Accurate brain tumour segmentation is critical for diagnosis and treatment planning, yet challenging due to tumour complexity. Manual segmentation is time-consuming and variable, necessitating automated methods. Deep ...Accurate brain tumour segmentation is critical for diagnosis and treatment planning, yet challenging due to tumour complexity. Manual segmentation is time-consuming and variable, necessitating automated methods. Deep learning, particularly 3D U-Net architectures, has revolutionised medical image analysis by leveraging volumetric data to capture spatial context, enhancing segmentation accuracy. This paper reviews brain tumour segmentation methods, emphasising 3D U-Net advancements. We analyse contributions from the Brain Tumour Segmentation (BraTS) challenges (2014-2023), highlighting key improvements and persistent challenges, including tumour heterogeneity, limited annotated data, varied imaging protocols, computational constraints, and model generalisation. Unlike previous reviews, we synthesise these challenges, proposing targeted research directions: enhancing model robustness through domain adaptation and multi-institutional data sharing, developing lightweight architectures for clinical deployment, integrating multi-modal and clinical data, and incorporating explainability techniques to build clinician trust. By addressing these challenges, we aim to guide future research toward developing more robust, generalisable, and clinically applicable segmentation models, ultimately improving patient outcomes in neuro-oncology.展开更多
Brain tumor segmentation from Magnetic Resonance Imaging(MRI)supports neurologists and radiologists in analyzing tumors and developing personalized treatment plans,making it a crucial yet challenging task.Supervised m...Brain tumor segmentation from Magnetic Resonance Imaging(MRI)supports neurologists and radiologists in analyzing tumors and developing personalized treatment plans,making it a crucial yet challenging task.Supervised models such as 3D U-Net perform well in this domain,but their accuracy significantly improves with appropriate preprocessing.This paper demonstrates the effectiveness of preprocessing in brain tumor segmentation by applying a pre-segmentation step based on the Generalized Gaussian Mixture Model(GGMM)to T1 contrastenhanced MRI scans from the BraTS 2020 dataset.The Expectation-Maximization(EM)algorithm is employed to estimate parameters for four tissue classes,generating a new pre-segmented channel that enhances the training and performance of the 3DU-Net model.The proposed GGMM+3D U-Net framework achieved a Dice coefficient of 0.88 for whole tumor segmentation,outperforming both the standard multiscale 3D U-Net(0.84)and MMU-Net(0.85).It also delivered higher Intersection over Union(IoU)scores compared to models trained without preprocessing or with simpler GMM-based segmentation.These results,supported by qualitative visualizations,suggest that GGMM-based preprocessing should be integrated into brain tumor segmentation pipelines to optimize performance.展开更多
The deep convolutional neural network U-net has been introduced into adaptive subtraction, which is a critical step in effectively suppressing seismic multiples. The U-net approach has higher precision than the tradit...The deep convolutional neural network U-net has been introduced into adaptive subtraction, which is a critical step in effectively suppressing seismic multiples. The U-net approach has higher precision than the traditional linear regression approach. However, the existing 2D U-net approach with 2D data windows can not deal with elaborate discrepancies between the actual and simulated multiples along the gather direction. It may lead to erroneous preservation of primaries or generate obvious vestigial multiples, especially in complex media. To further enhance the multiple suppression accuracy, we present an adaptive subtraction approach utilizing 3D U-net architecture, which can adaptively separate primaries and multiples utilizing 3D windows. The utilization of 3D windows allows for enhanced depiction of spatial continuity and anisotropy of seismic events along the gather direction in comparison to 2D windows. The 3D U-net approach with 3D windows can more effectively preserve the continuity of primaries and manage the complex disparities between the actual and simulated multiples. The proposed 3D U-net approach exhibits 1 dB improvement in the signal-to-noise ratio compared to the 2D U-net approach, as observed in the synthesis data section, and exhibits more outstanding performance in the preservation of primaries and removal of residual multiples in both synthesis and reality data sections. Moreover, to expedite network training in our proposed 3D U-net approach we employ the transfer learning (TL) strategy by utilizing the network parameters of 3D U-net estimated in the preceding data segment as the initial network parameters of 3D U-net for the subsequent data segment. In the reality data section, the 3D U-net approach incorporating TL reduces the computational expense by 70% compared to the one without TL.展开更多
With the intensification of global warming,marine heatwaves(MHWs)have emerged as a significant extreme hazard,garnering widespread attention and creating a pressing need for accurate prediction.The development of arti...With the intensification of global warming,marine heatwaves(MHWs)have emerged as a significant extreme hazard,garnering widespread attention and creating a pressing need for accurate prediction.The development of artificial intelligence,particularly the application of deep learning to sea surface temperature(SST),has significantly improved the feasibility of predictions.This study utilizes SST and Outgoing Longwave Radiation(OLR)data to train a 3D U-Net model for predicting MHWs in the South China Sea(SCS)with lead times ranging from 1 to 7 days,based on the characteristics of intraseasonal weather processes.Analysis of MHWs occurrences from 1982 to 2023 reveals distinct seasonal patterns,with summer MHWs primarily concentrated in the northern and central SCS,and the highest temperature centers located in the Gulf of Tonkin and west of the Philippines.The 2023 MHW forecast results demonstrate that the 3D U-Net model achieves low error rates and high correlation coefficients with observational data.Incorporating OLR data enhances forecast accuracy compared to SST-only inputs,and training the model exclusively with summer data further improves prediction accuracy.These findings indicate that the proposed method can significantly enhance the accuracy of MHW forecasts.展开更多
The gravity inversion is to restore genetic density distribution of the underground target to be explored for explaining the internal structure and distribution of the Earth.In this paper,we propose a new 3D gravity i...The gravity inversion is to restore genetic density distribution of the underground target to be explored for explaining the internal structure and distribution of the Earth.In this paper,we propose a new 3D gravity inversion method based on 3D U-Net++.Compared with two-dimensional gravity inversion,three-dimensional(3D)gravity inversion can more precisely describe the density distribution of underground space.However,conventional 3D gravity inversion method input is two-dimensional,the input and output of the network proposed in our method are three-dimensional.In the training stage,we design a large number of diversifi ed simulation model-data pairs by using the random walk method to improve the generalization ability of the network.In the test phase,we verify the network performance by using the model-data pairs generated by the simulation.To further illustrate the eff ectiveness of the algorithm,we apply the method to the inversion of the San Nicolas mining area,and the inversion results are basically consistent with the borehole measurement results.Moreover,the results of the 3D U-Net++inversion and the 3D U-Net inversion are compared.The density models of the 3D U-Net++inversion have higher resolution,more concentrated inversion results,and a clearer boundary of the density model.展开更多
With the widespread application of deep learning in the field of computer vision,gradually allowing medical image technology to assist doctors in making diagnoses has great practical and research significance.Aiming a...With the widespread application of deep learning in the field of computer vision,gradually allowing medical image technology to assist doctors in making diagnoses has great practical and research significance.Aiming at the shortcomings of the traditional U-Net model in 3D spatial information extraction,model over-fitting,and low degree of semantic information fusion,an improved medical image segmentation model has been used to achieve more accurate segmentation of medical images.In this model,we make full use of the residual network(ResNet)to solve the over-fitting problem.In order to process and aggregate data at different scales,the inception network is used instead of the traditional convolutional layer,and the dilated convolution is used to increase the receptive field.The conditional random field(CRF)can complete the contour refinement work.Compared with the traditional 3D U-Net network,the segmentation accuracy of the improved liver and tumor images increases by 2.89%and 7.66%,respectively.As a part of the image processing process,the method in this paper not only can be used for medical image segmentation,but also can lay the foundation for subsequent image 3D reconstruction work.展开更多
文摘Accurate brain tumour segmentation is critical for diagnosis and treatment planning, yet challenging due to tumour complexity. Manual segmentation is time-consuming and variable, necessitating automated methods. Deep learning, particularly 3D U-Net architectures, has revolutionised medical image analysis by leveraging volumetric data to capture spatial context, enhancing segmentation accuracy. This paper reviews brain tumour segmentation methods, emphasising 3D U-Net advancements. We analyse contributions from the Brain Tumour Segmentation (BraTS) challenges (2014-2023), highlighting key improvements and persistent challenges, including tumour heterogeneity, limited annotated data, varied imaging protocols, computational constraints, and model generalisation. Unlike previous reviews, we synthesise these challenges, proposing targeted research directions: enhancing model robustness through domain adaptation and multi-institutional data sharing, developing lightweight architectures for clinical deployment, integrating multi-modal and clinical data, and incorporating explainability techniques to build clinician trust. By addressing these challenges, we aim to guide future research toward developing more robust, generalisable, and clinically applicable segmentation models, ultimately improving patient outcomes in neuro-oncology.
基金Princess Nourah Bint Abdulrahman University Researchers Supporting Project number(PNURSP2025R826),Princess Nourah Bint Abdulrahman University,Riyadh,Saudi ArabiaNorthern Border University,Saudi Arabia,for supporting this work through project number(NBU-CRP-2025-2933).
文摘Brain tumor segmentation from Magnetic Resonance Imaging(MRI)supports neurologists and radiologists in analyzing tumors and developing personalized treatment plans,making it a crucial yet challenging task.Supervised models such as 3D U-Net perform well in this domain,but their accuracy significantly improves with appropriate preprocessing.This paper demonstrates the effectiveness of preprocessing in brain tumor segmentation by applying a pre-segmentation step based on the Generalized Gaussian Mixture Model(GGMM)to T1 contrastenhanced MRI scans from the BraTS 2020 dataset.The Expectation-Maximization(EM)algorithm is employed to estimate parameters for four tissue classes,generating a new pre-segmented channel that enhances the training and performance of the 3DU-Net model.The proposed GGMM+3D U-Net framework achieved a Dice coefficient of 0.88 for whole tumor segmentation,outperforming both the standard multiscale 3D U-Net(0.84)and MMU-Net(0.85).It also delivered higher Intersection over Union(IoU)scores compared to models trained without preprocessing or with simpler GMM-based segmentation.These results,supported by qualitative visualizations,suggest that GGMM-based preprocessing should be integrated into brain tumor segmentation pipelines to optimize performance.
基金supported by National Natural Science Foundation of China(42364008,41804110)in part by Guizhou Provincial Basic Research Program(Natural Science)(ZK[2022]060)+1 种基金in part by China Postdoctoral Science Foundation(2022M723127)in part by Youth Innovation Team Project of Shandong Provincial Education Department(2022KJ141).
文摘The deep convolutional neural network U-net has been introduced into adaptive subtraction, which is a critical step in effectively suppressing seismic multiples. The U-net approach has higher precision than the traditional linear regression approach. However, the existing 2D U-net approach with 2D data windows can not deal with elaborate discrepancies between the actual and simulated multiples along the gather direction. It may lead to erroneous preservation of primaries or generate obvious vestigial multiples, especially in complex media. To further enhance the multiple suppression accuracy, we present an adaptive subtraction approach utilizing 3D U-net architecture, which can adaptively separate primaries and multiples utilizing 3D windows. The utilization of 3D windows allows for enhanced depiction of spatial continuity and anisotropy of seismic events along the gather direction in comparison to 2D windows. The 3D U-net approach with 3D windows can more effectively preserve the continuity of primaries and manage the complex disparities between the actual and simulated multiples. The proposed 3D U-net approach exhibits 1 dB improvement in the signal-to-noise ratio compared to the 2D U-net approach, as observed in the synthesis data section, and exhibits more outstanding performance in the preservation of primaries and removal of residual multiples in both synthesis and reality data sections. Moreover, to expedite network training in our proposed 3D U-net approach we employ the transfer learning (TL) strategy by utilizing the network parameters of 3D U-net estimated in the preceding data segment as the initial network parameters of 3D U-net for the subsequent data segment. In the reality data section, the 3D U-net approach incorporating TL reduces the computational expense by 70% compared to the one without TL.
基金Guangdong Major Project of Basic and Applied Basic Research(2020B0301030004)。
文摘With the intensification of global warming,marine heatwaves(MHWs)have emerged as a significant extreme hazard,garnering widespread attention and creating a pressing need for accurate prediction.The development of artificial intelligence,particularly the application of deep learning to sea surface temperature(SST),has significantly improved the feasibility of predictions.This study utilizes SST and Outgoing Longwave Radiation(OLR)data to train a 3D U-Net model for predicting MHWs in the South China Sea(SCS)with lead times ranging from 1 to 7 days,based on the characteristics of intraseasonal weather processes.Analysis of MHWs occurrences from 1982 to 2023 reveals distinct seasonal patterns,with summer MHWs primarily concentrated in the northern and central SCS,and the highest temperature centers located in the Gulf of Tonkin and west of the Philippines.The 2023 MHW forecast results demonstrate that the 3D U-Net model achieves low error rates and high correlation coefficients with observational data.Incorporating OLR data enhances forecast accuracy compared to SST-only inputs,and training the model exclusively with summer data further improves prediction accuracy.These findings indicate that the proposed method can significantly enhance the accuracy of MHW forecasts.
基金supported by the Key Laboratory of Geological Survey and Evaluation of Ministry of Education (China University of Geosciences)(No. GLAB2020ZR13)
文摘The gravity inversion is to restore genetic density distribution of the underground target to be explored for explaining the internal structure and distribution of the Earth.In this paper,we propose a new 3D gravity inversion method based on 3D U-Net++.Compared with two-dimensional gravity inversion,three-dimensional(3D)gravity inversion can more precisely describe the density distribution of underground space.However,conventional 3D gravity inversion method input is two-dimensional,the input and output of the network proposed in our method are three-dimensional.In the training stage,we design a large number of diversifi ed simulation model-data pairs by using the random walk method to improve the generalization ability of the network.In the test phase,we verify the network performance by using the model-data pairs generated by the simulation.To further illustrate the eff ectiveness of the algorithm,we apply the method to the inversion of the San Nicolas mining area,and the inversion results are basically consistent with the borehole measurement results.Moreover,the results of the 3D U-Net++inversion and the 3D U-Net inversion are compared.The density models of the 3D U-Net++inversion have higher resolution,more concentrated inversion results,and a clearer boundary of the density model.
文摘With the widespread application of deep learning in the field of computer vision,gradually allowing medical image technology to assist doctors in making diagnoses has great practical and research significance.Aiming at the shortcomings of the traditional U-Net model in 3D spatial information extraction,model over-fitting,and low degree of semantic information fusion,an improved medical image segmentation model has been used to achieve more accurate segmentation of medical images.In this model,we make full use of the residual network(ResNet)to solve the over-fitting problem.In order to process and aggregate data at different scales,the inception network is used instead of the traditional convolutional layer,and the dilated convolution is used to increase the receptive field.The conditional random field(CRF)can complete the contour refinement work.Compared with the traditional 3D U-Net network,the segmentation accuracy of the improved liver and tumor images increases by 2.89%and 7.66%,respectively.As a part of the image processing process,the method in this paper not only can be used for medical image segmentation,but also can lay the foundation for subsequent image 3D reconstruction work.
