Publications by authors named "Heng Huang"

155 Publications

Improved Prediction of Cognitive Outcomes via Globally Aligned Imaging Biomarker Enrichments Over Progressions.

IEEE Trans Biomed Eng 2021 Apr 5;PP. Epub 2021 Apr 5.

Objective: Longitudinal neuroimaging data have been widely used to predict clinical scores for automatic diagnosis of Alzheimer's Disease (AD) in recent years. However, incomplete temporal neuroimaging records of the patients pose a major challenge to use these data for accurately diagnosing AD. In this paper, we propose a novel method to learn an enriched representation for imaging biomarkers, which simultaneously captures the information conveyed by both the baseline neuroimaging records of all the participants in a studied cohort and the progressive variations of the available follow-up records of every individual participant.

Methods: Taking into account that different participants usually take different numbers of medical records at different time points, we develop a robust learning objective that minimizes the summations of a number of not-squared L2-norm distances, which, though, is difficult to efficiently solve in general. Thus we derive a new efficient iterative algorithm with rigorously proved convergence.

Results: We have conducted extensive experiments using the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. Clear performance gains have been achieved when we predict different cognitive scores using the enriched biomarker representations learned by our new method. We further observe that the top selected biomarkers by our proposed method are in perfect accordance with the known knowledge in existing clinical AD studies.

Conclusion: All these promising experimental results have demonstrated the effectiveness of our new method.

Significance: We anticipate that our new method is of interest to biomedical engineering communities beyond AD research and have open-sourced the code of our method online.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TBME.2021.3070875DOI Listing
April 2021

4D Modeling of fMRI Data via Spatio-Temporal Convolutional Neural Networks (ST-CNN).

IEEE Trans Cogn Dev Syst 2020 Sep 14;12(3):451-460. Epub 2019 May 14.

Cortical Architecture Imaging and Discovery Lab, Department of Computer Science and Bioimaging Research Center, The University of Georgia, Athens, GA, USA.

Since the human brain functional mechanism has been enabled for investigation by the functional Magnetic Resonance Imaging (fMRI) technology, simultaneous modeling of both the spatial and temporal patterns of brain functional networks from 4D fMRI data has been a fundamental but still challenging research topic for neuroimaging and medical image analysis fields. Currently, general linear model (GLM), independent component analysis (ICA), sparse dictionary learning, and recently deep learning models, are major methods for fMRI data analysis in either spatial or temporal domains, but there are few joint spatial-temporal methods proposed, as far as we know. As a result, the 4D nature of fMRI data has not been effectively investigated due to this methodological gap. The recent success of deep learning applications for functional brain decoding and encoding greatly inspired us in this work to propose a novel framework called spatio-temporal convolutional neural network (ST-CNN) to extract both spatial and temporal characteristics from targeted networks jointly and automatically identify of functional networks. The identification of Default Mode Network (DMN) from fMRI data was used for evaluation of the proposed framework. Results show that only training the framework on one fMRI dataset is sufficiently generalizable to identify the DMN from different datasets of different cognitive tasks and resting state. Further investigation of the results shows that the joint-learning scheme can capture the intrinsic relationship between the spatial and temporal characteristics of DMN and thus it ensures the accurate identification of DMN from independent datasets. The ST-CNN model brings new tools and insights for fMRI analysis in cognitive and clinical neuroscience studies.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/tcds.2019.2916916DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7978010PMC
September 2020

Risk factors for lymph node metastasis in T1 esophageal squamous cell carcinoma: A systematic review and meta-analysis.

World J Gastroenterol 2021 Feb;27(8):737-750

Department of Thoracic Surgery, Affiliated Hospital of North Sichuan Medical College, Nanchong 637000, Sichuan Province, China.

Background: Lymph node metastasis (LNM) affects the application and outcomes of endoscopic resection in T1 esophageal squamous cell carcinoma (ESCC). However, reports of the risk factors for LNM have been controversial.

Aim: To evaluate risk factors for LNM in T1 ESCC.

Methods: We searched Embase, PubMed and Cochrane Library to select studies related to LNM in patients with T1 ESCC. Included studies were divided into LNM and non-LNM groups. We performed a meta-analysis to examine the relationship between LNM and clinicopathologic features. Odds ratio (OR), mean differences and 95% confidence interval (CI) were assessed using a fixed-effects or random-effects model.

Results: Seventeen studies involving a total of 3775 patients with T1 ESCC met the inclusion criteria. After excluding studies with heterogeneity based on influence analysis, tumor size (OR = 1.93, 95%CI = 1.49-2.50, < 0.001), tumor location (OR = 1.46, 95%CI = 1.17-1.82, < 0.001), macroscopic type (OR = 3.17, 95%CI = 2.33-4.31, < 0.001), T1 substage (OR = 6.28, 95%CI = 4.93-8.00, < 0.001), differentiation (OR = 2.11, 95%CI = 1.64-2.72, < 0.001) and lymphovascular invasion (OR = 5.86, 95%CI = 4.60-7.48, < 0.001) were found to be significantly associated with LNM. Conversely, sex, age and infiltrative growth pattern were not identified as risk factors for LNM.

Conclusion: A tumor size > 2 cm, lower location, nonflat macroscopic type, T1b stage, poor differentiation and lymphovascular invasion were associated with LNM in patients with T1 ESCC.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.3748/wjg.v27.i8.737DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7934003PMC
February 2021

Faster Stochastic Quasi-Newton Methods.

IEEE Trans Neural Netw Learn Syst 2021 Mar 5;PP. Epub 2021 Mar 5.

Stochastic optimization methods have become a class of popular optimization tools in machine learning. Especially, stochastic gradient descent (SGD) has been widely used for machine learning problems, such as training neural networks, due to low per-iteration computational complexity. In fact, the Newton or quasi-newton (QN) methods leveraging the second-order information are able to achieve a better solution than the first-order methods. Thus, stochastic QN (SQN) methods have been developed to achieve a better solution efficiently than the stochastic first-order methods by utilizing approximate second-order information. However, the existing SQN methods still do not reach the best known stochastic first-order oracle (SFO) complexity. To fill this gap, we propose a novel faster stochastic QN method (SpiderSQN) based on the variance reduced technique of SIPDER. We prove that our SpiderSQN method reaches the best known SFO complexity of O(n+n1/2ε⁻²) in the finite-sum setting to obtain an ε-first-order stationary point. To further improve its practical performance, we incorporate SpiderSQN with different momentum schemes. Moreover, the proposed algorithms are generalized to the online setting, and the corresponding SFO complexity of O(ε⁻³) is developed, which also matches the existing best result. Extensive experiments on benchmark data sets demonstrate that our new algorithms outperform state-of-the-art approaches for nonconvex optimization.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TNNLS.2021.3056947DOI Listing
March 2021

Scaling Up Generalized Kernel Methods.

IEEE Trans Pattern Anal Mach Intell 2021 Feb 16;PP. Epub 2021 Feb 16.

Kernel methods have achieved tremendous success in the past two decades. In the current big data era, data collection has grown tremendously. However, existing kernel methods are not scalable enough both at the training and predicting steps. To address this challenge, in this paper, we first introduce a general sparse kernel learning formulation based on the random feature approximation, where the loss functions are possibly non-convex. In order to reduce the scale of random features required in experiment, we also use that formulation based on the orthogonal random feature approximation. Then we propose a new asynchronous parallel doubly stochastic algorithm for large scale sparse kernel learning (AsyDSSKL). To the best our knowledge, AsyDSSKL is the first algorithm with the techniques of asynchronous parallel computation and doubly stochastic optimization. We also provide a comprehensive convergence guarantee to AsyDSSKL. Importantly, the experimental results on various large-scale real-world datasets show that, our AsyDSSKL method has the significant superiority on the computational efficiency at the training and predicting steps over the existing kernel methods.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TPAMI.2021.3059702DOI Listing
February 2021

Panoptic Feature Fusion Net: A Novel Instance Segmentation Paradigm for Biomedical and Biological Images.

IEEE Trans Image Process 2021 21;30:2045-2059. Epub 2021 Jan 21.

Instance segmentation is an important task for biomedical and biological image analysis. Due to the complicated background components, the high variability of object appearances, numerous overlapping objects, and ambiguous object boundaries, this task still remains challenging. Recently, deep learning based methods have been widely employed to solve these problems and can be categorized into proposal-free and proposal-based methods. However, both proposal-free and proposal-based methods suffer from information loss, as they focus on either global-level semantic or local-level instance features. To tackle this issue, we present a Panoptic Feature Fusion Net (PFFNet) that unifies the semantic and instance features in this work. Specifically, our proposed PFFNet contains a residual attention feature fusion mechanism to incorporate the instance prediction with the semantic features, in order to facilitate the semantic contextual information learning in the instance branch. Then, a mask quality sub-branch is designed to align the confidence score of each object with the quality of the mask prediction. Furthermore, a consistency regularization mechanism is designed between the semantic segmentation tasks in the semantic and instance branches, for the robust learning of both tasks. Extensive experiments demonstrate the effectiveness of our proposed PFFNet, which outperforms several state-of-the-art methods on various biomedical and biological datasets.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TIP.2021.3050668DOI Listing
January 2021

Allocation Strategies for Seed Nitrogen and Phosphorus in an Alpine Meadow Along an Altitudinal Gradient on the Tibetan Plateau.

Front Plant Sci 2020 9;11:614644. Epub 2020 Dec 9.

Plant Biology Section, School of Integrative Plant Science, Cornell University, Ithaca, NY, United States.

Nitrogen (N) and phosphorus (P) play important roles in many aspects of plant biology. The allocation of N and P in plant vegetative organs (i.e., leaves, stems, and fine roots) is critical to the regulation of plant growth and development. However, how these elements are allocated in seeds is unclear. The aim of this study was to explore the N and P allocation strategies of seeds in an alpine meadow along an altitudinal gradient. We measured the seed N and P contents of 253 herbaceous species in 37 families along an altitudinal gradient (2,000-4,200 m) in the east Tibetan alpine meadow. The geometric means of seed N and P concentrations and N:P ratios were 34.81 mg g, 5.06 mg g, and 6.88, respectively. Seed N and P concentrations varied across major taxonomic groups and among different altitude zones. N:P ratios showed no significant variations among different taxonomic groups with the exception of N-fixing species. The numerical value of the scaling exponent of seed N vs. P was 0.73, thus approaching 3/4, across the entire data set, but varied significantly across major taxonomic groups. In addition, the numerical value of the scaling exponent of N vs. P declined from 0.88 in the high altitude zone to 0.63 in the low altitude zone. These results indicate that the variations in the numerical value of the scaling exponent governing the seed N vs. P scaling relationship varies as a function of major taxonomic groups and among different altitude zones. We speculate that this variation reflects different adaptive strategies for survival and germination in an alpine meadow. If true, the data presented here advance our understanding of plant seed allocation strategies, and have important implications for modeling early plant growth and development.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.3389/fpls.2020.614644DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7756027PMC
December 2020

Effect of Capecitabine Maintenance Therapy Using Lower Dosage and Higher Frequency vs Observation on Disease-Free Survival Among Patients With Early-Stage Triple-Negative Breast Cancer Who Had Received Standard Treatment: The SYSUCC-001 Randomized Clinical Trial.

JAMA 2021 01;325(1):50-58

Department of Medical Oncology, Sun Yat-sen University Cancer Center, the State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangzhou, Guangdong, China.

Importance: Among all subtypes of breast cancer, triple-negative breast cancer has a relatively high relapse rate and poor outcome after standard treatment. Effective strategies to reduce the risk of relapse and death are needed.

Objective: To evaluate the efficacy and adverse effects of low-dose capecitabine maintenance after standard adjuvant chemotherapy in early-stage triple-negative breast cancer.

Design, Setting, And Participants: Randomized clinical trial conducted at 13 academic centers and clinical sites in China from April 2010 to December 2016 and final date of follow-up was April 30, 2020. Patients (n = 443) had early-stage triple-negative breast cancer and had completed standard adjuvant chemotherapy.

Interventions: Eligible patients were randomized 1:1 to receive capecitabine (n = 222) at a dose of 650 mg/m2 twice a day by mouth for 1 year without interruption or to observation (n = 221) after completion of standard adjuvant chemotherapy.

Main Outcomes And Measures: The primary end point was disease-free survival. Secondary end points included distant disease-free survival, overall survival, locoregional recurrence-free survival, and adverse events.

Results: Among 443 women who were randomized, 434 were included in the full analysis set (mean [SD] age, 46 [9.9] years; T1/T2 stage, 93.1%; node-negative, 61.8%) (98.0% completed the trial). After a median follow-up of 61 months (interquartile range, 44-82), 94 events were observed, including 38 events (37 recurrences and 32 deaths) in the capecitabine group and 56 events (56 recurrences and 40 deaths) in the observation group. The estimated 5-year disease-free survival was 82.8% in the capecitabine group and 73.0% in the observation group (hazard ratio [HR] for risk of recurrence or death, 0.64 [95% CI, 0.42-0.95]; P = .03). In the capecitabine group vs the observation group, the estimated 5-year distant disease-free survival was 85.8% vs 75.8% (HR for risk of distant metastasis or death, 0.60 [95% CI, 0.38-0.92]; P = .02), the estimated 5-year overall survival was 85.5% vs 81.3% (HR for risk of death, 0.75 [95% CI, 0.47-1.19]; P = .22), and the estimated 5-year locoregional recurrence-free survival was 85.0% vs 80.8% (HR for risk of locoregional recurrence or death, 0.72 [95% CI, 0.46-1.13]; P = .15). The most common capecitabine-related adverse event was hand-foot syndrome (45.2%), with 7.7% of patients experiencing a grade 3 event.

Conclusions And Relevance: Among women with early-stage triple-negative breast cancer who received standard adjuvant treatment, low-dose capecitabine maintenance therapy for 1 year, compared with observation, resulted in significantly improved 5-year disease-free survival.

Trial Registration: ClinicalTrials.gov Identifier: NCT01112826.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1001/jama.2020.23370DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7729589PMC
January 2021

A GENERALIZED FRAMEWORK OF PATHLENGTH ASSOCIATED COMMUNITY ESTIMATION FOR BRAIN STRUCTURAL NETWORK.

Proc IEEE Int Symp Biomed Imaging 2020 Apr 22;2020:288-291. Epub 2020 May 22.

Department of Electrical and Computer Engineering, University of Pittsburgh, PA, USA.

Diffusion MRI-derived brain structural network has been widely used in brain research and community or modular structure is one of popular network features, which can be extracted from network edge-derived pathlengths. Conceptually, brain structural network edges represent the connecting strength between pair of nodes, thus non-negative. The pathlength. Many studies have demonstrated that each brain network edge can be affected by many confounding factors (e.g. age, sex, etc.) and this influence varies on each edge. However, after applying generalized linear regression to remove those confounding's effects, some network edges may become negative, which leads to barriers in extracting the community structure. In this study, we propose a novel generalized framework to solve this negative edge issue in extracting the modular structure from brain structural network. We have compared our framework with traditional Q method. The results clearly demonstrated that our framework has significant advantages in both stability and sensitivity.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/isbi45749.2020.9098552DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7652406PMC
April 2020

Wavelet-Based Dual Recursive Network for Image Super-Resolution.

IEEE Trans Neural Netw Learn Syst 2020 Oct 27;PP. Epub 2020 Oct 27.

Although remarkable progress has been made on single-image super-resolution (SISR), deep learning methods cannot be easily applied to real-world applications due to the requirement of its heavy computation, especially for mobile devices. Focusing on the fewer parameters and faster inference SISR approach, we propose an efficient and time-saving wavelet transform-based network architecture, where the image super-resolution (SR) processing is carried out in the wavelet domain. Different from the existing methods that directly infer high-resolution (HR) image with the input low-resolution (LR) image, our approach first decomposes the LR image into a series of wavelet coefficients (WCs) and the network learns to predict the corresponding series of HR WCs and then reconstructs the HR image. Particularly, in order to further enhance the relationship between WCs and image deep characteristics, we propose two novel modules [wavelet feature mapping block (WFMB) and wavelet coefficients reconstruction block (WCRB)] and a dual recursive framework for joint learning strategy, thus forming a WCs prediction model to realize the efficient and accurate reconstruction of HR WCs. Experimental results show that the proposed method can outperform state-of-the-art methods with more than a 2x reduction in model parameters and computational complexity.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TNNLS.2020.3028688DOI Listing
October 2020

Multicentre, randomised, open-label, non-inferiority trial comparing the effectiveness and safety of ductal lavage versus oral corticosteroids for idiopathic granulomatous mastitis: a study protocol.

BMJ Open 2020 10 10;10(10):e036643. Epub 2020 Oct 10.

Breast Tumor Center, Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou, Guangdong, China

Introduction: The ideal treatment for idiopathic granulomatous mastitis (IGM) remains unclear. In a prospective, single-centre, pilot study, we reported that ductal lavage treatment for non-lactational mastitis patients had a 1-year clinical complete response (cCR) rate of >90%, without any significant adverse events. Thus, in this multicentre, randomised, open-label, non-inferiority trial, we will aim to compare the effectiveness and safety of ductal lavage vs oral corticosteroids as the first-line treatment for patients with IGM.

Methods And Analysis: The trial will be conducted at the Breast Tumor Center of Sun Yat-sen Memorial Hospital in China and at least at one participating regional centre. We plan to recruit 140 eligible IGM patients who will be randomised into the ductal lavage group or oral corticosteroid group with a 1:1 ratio. The patients in the oral corticosteroid group will receive meprednisone or prednisone for 6 months. The patients in the ductal lavage group will receive ductal lavage and breast massage, as previously reported. All the participants will be followed up at the clinic for 1 year post randomisation. The primary endpoint of this trial will be the 1-year cCR rate, and the secondary endpoints will include the time to cCR, treatment failure rate, relapse rate and protocol compliance rate. The trial was designed to determine whether ductal lavage is non-inferior to oral corticosteroids (1-year cCR rate assumed to be 90%), with a non-inferiority margin of 15%.

Ethics And Dissemination: The ethics committee of Sun Yat-sen Memorial Hospital at Sun Yat-sen University approved the study (2018-Lun-Shen-Yan-No. 30). The results of the trial will be communicated to the participating primary care practices, published in international journals and presented at international clinical and scientific conferences.

Trial Registration Number: ClinicalTrials.gov Registry (NCT03724903); Pre-results.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1136/bmjopen-2019-036643DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7552910PMC
October 2020

Virtual Molecular Projections and Convolutional Neural Networks for the End-to-End Modeling of Nanoparticle Activities and Properties.

Anal Chem 2020 10 7;92(20):13971-13979. Epub 2020 Oct 7.

Center for Computational and Integrative Biology, Rutgers University, 201 S Broadway, Camden, New Jersey 08103, United States.

Digitalizing complex nanostructures into data structures suitable for machine learning modeling without losing nanostructure information has been a major challenge. Deep learning frameworks, particularly convolutional neural networks (CNNs), are especially adept at handling multidimensional and complex inputs. In this study, CNNs were applied for the modeling of nanoparticle activities exclusively from nanostructures. The nanostructures were represented by virtual molecular projections, a multidimensional digitalization of nanostructures, and used as input data to train CNNs. To this end, 77 nanoparticles with various activities and/or physicochemical property results were used for modeling. The resulting CNN model predictions show high correlations with the experimental results. An analysis of a trained CNN quantitatively showed that neurons were able to recognize distinct nanostructure features critical to activities and physicochemical properties. This "end-to-end" deep learning approach is well suited to digitalize complex nanostructures for data-driven machine learning modeling and can be broadly applied to rationally design nanoparticles with desired activities.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1021/acs.analchem.0c02878DOI Listing
October 2020

Large-Scale Nonlinear AUC Maximization via Triply Stochastic Gradients.

IEEE Trans Pattern Anal Mach Intell 2020 Sep 18;PP. Epub 2020 Sep 18.

Learning to improve AUC performance for imbalanced data is an important machine learning research problem. Most methods of AUC maximization assume that the model function is linear in the original feature space. However, this assumption is not suitable for nonlinear separable problems. Although there have been several nonlinear methods of AUC maximization, scaling up nonlinear AUC maximization is still an open question. To address this challenging problem, in this paper, we propose a novel large-scale nonlinear AUC maximization method (named as TSAM) based on the triply stochastic gradient descents. Specifically, we first use the random Fourier feature to approximate the kernel function. After that, we use the triply stochastic gradients w.r.t. the pairwise loss and random feature to iteratively update the solution. Finally, we prove that TSAM converges to the optimal solution with the rate of O(1/t) after t iterations. Experimental results on a variety of benchmark datasets not only confirm the scalability of TSAM, but also show a significant reduction of computational time compared with existing batch learning algorithms, while retaining the similar generalization performance.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TPAMI.2020.3024987DOI Listing
September 2020

[Lateral ligament reconstruction with autogenous partial peroneus longus tendon for chronic lateral ankle instability].

Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi 2020 Sep;34(9):1114-1119

Department of Hand and Foot Surgery, Affiliated Hospital of Qingdao University, Qingdao Shandong, 266000, P.R.China.

Objective: To assess the effectiveness of lateral ligament reconstruction with autogenous partial peroneus longus tendon for chronic lateral ankle instability.

Methods: Between September 2014 and November 2018, 32 patients (32 sides) with chronic lateral ankle instability were treated with lateral ankle ligament reconstruction by using autogenous anterior half of the peroneus longus tendon. There were 25 males and 7 females, with an average age of 28.5 years (range, 20-51 years). The disease duration was 6-41 months (mean, 8.9 months). The preoperative Karlsson-Peterson ankle score was 53.7±9.7. The talar tilt angle was (14.9±3.7)°, and the anterior talar translation was (8.2±2.8) mm. Six patients combined with osteochondral lesion of talus and 4 patients combined with bony impingement.

Results: All incisions healed by first intention postoperatively. All patients were followed up 12-53 months (mean, 22.7 months). At last follow-up, the Karlsson-Peterson ankle score was 85.2±9.6; the talar tilt angle was (4.3±1.4)°; the anterior talar translation was (3.5±1.1) mm. There were significant differences in all indexes between pre- and post-operation ( <0.05). Seventeen patients were very satisfied with the results, 10 patients were satisfied, 4 patients were normal, and 1 patient was unsatisfied. After operation, the ankle sprain occurred in 7 cases, the tenderness around the compression screws at calcaneus in 5 cases, the anterolateral pain of ankle joint over 6 months in 4 cases. No patient had discomfort around the reciepient sites. At last follow-up, the ultrasonography examination showed that there was no significant difference in the density and diameter between bilateral peroneus longus tendons in 12 cases.

Conclusion: For chronic lateral ankle instability, the lateral ankle ligament reconstruction with the autogenous partial peroneus longus tendon is a safe and effective surgical option.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.7507/1002-1892.202002008DOI Listing
September 2020

PDAM: A Panoptic-Level Feature Alignment Framework for Unsupervised Domain Adaptive Instance Segmentation in Microscopy Images.

IEEE Trans Med Imaging 2021 Jan 29;40(1):154-165. Epub 2020 Dec 29.

In this work, we present an unsupervised domain adaptation (UDA) method, named Panoptic Domain Adaptive Mask R-CNN (PDAM), for unsupervised instance segmentation in microscopy images. Since there currently lack methods particularly for UDA instance segmentation, we first design a Domain Adaptive Mask R-CNN (DAM) as the baseline, with cross-domain feature alignment at the image and instance levels. In addition to the image- and instance-level domain discrepancy, there also exists domain bias at the semantic level in the contextual information. Next, we, therefore, design a semantic segmentation branch with a domain discriminator to bridge the domain gap at the contextual level. By integrating the semantic- and instance-level feature adaptation, our method aligns the cross-domain features at the panoptic level. Third, we propose a task re-weighting mechanism to assign trade-off weights for the detection and segmentation loss functions. The task re-weighting mechanism solves the domain bias issue by alleviating the task learning for some iterations when the features contain source-specific factors. Furthermore, we design a feature similarity maximization mechanism to facilitate instance-level feature adaptation from the perspective of representational learning. Different from the typical feature alignment methods, our feature similarity maximization mechanism separates the domain-invariant and domain-specific features by enlarging their feature distribution dependency. Experimental results on three UDA instance segmentation scenarios with five datasets demonstrate the effectiveness of our proposed PDAM method, which outperforms state-of-the-art UDA methods by a large margin.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TMI.2020.3023466DOI Listing
January 2021

Scalable Kernel Ordinal Regression via Doubly Stochastic Gradients.

IEEE Trans Neural Netw Learn Syst 2020 Aug 28;PP. Epub 2020 Aug 28.

Ordinal regression (OR) is one of the most important machine learning tasks. The kernel method is a major technique to achieve nonlinear OR. However, traditional kernel OR solvers are inefficient due to increased complexity introduced by multiple ordinal thresholds as well as the cost of kernel computation. Doubly stochastic gradient (DSG) is a very efficient and scalable kernel learning algorithm that combines random feature approximation with stochastic functional optimization. However, the theory and algorithm of DSG can only support optimization tasks within the unique reproducing kernel Hilbert space (RKHS), which is not suitable for OR problems where the multiple ordinal thresholds usually lead to multiple RKHSs. To address this problem, we construct a kernel whose RKHS can contain the decision function with multiple thresholds. Based on this new kernel, we further propose a novel DSG-like algorithm, DSGOR. In each iteration of DSGOR, we update the decision functional as well as the function bias with appropriately set learning rates for each. Our theoretic analysis shows that DSGOR can achieve O(1/t) convergence rate, which is as good as DSG, even though dealing with a much harder problem. Extensive experimental results demonstrate that our algorithm is much more efficient than traditional kernel OR solvers, especially on large-scale problems.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TNNLS.2020.3015937DOI Listing
August 2020

Efficient Active Learning by Querying Discriminative and Representative Samples and Fully Exploiting Unlabeled Data.

IEEE Trans Neural Netw Learn Syst 2020 Aug 26;PP. Epub 2020 Aug 26.

Active learning is an important learning paradigm in machine learning and data mining, which aims to train effective classifiers with as few labeled samples as possible. Querying discriminative (informative) and representative samples are the state-of-the-art approach for active learning. Fully utilizing a large amount of unlabeled data provides a second chance to improve the performance of active learning. Although there have been several active learning methods proposed by combining with semisupervised learning, fast active learning with fully exploiting unlabeled data and querying discriminative and representative samples is still an open question. To overcome this challenging issue, in this article, we propose a new efficient batch mode active learning algorithm. Specifically, we first provide an active learning risk bound by fully considering the unlabeled samples in characterizing the informativeness and representativeness. Based on the risk bound, we derive a new objective function for batch mode active learning. After that, we propose a wrapper algorithm to solve the objective function, which essentially trains a semisupervised classifier and selects discriminative and representative samples alternately. Especially, to avoid retraining the semisupervised classifier from scratch after each query, we design two unique procedures based on the path-following technique, which can remove multiple queried samples from the unlabeled data set and add the queried samples into the labeled data set efficiently. Extensive experimental results on a variety of benchmark data sets not only show that our algorithm has a better generalization performance than the state-of-the-art active learning approaches but also show its significant efficiency.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TNNLS.2020.3016928DOI Listing
August 2020

Sparse Modal Additive Model.

IEEE Trans Neural Netw Learn Syst 2020 Jul 23;PP. Epub 2020 Jul 23.

Sparse additive models have been successfully applied to high-dimensional data analysis due to the flexibility and interpretability of their representation. However, the existing methods are often formulated using the least-squares loss with learning the conditional mean, which is sensitive to data with the non-Gaussian noises, e.g., skewed noise, heavy-tailed noise, and outliers. To tackle this problem, we propose a new robust regression method, called as sparse modal additive model (SpMAM), by integrating the modal regression metric, the data-dependent hypothesis space, and the weighted ℓq,1-norm regularizer (q ≥ 1) into the additive models. Specifically, the modal regression metric assures the model robustness to complex noises via learning the conditional mode, the data-dependent hypothesis space offers the model adaptivity via sample-based presentation, and the ℓq,1-norm regularizer addresses the algorithmic interpretability via sparse variable selection. In theory, the proposed SpMAM enjoys statistical guarantees on asymptotic consistency for regression estimation and variable selection simultaneously. Experimental results on both synthetic and real-world benchmark data sets validate the effectiveness and robustness of the proposed model.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TNNLS.2020.3005144DOI Listing
July 2020

Water content quantitatively affects metabolic rates over the course of plant ontogeny.

New Phytol 2020 12 10;228(5):1524-1534. Epub 2020 Aug 10.

State Key Laboratory of Grassland Agro-Ecosystem, School of Life Sciences, Lanzhou University, Lanzhou, 730000, China.

Plant metabolism determines the structure and dynamics of ecological systems across many different scales. The metabolic theory of ecology quantitatively predicts the scaling of metabolic rate as a function of body size and temperature. However, the role of tissue water content has been neglected even though hydration significantly affects metabolism, and thus ecosystem structure and functioning. Here, we use a general model based on biochemical kinetics to quantify the combined effects of water content, body size and temperature on plant metabolic rates. The model was tested using a comprehensive dataset from 205 species across 10 orders of magnitude in body size from seeds to mature large trees. We show that water content significantly influences mass-specific metabolic rates as predicted by the model. The scaling exponents of whole-plant metabolic rate vs body size numerically converge onto 1.0 after water content is corrected regardless of body size or ontogenetic stage. The model provides novel insights into how water content together with body size and temperature quantitatively influence plant growth and metabolism, community dynamics and ecosystem energetics.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1111/nph.16808DOI Listing
December 2020

Noninvasive methods for detection of chronic lung allograft dysfunction in lung transplantation.

Transplant Rev (Orlando) 2020 07 8;34(3):100547. Epub 2020 May 8.

Department of Thoracic Surgery, Affiliated Hospital of North Sichuan Medical College, Nanchong, China.

Lung transplantation (LTx) is the only therapeutic option for end-stage lung diseases. Chronic lung allograft dysfunction (CLAD), which manifests as airflow restriction and/or obstruction, is the primary factor limiting the long-term survival of patients after surgery. According to histopathological and radiographic findings, CLAD comprises two phenotypes, bronchiolitis obliterans syndrome and restrictive allograft syndrome. Half of all lung recipients will develop CLAD in 5 years, and this rate may increase up to 75% 10 years after surgery owing to the paucity in accurate and effective early detection and treatment methods. Recently, many studies have presented noninvasive methods for detecting CLAD and improving diagnosis and intervention. However, the significance of accurately detecting CLAD remains controversial. We reviewed published studies that have presented noninvasive methods for detecting CLAD to highlight the current knowledge on clinical symptoms, spirometry, imaging examinations, and other methods to detect the disease.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.trre.2020.100547DOI Listing
July 2020

Critical transition to woody plant dominance through microclimate feedbacks in North American coastal ecosystems.

Ecology 2020 09 26;101(9):e03107. Epub 2020 Jun 26.

Department of Environmental Science, Policy, and Management, University of California, Berkeley, California, 94720, USA.

Climate warming is facilitating the expansion of many cold-sensitive woody species in woodland-grassland ecotones worldwide. Recent research has demonstrated that this range expansion can be further enhanced by positive vegetation-microclimate feedbacks whereby woody canopies induce local nocturnal warming, which reduces freeze-induced damage and favors the establishment of woody plants. However, this local positive feedback can be counteracted by biotic drivers such as browsing and the associated consumption of shrub biomass. The joint effects of large-scale climate warming and local-scale microclimate feedbacks on woody vegetation dynamics in these ecotones remain poorly understood. Here, we used a combination of experimental and modeling approaches to investigate the effects of woody cover on microclimate and the consequent implications on ecological stability in North American coastal ecosystems. We found greater browsing pressure and significant warming (~2°C) beneath shrub canopies compared to adjacent grasslands, which reduces shrub seedlings' exposure to cold damage. Cold sensitivity is evidenced by the significant decline in xylem hydraulic conductivity in shrub seedlings when temperatures dropped below -2°C. Despite the negative browsing-vegetation feedback, a small increase in minimum temperature can induce critical transitions from grass to woody plant dominance. Our framework also predicts the threshold temperature of -7°C for mangrove-salt marsh ecotones on the Atlantic coast of Florida. Above this reference temperature a critical transition may occur from salt marsh to mangrove vegetation, in agreement with empirical studies. Thus, the interaction between ongoing global warming trends and microclimate feedbacks may significantly alter woody vegetation dynamics and ecological stability in coastal ecosystems where woody plant expansion is primarily constrained by extreme low temperature events.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/ecy.3107DOI Listing
September 2020

BREM-SC: a bayesian random effects mixture model for joint clustering single cell multi-omics data.

Nucleic Acids Res 2020 06;48(11):5814-5824

Department of Biostatistics, University of Pittsburgh, Pittsburgh, PA, USA.

Droplet-based single cell transcriptome sequencing (scRNA-seq) technology, largely represented by the 10× Genomics Chromium system, is able to measure the gene expression from tens of thousands of single cells simultaneously. More recently, coupled with the cutting-edge Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq), the droplet-based system has allowed for immunophenotyping of single cells based on cell surface expression of specific proteins together with simultaneous transcriptome profiling in the same cell. Despite the rapid advances in technologies, novel statistical methods and computational tools for analyzing multi-modal CITE-Seq data are lacking. In this study, we developed BREM-SC, a novel Bayesian Random Effects Mixture model that jointly clusters paired single cell transcriptomic and proteomic data. Through simulation studies and analysis of public and in-house real data sets, we successfully demonstrated the validity and advantages of this method in fully utilizing both types of data to accurately identify cell clusters. In addition, as a probabilistic model-based approach, BREM-SC is able to quantify the clustering uncertainty for each single cell. This new method will greatly facilitate researchers to jointly study transcriptome and surface proteins at the single cell level to make new biological discoveries, particularly in the area of immunology.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1093/nar/gkaa314DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7293045PMC
June 2020

Clinical nomogram for lymph node metastasis in pathological T1 esophageal squamous cell carcinoma: a multicenter retrospective study.

Ann Transl Med 2020 Mar;8(6):292

Department of Thoracic Surgery, Affiliated Hospital of North Sichuan Medical College, Nanchong 637000, China.

Background: Endoscopic resection is increasingly used to treat pathological T1 (pT1) esophageal cancer (EC) patients. However, the procedures are limited by lymph node metastasis (LNM) and remain controversial. We aimed to construct a nomogram to predict the risk of LNM in patients with pT1 esophageal squamous cell carcinoma (ESCC).

Methods: A total of 243 patients with pT1 ESCC who underwent esophagectomy and lymph node dissection at two different institutes between February 2013 and June 2019 were analyzed retrospectively. Patients were categorized into the negative group and the positive group according to whether there was LNM. Risk factors for LNM were evaluated by univariate and multivariate analyses. The nomogram was used to estimate the individual risk of LNM.

Results: Forty-six (18.9%) of the 243 patients with pT1 ESCC exhibited LNM. The LNM rate in patients with stage T1a disease was 5.7% (5/88), and the rate in patients with stage T1b disease was 26.5% (41/155). Multivariable logistic regression analysis showed that tumor differentiation [odds ratio (OR) =1.942, 95% confidence interval (CI): 1.067-3.536, P=0.030], the T1 sub-stage (OR =4.750, 95% CI: 1.658-13.611, P=0.004), the preoperative alanine aminotransferase/aspartate aminotransferase ratio (LSR) (OR =5.371, 95% CI: 1.676-17.210, P=0.005), and the high-density lipoprotein cholesterol (HDL-C) level (OR =5.894, 95% CI: 1.917-18.124, P=0.002) were independent risk factors for LNM. The nomogram had relatively high accuracy, with an area under the receiver operating characteristic curve (AUC) of 0.803 (95% CI: 0.732-0.873). The calibration curve showed that the predicted probability of LNM was in good agreement with the actual probability.

Conclusions: Clinicopathological and hematological parameters of tumor differentiation, the T1 sub-stage, the preoperative LSR, and the HDL-C level may predict the risk of LNM in T1 ESCC. The risk of LNM can be predicted by the nomogram.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.21037/atm.2020.02.185DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7186726PMC
March 2020

Deep-learning-based Prediction of Late Age-Related Macular Degeneration Progression.

Nat Mach Intell 2020 Feb 14;2(2):141-150. Epub 2020 Feb 14.

Division of Pulmonary Medicine, Department of Pediatrics, UPMC Children's Hospital of Pittsburgh, University of Pittsburgh, Pittsburgh, PA.

Both genetic and environmental factors influence the etiology of age-related macular degeneration (AMD), a leading cause of blindness. AMD severity is primarily measured by fundus images and recently developed machine learning methods can successfully predict AMD progression using image data. However, none of these methods have utilized both genetic and image data for predicting AMD progression. Here we jointly used genotypes and fundus images to predict an eye as having progressed to late AMD with a modified deep convolutional neural network (CNN). In total, we used 31,262 fundus images and 52 AMD-associated genetic variants from 1,351 subjects from the Age-Related Eye Disease Study (AREDS) with disease severity phenotypes and fundus images available at baseline and follow-up visits over a period of 12 years. Our results showed that fundus images coupled with genotypes could predict late AMD progression with an averaged area under the curve (AUC) value of 0.85 (95%CI: 0.83-0.86). The results using fundus images alone showed an averaged AUC of 0.81 (95%CI: 0.80-0.83). We implemented our model in a cloud-based application for individual risk assessment.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1038/s42256-020-0154-9DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7153739PMC
February 2020

A General Model for Seed and Seedling Respiratory Metabolism.

Am Nat 2020 03 15;195(3):534-546. Epub 2020 Jan 15.

The ontogeny of seed plants usually involves a dormant dehydrated state and the breaking of dormancy and germination, which distinguishes it from that of most organisms. Seed germination and seedling establishment are critical ontogenetic stages in the plant life cycle, and both are fueled by respiratory metabolism. However, the scaling of metabolic rate with respect to individual traits remains poorly understood. Here, we tested metabolic scaling theory during seed germination and early establishment growth using a recently developed model and empirical data collected from 41 species. The results show that (i) the mass-specific respiration rate () was weakly correlated with body mass, mass-specific N content, and mass-specific C content; (ii) conformed to a single Michaelis-Menten curve as a function of tissue water content; and (iii) the central parameters in the model were highly correlated with DNA content and critical enzyme activities. The model offers new insights and a more integrative scaling theory that quantifies the combined effects of tissue water content and body mass on respiratory metabolism during early plant ontogeny.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1086/707072DOI Listing
March 2020

CAM plant expansion favored indirectly by asymmetric climate warming and increased rainfall variability.

Oecologia 2020 May 20;193(1):1-13. Epub 2020 Feb 20.

Department of Environmental Science, Policy, and Management, University of California, Berkeley, CA, 94720, USA.

Recent observational evidence suggests that nighttime temperatures are increasing faster than daytime temperatures, while in some regions precipitation events are becoming less frequent and more intense. The combined ecological impacts of these climatic changes on crassulacean acid metabolism (CAM) plants and their interactions with other functional groups (i.e., grass communities) remain poorly understood. Here we developed a growth chamber experiment to investigate how two CAM-grass communities in desert ecosystems of the southwestern United States and northern Mexico respond to asymmetric warming and increasing rainfall variability. Grasses generally showed competitive advantages over CAM plants with increasing rainfall variability under ambient temperature conditions. In contrast, asymmetric warming caused mortality of both grass species (Bouteloua eriopoda and Bouteloua curtipendula) in both rainfall treatments due to enhanced drought stress. Grass mortality indirectly favored CAM plants even though the biomass of both CAM species Cylindropuntia imbricata and Opuntia phaeacantha significantly decreased. The stem's volume-to-surface ratio of C. imbricata was significantly higher in mixture than in monoculture under ambient temperature (both P < 0.0014); however, the difference became insignificant under asymmetric warming (both P > 0.1625), suggesting that warming weakens the negative effects of interspecific competition on CAM plant growth. Our findings suggest that while the increase in intra-annual rainfall variability enhances grass productivity, asymmetric warming may lead to grass mortality, thereby indirectly favoring the expansion of co-existing CAM plants. This study provides novel experimental evidence showing how the ongoing changes in global warming and rainfall variability affect CAM-grass growth and interactions in dryland ecosystems.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1007/s00442-020-04624-wDOI Listing
May 2020

Hierarchical Organization of Functional Brain Networks Revealed by Hybrid Spatiotemporal Deep Learning.

Brain Connect 2020 03 5;10(2):72-82. Epub 2020 Mar 5.

Cortical Architecture Imaging and Discovery Lab, Department of Computer Science and Bioimaging Research Center, The University of Georgia, Athens, Georgia.

Hierarchical organization of brain function has been an established concept in the neuroscience field for a long time, however, it has been rarely demonstrated how such hierarchical macroscale functional networks are actually organized in the human brain. In this study, to answer this question, we propose a novel methodology to provide an evidence of hierarchical organization of functional brain networks. This article introduces the hybrid spatiotemporal deep learning (HSDL), by jointly using deep belief networks (DBNs) and deep least absolute shrinkage and selection operator (LASSO) to reveal the temporal hierarchical features and spatial hierarchical maps of brain networks based on the Human Connectome Project 900 functional magnetic resonance imaging (fMRI) data sets. Briefly, the key idea of HSDL is to extract the weights between two adjacent layers of DBNs, which are then treated as the hierarchical dictionaries for deep LASSO to identify the corresponding hierarchical spatial maps. Our results demonstrate that both spatial and temporal aspects of dozens of functional networks exhibit multiscale properties that can be well characterized and interpreted based on existing computational tools and neuroscience knowledge. Our proposed novel hybrid deep model is used to provide the first insightful opportunity to reveal the potential hierarchical organization of time series and functional brain networks, using task-based fMRI signals of human brain.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1089/brain.2019.0701DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7099414PMC
March 2020

Critical Transitions in Plant-Pollinator Systems Induced by Positive Inbreeding-Reward-Pollinator Feedbacks.

iScience 2020 Feb 7;23(2):100819. Epub 2020 Jan 7.

Department of Environmental Science, Policy, and Management, University of California, Berkeley, CA 94720, USA.

In many regions of the world pollinator populations are rapidly declining, a trend that is expected to disrupt major ecosystem functions and services. These changes in pollinator abundance may be prone to critical transitions with abrupt shifts to a state strongly depleted both in pollinator and vegetation abundance. Here we develop a process-based model to investigate the effect of a positive pollinator-vegetation feedback, whereby an initial decline in plant density increases selfing thereby reducing floral resources and negatively affecting pollinators. We show that a decline in resource availability and an increase in disturbance intensity can induce an abrupt shift in vegetation and pollinator dynamics and potentially lead to the collapse of plant-pollinator systems. Thus, endogenous feedbacks can induce strong non-linearities in plant-pollinator dynamics, making them vulnerable to critical transitions to a state depleted of both plants and pollinators in response to resource deficiency and natural or anthropogenic disturbance.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.isci.2020.100819DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6976937PMC
February 2020

Robust Cumulative Crowdsourcing Framework Using New Incentive Payment Function and Joint Aggregation Model.

IEEE Trans Neural Netw Learn Syst 2020 11 29;31(11):4610-4621. Epub 2020 Oct 29.

In recent years, crowdsourcing has gained tremendous attention in the machine learning community due to the increasing demand for labeled data. However, the labels collected by crowdsourcing are usually unreliable and noisy. This issue is mainly caused by: 1) nonflexible data collection mechanisms; 2) nonincentive payment functions; and 3) inexpert crowd workers. We propose a new robust crowdsourcing framework as a comprehensive solution for all these challenging problems. Our unified framework consists of three novel components. First, we introduce a new flexible data collection mechanism based on the cumulative voting system, allowing crowd workers to express their confidence for each option in multi-choice questions. Second, we design a novel payment function regarding the settings of our data collection mechanism. The payment function is theoretically proved to be incentive-compatible, encouraging crowd workers to disclose truthfully their beliefs to get the maximum payment. Third, we propose efficient aggregation models, which are compatible with both single-option and multi-option crowd labels. We define a new aggregation model, called simplex constrained majority voting (SCMV), and enhance it by using the probabilistic generative model. Furthermore, fast optimization algorithms are derived for the proposed aggregation models. Experimental results indicate higher quality for the crowd labels collected by our proposed mechanism without increasing the cost. Our aggregation models also outperform the state-of-the-art models on multiple crowdsourcing data sets in terms of accuracy and convergence speed.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TNNLS.2019.2956523DOI Listing
November 2020

Joint Multi-Modal Longitudinal Regression and Classification for Alzheimer's Disease Prediction.

IEEE Trans Med Imaging 2020 06 13;39(6):1845-1855. Epub 2019 Dec 13.

Alzheimer's disease (AD) is a serious neurodegenerative condition that affects millions of individuals across the world. As the average age of individuals in the United States and the world increases, the prevalence of AD will continue to grow. To address this public health problem, the research community has developed computational approaches to sift through various aspects of clinical data and uncover their insights, among which one of the most challenging problem is to determine the biological mechanisms that cause AD to develop. To study this problem, in this paper we present a novel Joint Multi-Modal Longitudinal Regression and Classification method and show how it can be used to identify the cognitive status of the participants in the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort and the underlying biological mechanisms. By intelligently combining clinical data of various modalities (i.e., genetic information and brain scans) using a variety of regularizations that can identify AD-relevant biomarkers, we perform the regression and classification tasks simultaneously. Because the proposed objective is a non-smooth optimization problem that is difficult to solve in general, we derive an efficient iterative algorithm and rigorously prove its convergence. To validate our new method in predicting the cognitive scores of patients and their clinical diagnosis, we conduct comprehensive experiments on the ADNI cohort. Our promising results demonstrate the benefits and flexibility of the proposed method. We anticipate that our new method is of interest to clinical communities beyond AD research and have open-sourced the code of our method online.11 The code package for the proposed Joint Multi-Modal Longitudinal Regression and Classification model have been made publicly available online at https://github.com/minds-mines/jmmlrc.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1109/TMI.2019.2958943DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7380699PMC
June 2020