Glioma survival risk prediction is of great significance for the individualized treatment and assessment programs.Currently,most deep learning based survival prediction paradigms rely on invasive and expensive histopa...Glioma survival risk prediction is of great significance for the individualized treatment and assessment programs.Currently,most deep learning based survival prediction paradigms rely on invasive and expensive histopathology and genomics methods.However,magnetic resonance imaging(MRI)has emerged as a promising non-invasive alternative with significant prognostic potential.To leverage the benefits of MRI,we propose a segmentation-guided fully automated multimodal MRI-based survival network(SGS-Net),which can simultaneously perform glioma segmentation and survival risk prediction.Specifically,the task interrelation is addressed using a hybrid convolutional neural network-Transformer(CNN-Transformer)encoder to represent the shared high-level semantic features by co-training a decoder for glioma segmentation and a Cox model for survival prediction.Then,to ensure the effective representation of the high-level features,glioma segmentation as an auxiliary task is utilized to guide survival prediction by jointly optimizing the segmentation loss and the Cox partial log-likelihood loss.Furthermore,a pair-wise ranking loss is designed to allow the network to learn the survival difference between patients.To balance the multi-task losses,an uncertain weight manner is adopted to adaptively adjust the weights for preventing task bias.Finally,the proposed SGS-Net is assessed using a publicly available multi-institutional dataset.Experimental and visual results show that SGS-Net achieves promising segmentation performance and obtains a C-index of 81.07%for survival risk prediction,which outperforms several existing state-of-the-art methods and even histopathology-based methods.In addition,Kaplan-Meier survival analysis confirms that the prognosis risk generated by SGS-Net is consistent with the prior prognosis based on the grading or genotyping paradigms.展开更多
Gliomas are the most common malignant tumors in the central nervous system and are known for their inherent diversity and propensity to invade surrounding tissue.These features pose significant challenges in diagnosin...Gliomas are the most common malignant tumors in the central nervous system and are known for their inherent diversity and propensity to invade surrounding tissue.These features pose significant challenges in diagnosing and treating these tumors.Magnetic resonance imaging(MRI)has not only remained at the forefront of glioma management but has also evolved significantly with the advent of multimodal MRI.The rise of multimodal MRI represents a pivotal leap forward,as it seamlessly integrates diverse MRI sequences and advanced techniques to offer an unprecedented,comprehensive,and multidimensional glimpse into the complexities of glioma pathology,including encompassing structural,functional,and even molecular imaging.This holistic approach empowers clinicians with a deeper understanding of tumor characteristics,enabling more precise diagnoses,tailored treatment strategies,and enhanced monitoring capabilities,ultimately improving patient outcomes.Looking ahead,the integration of artificial intelligence(AI)with MRI data heralds a new era of unparalleled precision in glioma diagnosis and therapy.This integration holds the promise to revolutionize the field,enabling more sophisticated analyses that fully leverage all aspects of multimodal MRI.In summary,with the continuous advancement of multimodal MRI techniques and future deep integrations with artificial intelligence,glioma care is poised to evolve toward increasingly personalized,precise,and efficacious strategies.展开更多
Purpose: Magnetic resonance imaging (MRI) is the gold standard in visualizing brain tumors and their effects on adjacent structures. However, no reliable information concerning different tumor components and borders b...Purpose: Magnetic resonance imaging (MRI) is the gold standard in visualizing brain tumors and their effects on adjacent structures. However, no reliable information concerning different tumor components and borders between perifocal edema and infiltration areas can be received. The aim of the study was to establish and evaluate a multimodal imaging concept, in order to differentiate different biological tumor components and to determine tumor borders. Materials and Methods: 12 patients with cerebral gliomas (four low and eight high grade) received a “morphological” MRI, a 3D MR spectroscopy and a T2* MR perfusion examination prior to surgery. Data was evaluated by defining different tumor components, which were entitled based upon their multimodal characteristics and histological data. Results: In high grade gliomas different components can be differentiated, which were described as: “true edema”, “cellular proliferation”, “vascular proliferation”, “cellular infiltration”, “tumor” and “necrosis”. In low grade gliomas, four different tumor components were found: “true edema”, “cellular infiltration”, “cellular proliferation” and “tumor”. Conclusion: With the applied multimodal imaging and a novel evaluation concept, it was possible to detect different tumor components, which could be helpful in detecting the optimal sites for tumor biopsy. Especially in morphological “edema appearing” sites, this knowledge could be important for the adaption of tumor resection borders and the planning of radiation therapy. Further studies with more patients and histological correlation are needed.展开更多
Gliomas,the most prevalent primary brain tumors,require accurate segmentation for diagnosis and risk assess-ment.In this paper,we develop a novel deep learning-based method,the Dynamic Hierarchical Attention for Impro...Gliomas,the most prevalent primary brain tumors,require accurate segmentation for diagnosis and risk assess-ment.In this paper,we develop a novel deep learning-based method,the Dynamic Hierarchical Attention for Improved Segmentation and Survival Prognosis(DHA-ISSP)model.The DHA-ISSP model combines a three-band 3D convolutional neural network(CNN)U-Net architecture with dynamic hierarchical attention mechanisms,enabling precise tumor segmentation and survival prediction.The DHA-ISSP model captures fine-grained details and contextual information by leveraging attention mechanisms at multiple levels,enhancing segmentation accuracy.By achieving remarkable results,our approach surpasses 369 competing teams in the 2020 Multimodal Brain Tumor Segmentation Challenge.With a Dice similarity coefficient of 0.89 and a Hausdorff distance of 4.8 mm,the DHA-ISSP model demonstrates its effectiveness in accurately segmenting brain tumors.We also extract radio mic characteristics from the segmented tumor areas using the DHA-ISSP model.By applying cross-validation of decision trees to the selected features,we identify crucial predictors for glioma survival,enabling personalized treatment strategies.Utilizing the DHA-ISSP model and the desired features,we assess patients’overall survival and categorize survivors into short,mid,in addition to long survivors.The proposed work achieved impressive performance metrics,including the highest accuracy of 0.91,precision of 0.84,recall of 0.92,F1 score of 0.88,specificity of 0.94,sensitivity of 0.92,area under the curve(AUC)value of 0.96,and the lowest mean absolute error value of 0.09 and mean squared error value of 0.18.These results clearly demonstrate the superiority of the proposed system in accurately segmenting brain tumors and predicting survival outcomes,highlighting its significant merit and potential for clinical applications.展开更多
Integrase strand transfer inhibitors(INSTIs)have emerged as the first‐line choice for treating human immunodeficiency virus(HIV)infection due to their superior efficacy and safety.However,the impact of INSTIs on the ...Integrase strand transfer inhibitors(INSTIs)have emerged as the first‐line choice for treating human immunodeficiency virus(HIV)infection due to their superior efficacy and safety.However,the impact of INSTIs on the development of neuropsychiatric conditions in people living with HIV(PLWH)is not fully understood due to limited data.In this study,we conducted a cross‐sectional examination of PLWH receiving antiretroviral therapy,with a specific focus on HIV‐positive men who have sex with men(MSM)on INSTI‐based regimens(n=61)and efavirenz(EFV)‐based regimens(n=28).Participants underwent comprehensive neuropsychiatric evaluations and multimodal magnetic resonance imaging(MRI)scans,including T1‐weighted images and resting‐state functional MRI.Compared to the EFV group,the INSTI group exhibited primarily reduced gray matter volume(GMV)in the right superior parietal gyrus,higher regional homogeneity(ReHo)in the left postcentral gyrus,lower ReHo in the right orbital part of the inferior frontal gyrus,and increased voxel‐wise functional connectivity for the seed region in the left inferior temporal gyrus with clusters in the right cuneus.Furthermore,the analysis revealed a main effect of antiretroviral drugs on GMV changes,but no main effect of neuropsychiatric disorders or their interaction.The repeated analysis of participants who did not switch regimens confirmed the GMV changes in the INSTI group,validating the initial findings.Our study demonstrated gray matter atrophy and functional brain changes in PLWH on INSTI‐based regimens compared to those on EFV‐based regimens.These neuroimaging results provide valuable insights into the characteristics of brain network modifications in PLWH receiving INSTI‐based regimens。展开更多
基金supported in part by the National Key Research and Development Program of China(No.2021YFF1201200)National Natural Science Foundation of China(Nos.62302119,62172444,and 62476291)+3 种基金Science and Technology Innovation Program of Hunan Province(No.2022RC1031)Guizhou Provincial Basic Research Program(Natural Science)(No.QKHZK[2024]603)Guiyang City Science and Technology Plan Project(No.[2024]2-18)High Performance Computing Center of Central South University.
文摘Glioma survival risk prediction is of great significance for the individualized treatment and assessment programs.Currently,most deep learning based survival prediction paradigms rely on invasive and expensive histopathology and genomics methods.However,magnetic resonance imaging(MRI)has emerged as a promising non-invasive alternative with significant prognostic potential.To leverage the benefits of MRI,we propose a segmentation-guided fully automated multimodal MRI-based survival network(SGS-Net),which can simultaneously perform glioma segmentation and survival risk prediction.Specifically,the task interrelation is addressed using a hybrid convolutional neural network-Transformer(CNN-Transformer)encoder to represent the shared high-level semantic features by co-training a decoder for glioma segmentation and a Cox model for survival prediction.Then,to ensure the effective representation of the high-level features,glioma segmentation as an auxiliary task is utilized to guide survival prediction by jointly optimizing the segmentation loss and the Cox partial log-likelihood loss.Furthermore,a pair-wise ranking loss is designed to allow the network to learn the survival difference between patients.To balance the multi-task losses,an uncertain weight manner is adopted to adaptively adjust the weights for preventing task bias.Finally,the proposed SGS-Net is assessed using a publicly available multi-institutional dataset.Experimental and visual results show that SGS-Net achieves promising segmentation performance and obtains a C-index of 81.07%for survival risk prediction,which outperforms several existing state-of-the-art methods and even histopathology-based methods.In addition,Kaplan-Meier survival analysis confirms that the prognosis risk generated by SGS-Net is consistent with the prior prognosis based on the grading or genotyping paradigms.
基金funded by Zhejiang Traditional Chinese Medicine Science and Technology Plan Project,grant number 2023ZL073the Key Science and Technology Plan of the Coconstruction Project of the National Traditional Chinese Medicine Administration Science and Technology Department and Zhejiang Province Traditional Chinese Medicine Administration,grant number GZY-ZJ-KJ-24021.
文摘Gliomas are the most common malignant tumors in the central nervous system and are known for their inherent diversity and propensity to invade surrounding tissue.These features pose significant challenges in diagnosing and treating these tumors.Magnetic resonance imaging(MRI)has not only remained at the forefront of glioma management but has also evolved significantly with the advent of multimodal MRI.The rise of multimodal MRI represents a pivotal leap forward,as it seamlessly integrates diverse MRI sequences and advanced techniques to offer an unprecedented,comprehensive,and multidimensional glimpse into the complexities of glioma pathology,including encompassing structural,functional,and even molecular imaging.This holistic approach empowers clinicians with a deeper understanding of tumor characteristics,enabling more precise diagnoses,tailored treatment strategies,and enhanced monitoring capabilities,ultimately improving patient outcomes.Looking ahead,the integration of artificial intelligence(AI)with MRI data heralds a new era of unparalleled precision in glioma diagnosis and therapy.This integration holds the promise to revolutionize the field,enabling more sophisticated analyses that fully leverage all aspects of multimodal MRI.In summary,with the continuous advancement of multimodal MRI techniques and future deep integrations with artificial intelligence,glioma care is poised to evolve toward increasingly personalized,precise,and efficacious strategies.
文摘Purpose: Magnetic resonance imaging (MRI) is the gold standard in visualizing brain tumors and their effects on adjacent structures. However, no reliable information concerning different tumor components and borders between perifocal edema and infiltration areas can be received. The aim of the study was to establish and evaluate a multimodal imaging concept, in order to differentiate different biological tumor components and to determine tumor borders. Materials and Methods: 12 patients with cerebral gliomas (four low and eight high grade) received a “morphological” MRI, a 3D MR spectroscopy and a T2* MR perfusion examination prior to surgery. Data was evaluated by defining different tumor components, which were entitled based upon their multimodal characteristics and histological data. Results: In high grade gliomas different components can be differentiated, which were described as: “true edema”, “cellular proliferation”, “vascular proliferation”, “cellular infiltration”, “tumor” and “necrosis”. In low grade gliomas, four different tumor components were found: “true edema”, “cellular infiltration”, “cellular proliferation” and “tumor”. Conclusion: With the applied multimodal imaging and a novel evaluation concept, it was possible to detect different tumor components, which could be helpful in detecting the optimal sites for tumor biopsy. Especially in morphological “edema appearing” sites, this knowledge could be important for the adaption of tumor resection borders and the planning of radiation therapy. Further studies with more patients and histological correlation are needed.
基金study conception and design:S.Kannan,S.Anusuyadata collection:S.Kannan+1 种基金analysis and interpretation of results:S.Kannan,S.Anusuyadraft manuscript preparation:S.Kannan.All authors reviewed the results and approved the final version of the manuscript.
文摘Gliomas,the most prevalent primary brain tumors,require accurate segmentation for diagnosis and risk assess-ment.In this paper,we develop a novel deep learning-based method,the Dynamic Hierarchical Attention for Improved Segmentation and Survival Prognosis(DHA-ISSP)model.The DHA-ISSP model combines a three-band 3D convolutional neural network(CNN)U-Net architecture with dynamic hierarchical attention mechanisms,enabling precise tumor segmentation and survival prediction.The DHA-ISSP model captures fine-grained details and contextual information by leveraging attention mechanisms at multiple levels,enhancing segmentation accuracy.By achieving remarkable results,our approach surpasses 369 competing teams in the 2020 Multimodal Brain Tumor Segmentation Challenge.With a Dice similarity coefficient of 0.89 and a Hausdorff distance of 4.8 mm,the DHA-ISSP model demonstrates its effectiveness in accurately segmenting brain tumors.We also extract radio mic characteristics from the segmented tumor areas using the DHA-ISSP model.By applying cross-validation of decision trees to the selected features,we identify crucial predictors for glioma survival,enabling personalized treatment strategies.Utilizing the DHA-ISSP model and the desired features,we assess patients’overall survival and categorize survivors into short,mid,in addition to long survivors.The proposed work achieved impressive performance metrics,including the highest accuracy of 0.91,precision of 0.84,recall of 0.92,F1 score of 0.88,specificity of 0.94,sensitivity of 0.92,area under the curve(AUC)value of 0.96,and the lowest mean absolute error value of 0.09 and mean squared error value of 0.18.These results clearly demonstrate the superiority of the proposed system in accurately segmenting brain tumors and predicting survival outcomes,highlighting its significant merit and potential for clinical applications.
基金supported by the National Natural Science Foundation of China(82072271,82241072,82072294)the National Key Research and Development Program of China(2021YFC2501402,2021YFC0122601)+8 种基金the Beijing Natural Science Foundation(7222095,7222091)the Peak Talent Program of Beijing Hospital Authority(DFL20191701)the Capital’s Funds for Health Improvement and Research(2022-1-1151)the Research and Translational Application of Clinical Characteristic Diagnostic and Treatment Techniques in Capital City(Z221100007422055)the Beijing Research Center for Respiratory Infectious Diseases(BJRID2024-001)the Beijing Hospitals Authority Innovation Studio of Young Staff Funding Support(2021037)the High-level Public Health Technical Personnel Construction Project(2022-1-007)the High-level Public Health Specialized Talents Project of Beijing Municipal Health commission(2022-02-20)the Beijing Key Laboratory for HIV/AIDS Research(BZ0089).
文摘Integrase strand transfer inhibitors(INSTIs)have emerged as the first‐line choice for treating human immunodeficiency virus(HIV)infection due to their superior efficacy and safety.However,the impact of INSTIs on the development of neuropsychiatric conditions in people living with HIV(PLWH)is not fully understood due to limited data.In this study,we conducted a cross‐sectional examination of PLWH receiving antiretroviral therapy,with a specific focus on HIV‐positive men who have sex with men(MSM)on INSTI‐based regimens(n=61)and efavirenz(EFV)‐based regimens(n=28).Participants underwent comprehensive neuropsychiatric evaluations and multimodal magnetic resonance imaging(MRI)scans,including T1‐weighted images and resting‐state functional MRI.Compared to the EFV group,the INSTI group exhibited primarily reduced gray matter volume(GMV)in the right superior parietal gyrus,higher regional homogeneity(ReHo)in the left postcentral gyrus,lower ReHo in the right orbital part of the inferior frontal gyrus,and increased voxel‐wise functional connectivity for the seed region in the left inferior temporal gyrus with clusters in the right cuneus.Furthermore,the analysis revealed a main effect of antiretroviral drugs on GMV changes,but no main effect of neuropsychiatric disorders or their interaction.The repeated analysis of participants who did not switch regimens confirmed the GMV changes in the INSTI group,validating the initial findings.Our study demonstrated gray matter atrophy and functional brain changes in PLWH on INSTI‐based regimens compared to those on EFV‐based regimens.These neuroimaging results provide valuable insights into the characteristics of brain network modifications in PLWH receiving INSTI‐based regimens。
