiple‐sampling capsule robot capable of collecting gut microbiota from various locations within the GI tract with minimal cross‐contamination. The proposed capsule comprises a body, a driving unit, six sampling tools, a central rod, and two heads. Electromagnetic field control facilitates control of the orientation and position of the capsule, particularly to align the channel of the capsule where the sample is collected facing downward. The capsule can collect six gut microbiota samples preventing contamination before and after sampling. The active locomotion and multiple sampling performance of the capsule are evaluated through basic performance tests (propulsion direction precision: 0.76 ± 0.52°, channel alignment precision: 0.84 ± 0.55°), phantom tests (average amount per sample: 10.3 ± 2.4 mg, cross‐contamination: 0.6 ± 0.4%), and ex‐vivo tests (average amount per sample: 9.9 ± 1.7 mg). The possibility of integration and clinical application of the capsule is confirmed through preclinical tests using a porcine model.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"32 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141273145","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Adolescent psychiatric disorders arise from intricate interactions of clinical histories and disruptions in brain development. While connections between psychopathology and brain functional connectivity are studied, the use of deep learning to elucidate overlapping neural mechanisms through multimodal brain images remains nascent. Utilizing two adolescent datasets—the Philadelphia Neurodevelopmental Cohort (PNC, n = 1100) and the Adolescent Brain Cognitive Development (ABCD, n = 7536)—this study employs interpretable neural networks and demonstrates that incorporating brain morphology, along with functional and structural networks, augments traditional clinical characteristics (age, gender, race, parental education, medical history, and trauma exposure). Predictive accuracy reaches 0.37–0.464 between real and predicted general psychopathology and four psychopathology dimensions (externalizing, psychosis, anxiety, and fear). The brain morphology and connectivities within the frontoparietal, default mode network, and visual associate networks are recurrent across general psychopathology and four psychopathology dimensions. Unique structural and functional pathways originating from the cerebellum, amygdala, and visual‐sensorimotor cortex are linked with these individual dimensions. Consistent findings across both PNC and ABCD affirm the generalizability. The results underscore the potential of diverse sensory inputs in steering executive processes tied to psychopathology dimensions in adolescents, hinting at neural avenues for targeted therapeutic interventions and preventive strategies.
{"title":"Unraveling Multimodal Brain Signatures: Deciphering Transdiagnostic Dimensions of Psychopathology in Adolescents","authors":"Jing Xia, Nanguang Chen, Anqi Qiu","doi":"10.1002/aisy.202300577","DOIUrl":"https://doi.org/10.1002/aisy.202300577","url":null,"abstract":"Adolescent psychiatric disorders arise from intricate interactions of clinical histories and disruptions in brain development. While connections between psychopathology and brain functional connectivity are studied, the use of deep learning to elucidate overlapping neural mechanisms through multimodal brain images remains nascent. Utilizing two adolescent datasets—the Philadelphia Neurodevelopmental Cohort (PNC, n = 1100) and the Adolescent Brain Cognitive Development (ABCD, n = 7536)—this study employs interpretable neural networks and demonstrates that incorporating brain morphology, along with functional and structural networks, augments traditional clinical characteristics (age, gender, race, parental education, medical history, and trauma exposure). Predictive accuracy reaches 0.37–0.464 between real and predicted general psychopathology and four psychopathology dimensions (externalizing, psychosis, anxiety, and fear). The brain morphology and connectivities within the frontoparietal, default mode network, and visual associate networks are recurrent across general psychopathology and four psychopathology dimensions. Unique structural and functional pathways originating from the cerebellum, amygdala, and visual‐sensorimotor cortex are linked with these individual dimensions. Consistent findings across both PNC and ABCD affirm the generalizability. The results underscore the potential of diverse sensory inputs in steering executive processes tied to psychopathology dimensions in adolescents, hinting at neural avenues for targeted therapeutic interventions and preventive strategies.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"47 2","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141103467","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To solve the problems of existing hybrid networks based on convolutional neural networks (CNN) and Transformers, we propose a new encoder–decoder network FI‐Net based on CNN‐Transformer for medical image segmentation. In the encoder part, a dual‐stream encoder is used to capture local details and long‐range dependencies. Moreover, the attentional feature fusion module is used to perform interactive feature fusion of dual‐branch features, maximizing the retention of local details and global semantic information in medical images. At the same time, the multi‐scale feature aggregation module is used to aggregate local information and capture multi‐scale context to mine more semantic details. The multi‐level feature bridging module is used in skip connections to bridge multi‐level features and mask information to assist multi‐scale feature interaction. Experimental results on seven public medical image datasets fully demonstrate the effectiveness and advancement of our method. In future work, we plan to extend FI‐Net to support 3D medical image segmentation tasks and combine self‐supervised learning and knowledge distillation to alleviate the overfitting problem of limited data training.
{"title":"FI‐Net: Rethinking Feature Interactions for Medical Image Segmentation","authors":"Yuhan Ding, Jinhui Liu, Yunbo He, Jinliang Huang, Haisu Liang, Zhenglin Yi, Yongjie Wang","doi":"10.1002/aisy.202400201","DOIUrl":"https://doi.org/10.1002/aisy.202400201","url":null,"abstract":"To solve the problems of existing hybrid networks based on convolutional neural networks (CNN) and Transformers, we propose a new encoder–decoder network FI‐Net based on CNN‐Transformer for medical image segmentation. In the encoder part, a dual‐stream encoder is used to capture local details and long‐range dependencies. Moreover, the attentional feature fusion module is used to perform interactive feature fusion of dual‐branch features, maximizing the retention of local details and global semantic information in medical images. At the same time, the multi‐scale feature aggregation module is used to aggregate local information and capture multi‐scale context to mine more semantic details. The multi‐level feature bridging module is used in skip connections to bridge multi‐level features and mask information to assist multi‐scale feature interaction. Experimental results on seven public medical image datasets fully demonstrate the effectiveness and advancement of our method. In future work, we plan to extend FI‐Net to support 3D medical image segmentation tasks and combine self‐supervised learning and knowledge distillation to alleviate the overfitting problem of limited data training.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140970204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Determining the target‐bound conformation of a drug‐like molecule is a crucial step in drug design, as it affects the outcome of virtual screening (VS), and paves the way for hit‐to‐lead and lead optimization. While most docking programs usually manage to produce at least a near‐native pose for a bioactive molecule inside its binding pocket, their integrated classical scoring functions (SFs) generally fail to prioritize this pose. Many studies have been carried out to tackle this SF problem, offering multiple pose refinement and/or classification methods, albeit with limitations. This study presents a new support vector machine model for pose classification, called “ClassyPose”, which predicts the probability that a receptor‐bound ligand conformation could be near‐native, without any additional pose optimization step. Trained on protein‐ligand extended connectivity features extracted from over 21 600 crystal and docking poses of diverse ligands, this model outperformed other machine‐learning algorithms and three existing SFs in terms of docking power, identifying the native ligand pose as top‐ranked solution for more than 90% of entries in two test sets. It also achieved high specificity (above 0.96), and improved VS performance when used for pose selection. This efficient, user‐friendly tool and all related data are available at https://github.com/vktrannguyen/Classy_Pose.
确定类药物分子的靶标结合构象是药物设计的关键一步，因为它影响着虚拟筛选（VS）的结果，并为 "命中先导"（hit-to-lead）和 "先导优化"（lead optimization）铺平了道路。虽然大多数对接程序通常都能为生物活性分子在其结合口袋内生成至少一个接近原生的姿势，但其集成的经典评分函数（SF）通常无法优先考虑这一姿势。为解决 SF 问题，已有许多研究提供了多种姿势改进和/或分类方法，但这些方法都有局限性。本研究提出了一种新的姿势分类支持向量机模型，称为 "ClassyPose"，它可以预测受体结合配体构象接近原生的概率，而无需任何额外的姿势优化步骤。该模型以从超过 21 600 个不同配体的晶体和对接姿势中提取的蛋白质-配体扩展连接特征为基础进行训练，在对接能力方面优于其他机器学习算法和现有的三种 SF，在两个测试集中，90% 以上的条目都能将原生配体姿势识别为排名靠前的解决方案。它还实现了较高的特异性（高于 0.96），并在用于姿势选择时提高了 VS 性能。这一高效、用户友好的工具和所有相关数据可在 https://github.com/vktrannguyen/Classy_Pose 网站上查阅。
{"title":"ClassyPose: A Machine‐Learning Classification Model for Ligand Pose Selection Applied to Virtual Screening in Drug Discovery","authors":"V. Tran-Nguyen, A. Camproux, Olivier Taboureau","doi":"10.1002/aisy.202400238","DOIUrl":"https://doi.org/10.1002/aisy.202400238","url":null,"abstract":"Determining the target‐bound conformation of a drug‐like molecule is a crucial step in drug design, as it affects the outcome of virtual screening (VS), and paves the way for hit‐to‐lead and lead optimization. While most docking programs usually manage to produce at least a near‐native pose for a bioactive molecule inside its binding pocket, their integrated classical scoring functions (SFs) generally fail to prioritize this pose. Many studies have been carried out to tackle this SF problem, offering multiple pose refinement and/or classification methods, albeit with limitations. This study presents a new support vector machine model for pose classification, called “ClassyPose”, which predicts the probability that a receptor‐bound ligand conformation could be near‐native, without any additional pose optimization step. Trained on protein‐ligand extended connectivity features extracted from over 21 600 crystal and docking poses of diverse ligands, this model outperformed other machine‐learning algorithms and three existing SFs in terms of docking power, identifying the native ligand pose as top‐ranked solution for more than 90% of entries in two test sets. It also achieved high specificity (above 0.96), and improved VS performance when used for pose selection. This efficient, user‐friendly tool and all related data are available at https://github.com/vktrannguyen/Classy_Pose.","PeriodicalId":7187,"journal":{"name":"Advanced Intelligent Systems","volume":"110 8","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140986898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：