TractLearn: a geodesic learning framework for quantitative analysis of brain bundles

ABSTRACTDeep learning-based convolutional neural networks have recently proved their efficiency in providing fast segmentation of major brain fascicles structures, based on diffusion-weighted imaging. The quantitative analysis of brain fascicles then relies on metrics either coming from the tractography process itself or from each voxel along the bundle.Statistical detection of abnormal voxels in the context of disease usually relies on univariate and multivariate statistics models, such as the General Linear Model (GLM). Yet in the case of high-dimensional low sample size data, the GLM often implies high standard deviation range in controls due to anatomical variability, despite the commonly used smoothing process. This can lead to difficulties to detect subtle quantitative alterations from a brain bundle at the voxel scale.Here we introduce TractLearn, a unified framework for brain fascicles quantitative analyses by using geodesic learning as a data-driven learning task. TractLearn allows a mapping between the image high-dimensional domain and the reduced latent space of brain fascicles using a Riemannian approach.We illustrate the robustness of this method on a healthy population with test-retest acquisition of multi-shell diffusion MRI data, demonstrating that it is possible to separately study the global effect due to different MRI sessions from the effect of local bundle alterations. We have then tested the efficiency of our algorithm on a sample of 5 age-matched subjects referred with mild traumatic brain injury.Our contributions are to propose an algorithm based on:1/ A manifold approach to capture controls variability as standard reference instead of an atlas approach based on a Euclidean mean2/ The ability to detect global variation of voxels quantitative values, which means that all the voxels interaction in a structure are considered rather than analyzing each voxel independently.With this regard, TractLearn is a ready-to-use algorithm for precision medicine.KEY POINTWe provide a statistical test taking into account the interaction between voxelsWe propose to use a Riemaniann manifold as reference instead of a Euclidean meanWe demonstrate the usefulness and reliability of the track-weighted contrast

Download Full-text

Quantitative Analysis of Nearest-Neighbors Search in High-Dimensional Sampling-Based Motion Planning

Springer Tracts in Advanced Robotics - Algorithmic Foundation of Robotics VII ◽

10.1007/978-3-540-68405-3_1 ◽

2008 ◽

pp. 3-18 ◽

Cited By ~ 8

Author(s):

Erion Plaku ◽

Lydia E. Kavraki

Keyword(s):

Quantitative Analysis ◽

Motion Planning ◽

Nearest Neighbors ◽

High Dimensional

Download Full-text

Unsupervised Learning for Large Scale Data: The ATHLOS Project

10.1101/2021.04.01.21254751 ◽

2021 ◽

Author(s):

Petros Barmpas ◽

Sotiris Tasoulis ◽

Aristidis G. Vrahatis ◽

Panagiotis Anagnostou ◽

Spiros Georgakopoulos ◽

...

Keyword(s):

Unsupervised Learning ◽

Real World ◽

Large Scale ◽

High Dimensional Data ◽

Experimental Studies ◽

Mixed Data ◽

Categorical Variables ◽

High Dimensional ◽

Data Types ◽

Unified Framework

1AbstractRecent technological advancements in various domains, such as the biomedical and health, offer a plethora of big data for analysis. Part of this data pool is the experimental studies that record various and several features for each instance. It creates datasets having very high dimensionality with mixed data types, with both numerical and categorical variables. On the other hand, unsupervised learning has shown to be able to assist in high-dimensional data, allowing the discovery of unknown patterns through clustering, visualization, dimensionality reduction, and in some cases, their combination. This work highlights unsupervised learning methodologies for large-scale, high-dimensional data, providing the potential of a unified framework that combines the knowledge retrieved from clustering and visualization. The main purpose is to uncover hidden patterns in a high-dimensional mixed dataset, which we achieve through our application in a complex, real-world dataset. The experimental analysis indicates the existence of notable information exposing the usefulness of the utilized methodological framework for similar high-dimensional and mixed, real-world applications.

Download Full-text

Comprehensive quantification of the modified proteome reveals oxidative heart damage in mitochondrial heteroplasmy

10.1101/296392 ◽

2018 ◽

Author(s):

Navratan Bagwan ◽

Elena Bonzon-Kulichenko ◽

Enrique Calvo ◽

Ana Victoria Lechuga-Vieco ◽

Spiros Michalakopoulos ◽

...

Keyword(s):

Mass Spectrometry ◽

Quantitative Analysis ◽

Oxidative Damage ◽

Cardiac Tissue ◽

List Type ◽

Post Translational Modifications ◽

Mitochondrial Heteroplasmy ◽

Peptide Modifications ◽

Oxidative Damage To Proteins ◽

The Impact

SUMMARYPost-translational modifications hugely increase the functional diversity of proteomes. Recent algorithms based on ultratolerant database searching are forging a path to unbiased analysis of peptide modifications by shotgun mass spectrometry. However, these approaches identify only half of the modified forms potentially detectable and do not map the modified residue. Moreover, tools for the quantitative analysis of peptide modifications are currently lacking. Here, we present a suite of algorithms that allow comprehensive identification of detectable modifications, pinpoint the modified residues, and enable their quantitative analysis through an integrated statistical model. These developments were used to characterize the impact of mitochondrial heteroplasmy on the proteome and on the modified peptidome in several tissues from 12-week old mice. Our results reveal that heteroplasmy mainly affects cardiac tissue, inducing oxidative damage to proteins of the oxidative phosphorylation system, and provide a molecular mechanism that explains the structural and functional alterations produced in heart mitochondria.HighlightsIdentifies all protein modifications detectable by mass spectrometryLocates the modified site with 85% accuracyIntegrates quantitative analysis of the proteome and the modified peptidomeReveals that mtDNA heteroplasmy causes oxidative damage in heart OXPHOS proteins

Download Full-text

Nonlinear Regression with High-Dimensional Space Mapping for Blood Component Spectral Quantitative Analysis

Journal of Spectroscopy ◽

10.1155/2018/2689750 ◽

2018 ◽

Vol 2018 ◽

pp. 1-8 ◽

Cited By ~ 2

Author(s):

Xiaoyan Ma ◽

Yanbin Zhang ◽

Hui Cao ◽

Shiliang Zhang ◽

Yan Zhou

Keyword(s):

Quantitative Analysis ◽

Spectral Data ◽

Nonlinear Regression ◽

Dimensional Space ◽

Space Mapping ◽

Partial Least Square ◽

Gaussian Kernel ◽

High Dimensional ◽

Blood Component ◽

High Dimensional Space

Accurate and fast determination of blood component concentration is very essential for the efficient diagnosis of patients. This paper proposes a nonlinear regression method with high-dimensional space mapping for blood component spectral quantitative analysis. Kernels are introduced to map the input data into high-dimensional space for nonlinear regression. As the most famous kernel, Gaussian kernel is usually adopted by researchers. More kernels need to be studied for each kernel describes its own high-dimensional feature space mapping which affects regression performance. In this paper, eight kernels are used to discuss the influence of different space mapping to the blood component spectral quantitative analysis. Each kernel and corresponding parameters are assessed to build the optimal regression model. The proposed method is conducted on a real blood spectral data obtained from the uric acid determination. Results verify that the prediction errors of proposed models are more precise than the ones obtained by linear models. Support vector regression (SVR) provides better performance than partial least square (PLS) when combined with kernels. The local kernels are recommended according to the blood spectral data features. SVR with inverse multiquadric kernel has the best predictive performance that can be used for blood component spectral quantitative analysis.

Download Full-text

Root system influence on high dimensional leaf phenotypes over the grapevine growing season

10.1101/2020.11.10.376947 ◽

2020 ◽

Author(s):

Zachary N. Harris ◽

Laura L. Klein ◽

Mani Awale ◽

Joel F. Swift ◽

Zoë Migicovsky ◽

...

Keyword(s):

Root System ◽

Growing Season ◽

High Dimensional ◽

List Type ◽

Phenological Stage ◽

Multi Scale ◽

Experimental Systems ◽

Shoot System ◽

Scale Data ◽

Broad Understanding

SummaryIn many perennial crops, grafting the root system of one individual to the shoot system of another individual has become an integral part of propagation performed at industrial scales to enhance pest, disease, and stress tolerance and to regulate yield and vigor. Grafted plants offer important experimental systems for understanding the extent and seasonality of root system effects on shoot system biology.Using an experimental vineyard where a common scion ‘Chambourcin’ is growing ungrafted and grafted to three different rootstocks, we explore associations between root system genotype and leaf phenotypes in grafted grapevines across a growing season. We quantified five high-dimensional leaf phenotyping modalities: ionomics, metabolomics, transcriptomics, morphometrics, and physiology and show that rootstock influence is subtle but ubiquitous across modalities.We find strong signatures of rootstock influence on the leaf ionome, with unique signatures detected at each phenological stage. Moreover, all phenotypes and patterns of phenotypic covariation were highly dynamic across the season.These findings expand upon previously identified patterns to suggest that the influence of root system on shoot system phenotypes is complex and broad understanding necessitates volumes of high-dimensional, multi-scale data previously unmet.

Download Full-text

An autoimmune cause of confusion in a patient with a background of hypothyroidism

Endocrinology Diabetes and Metabolism Case Reports ◽

10.1530/edm-19-0014 ◽

2019 ◽

Vol 2019 ◽

Author(s):

Jonathan Brown ◽

Luqman Sardar

Keyword(s):

Thyroid Function ◽

Thyroid Stimulating Hormone ◽

Healthy Population ◽

List Type ◽

Thyroid Function Tests ◽

Thyroid Antibodies ◽

Fluid Analysis ◽

Normal Thyroid Function ◽

Function Tests ◽

Antibody Titres

Summary A 68-year-old previously independent woman presented multiple times to hospital over the course of 3 months with a history of intermittent weakness, vacant episodes, word finding difficulty and reduced cognition. She was initially diagnosed with a TIA, and later with a traumatic subarachnoid haemorrhage following a fall; however, despite resolution of the haemorrhage, symptoms were ongoing and continued to worsen. Confusion screen blood tests showed no cause for the ongoing symptoms. More specialised investigations, such as brain imaging, cerebrospinal fluid analysis, electroencephalogram and serology also gave no clear diagnosis. The patient had a background of hypothyroidism, with plasma thyroid function tests throughout showing normal free thyroxine and a mildly raised thyroid-stimulating hormone (TSH). However plasma anti-thyroid peroxidise (TPO) antibody titres were very high. After discussion with specialists, it was felt she may have a rare and poorly understood condition known as Hashimoto’s encephalopathy (HE). After a trial with steroids, her symptoms dramatically improved and she was able to live independently again, something which would have been impossible at presentation. Learning points: In cases of subacute onset confusion where most other diagnoses have already been excluded, testing for anti-thyroid antibodies can identify patients potentially suffering from HE. In these patients, and under the guidance of specialists, a trial of steroids can dramatically improve patient’s symptoms. The majority of patients are euthyroid at the time of presentation, and so normal thyroid function tests should not prevent anti-thyroid antibodies being tested for. Due to high titres of anti-thyroid antibodies being found in a small percentage of the healthy population, HE should be treated as a diagnosis of exclusion, particularly as treatment with steroids may potentially worsen the outcome in other causes of confusion, such as infection.

Download Full-text

Semantics for interoperability of distributed data and models: Foundations for better-connected information

F1000Research ◽

10.12688/f1000research.11638.1 ◽

2017 ◽

Vol 6 ◽

pp. 686 ◽

Cited By ~ 7

Author(s):

Ferdinando Villa ◽

Stefano Balbi ◽

Ioannis N. Athanasiadis ◽

Caterina Caracciolo

Keyword(s):

Large Scale ◽

List Item ◽

Semantic Annotation ◽

System Modeling ◽

High Standard ◽

List Type ◽

Distributed Data ◽

Software Infrastructure ◽

Conceptual Foundation ◽

Definition Of

Correct and reliable linkage of independently produced information is a requirement to enable sophisticated applications and processing workflows. These can ultimately help address the challenges posed by complex systems (such as socio-ecological systems), whose many components can only be described through independently developed data and model products. We discuss the first outcomes of an investigation in the conceptual and methodological aspects of semantic annotation of data and models, aimed to enable a high standard of interoperability of information. The results, operationalized in the context of a long-term, active, large-scale project on ecosystem services assessment, include: A definition of interoperability based on semantics and scale;A conceptual foundation for the phenomenology underlying scientific observations, aimed to guide the practice of semantic annotation in domain communities;A dedicated language and software infrastructure that operationalizes the findings and allows practitioners to reap the benefits of data and model interoperability. The work presented is the first detailed description of almost a decade of work with communities active in socio-ecological system modeling. After defining the boundaries of possible interoperability based on the understanding of scale, we discuss examples of the practical use of the findings to obtain consistent, interoperable and machine-ready semantic specifications that can integrate semantics across diverse domains and disciplines.

Download Full-text

Fast Markov Blanket Discovery Without Causal Sufficiency

10.48011/asba.v2i1.1663 ◽

2020 ◽

Author(s):

Pedro V. B. Jeronymo ◽

Carlos D. Maciel

Keyword(s):

Causal Structure ◽

Learning Task ◽

High Dimensional ◽

Target Variable ◽

Common Assumption ◽

Markov Blanket ◽

Common Causes ◽

Real World Datasets ◽

Selection Algorithms ◽

Discovery Algorithms

Faster feature selection algorithms become a necessity as Big Data dictates the zeitgeist. An important class of feature selectors are Markov Blanket (MB) learning algorithms. They are Causal Discovery algorithms that learn the local causal structure of a target variable. A common assumption in their theoretical basis, yet often violated in practice, is causal sufficiency: the requirement that all common causes of the measured variables in the dataset are also in the dataset. Recently, Yu et al. (2018) proposed the M3B algorithm, the first to directly learn the MB without demanding causal sufficiency. The main drawback of M3B is that it is time inefficient, being intractable for high-dimensional inputs. In this paper, we derive the Fast Markov Blanket Discovery Algorithm (FMMB). Empirical results that compare FMMB to M3B on the structural learning task show that FMMB outperforms M3B in terms of time efficiency while preserving structural accuracy. Five real-world datasets where used to contrast both algorithms as feature selectors. Applying NB and SVM classifiers, FMMB achieved a competitive outcome. This method mitigates the curse of dimensionality and inspires the development of local-toglobal algorithms.

Download Full-text

Spatial-Structure Siamese Network for Plant Identification

International Journal of Pattern Recognition and Artificial Intelligence ◽

10.1142/s0218001418500350 ◽

2018 ◽

Vol 32 (11) ◽

pp. 1850035 ◽

Cited By ~ 1

Author(s):

Zhi-Yong Gao ◽

Heng-Xing Xie ◽

Ji-Feng Li ◽

Shi-Li Liu

Keyword(s):

Spatial Structure ◽

State Of The Art ◽

Learning Task ◽

Sequential Data ◽

Unified Framework ◽

Plant Identification ◽

Identification Methods ◽

Siamese Network ◽

Large Numbers ◽

Image Pairs

Plant identification is now attracting considerable attention due to its important applications in agriculture automation and ecosystems. Recently, deep learning-based plant identification methods have drawn increasing interest and shown favorable performance. However, existing methods do not consider plant spatial structure and their similarities explicitly. In this paper, we propose a robust spatial-structure siamese network (3SN) for plant identification, which has the following advantages: (1) It models the spatial structure of a plant by recurrent neural networks exploiting their capability to capture long-range dependencies among sequential data, which enables it to capture even a slight difference between a specific plant and distractors. (2) The plant similarity modeling is achieved effectively by a siamese network with large numbers of image pairs. In this way, the plant classification task and siamese learning task are learned jointly in a unified framework, where both can enhance and complement each other. Extensive experimental results show that the proposed 3SN method outperforms the state-of-the-art methods consistently.

Download Full-text