scholarly journals Three-dimensional deep learning with spatial erasing for unsupervised anomaly segmentation in brain MRI

2021 ◽  
Author(s):  
Marcel Bengs ◽  
Finn Behrendt ◽  
Julia Krüger ◽  
Roland Opfer ◽  
Alexander Schlaefer
Keyword(s):  
Deep Learning ◽  
Brain Mri ◽  
Magnetic Resonance Images ◽  
Spatial Context ◽  
Data Sets ◽  
Learning Methods ◽  
Data Set ◽  
Performance Improvements ◽  
Wide Range ◽  
The Impact

Abstract Purpose Brain Magnetic Resonance Images (MRIs) are essential for the diagnosis of neurological diseases. Recently, deep learning methods for unsupervised anomaly detection (UAD) have been proposed for the analysis of brain MRI. These methods rely on healthy brain MRIs and eliminate the requirement of pixel-wise annotated data compared to supervised deep learning. While a wide range of methods for UAD have been proposed, these methods are mostly 2D and only learn from MRI slices, disregarding that brain lesions are inherently 3D and the spatial context of MRI volumes remains unexploited. Methods We investigate whether using increased spatial context by using MRI volumes combined with spatial erasing leads to improved unsupervised anomaly segmentation performance compared to learning from slices. We evaluate and compare 2D variational autoencoder (VAE) to their 3D counterpart, propose 3D input erasing, and systemically study the impact of the data set size on the performance. Results Using two publicly available segmentation data sets for evaluation, 3D VAEs outperform their 2D counterpart, highlighting the advantage of volumetric context. Also, our 3D erasing methods allow for further performance improvements. Our best performing 3D VAE with input erasing leads to an average DICE score of 31.40% compared to 25.76% for the 2D VAE. Conclusions We propose 3D deep learning methods for UAD in brain MRI combined with 3D erasing and demonstrate that 3D methods clearly outperform their 2D counterpart for anomaly segmentation. Also, our spatial erasing method allows for further performance improvements and reduces the requirement for large data sets.

scholarly journals Investigating the impact of pre-processing techniques and pre-trained word embeddings in detecting Arabic health information on social media

2021 ◽  
Vol 8 (1) ◽  
Author(s):  
Yahya Albalawi ◽  
Jim Buckley ◽  
Nikola S. Nikolov
Keyword(s):  
Social Media ◽  
Deep Learning ◽  
Comprehensive Evaluation ◽  
Classification Problem ◽  
Data Sets ◽  
Word Embeddings ◽  
Data Set ◽  
Lower Accuracy ◽  
Health Related ◽  
The Impact

AbstractThis paper presents a comprehensive evaluation of data pre-processing and word embedding techniques in the context of Arabic document classification in the domain of health-related communication on social media. We evaluate 26 text pre-processings applied to Arabic tweets within the process of training a classifier to identify health-related tweets. For this task we use the (traditional) machine learning classifiers KNN, SVM, Multinomial NB and Logistic Regression. Furthermore, we report experimental results with the deep learning architectures BLSTM and CNN for the same text classification problem. Since word embeddings are more typically used as the input layer in deep networks, in the deep learning experiments we evaluate several state-of-the-art pre-trained word embeddings with the same text pre-processing applied. To achieve these goals, we use two data sets: one for both training and testing, and another for testing the generality of our models only. Our results point to the conclusion that only four out of the 26 pre-processings improve the classification accuracy significantly. For the first data set of Arabic tweets, we found that Mazajak CBOW pre-trained word embeddings as the input to a BLSTM deep network led to the most accurate classifier with F1 score of 89.7%. For the second data set, Mazajak Skip-Gram pre-trained word embeddings as the input to BLSTM led to the most accurate model with F1 score of 75.2% and accuracy of 90.7% compared to F1 score of 90.8% achieved by Mazajak CBOW for the same architecture but with lower accuracy of 70.89%. Our results also show that the performance of the best of the traditional classifier we trained is comparable to the deep learning methods on the first dataset, but significantly worse on the second dataset.

scholarly journals Training confounder-free deep learning models for medical applications

2020 ◽  
Vol 11 (1) ◽  
Author(s):  
Qingyu Zhao ◽  
Ehsan Adeli ◽  
Kilian M. Pohl
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Bone Age ◽  
Magnetic Resonance Images ◽  
Data Sets ◽  
Large Set ◽  
Learning Models ◽  
End To End ◽  
The Impact ◽  
The Relationship

AbstractThe presence of confounding effects (or biases) is one of the most critical challenges in using deep learning to advance discovery in medical imaging studies. Confounders affect the relationship between input data (e.g., brain MRIs) and output variables (e.g., diagnosis). Improper modeling of those relationships often results in spurious and biased associations. Traditional machine learning and statistical models minimize the impact of confounders by, for example, matching data sets, stratifying data, or residualizing imaging measurements. Alternative strategies are needed for state-of-the-art deep learning models that use end-to-end training to automatically extract informative features from large set of images. In this article, we introduce an end-to-end approach for deriving features invariant to confounding factors while accounting for intrinsic correlations between the confounder(s) and prediction outcome. The method does so by exploiting concepts from traditional statistical methods and recent fair machine learning schemes. We evaluate the method on predicting the diagnosis of HIV solely from Magnetic Resonance Images (MRIs), identifying morphological sex differences in adolescence from those of the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA), and determining the bone age from X-ray images of children. The results show that our method can accurately predict while reducing biases associated with confounders. The code is available at https://github.com/qingyuzhao/br-net.

scholarly journals The Impact of Normalization Methods on RNA-Seq Data Analysis

2015 ◽  
Vol 2015 ◽  
pp. 1-10 ◽  
Author(s):  
J. Zyprych-Walczak ◽  
A. Szabelska ◽  
L. Handschuh ◽  
K. Górczak ◽  
K. Klamecka ◽  
...  
Keyword(s):  
High Throughput Sequencing ◽  
Data Sets ◽  
Complex Data ◽  
Rna Seq ◽  
Medical Problems ◽  
Data Set ◽  
Normalization Methods ◽  
Wide Range ◽  
The Impact ◽  
Selection Of

High-throughput sequencing technologies, such as the Illumina Hi-seq, are powerful new tools for investigating a wide range of biological and medical problems. Massive and complex data sets produced by the sequencers create a need for development of statistical and computational methods that can tackle the analysis and management of data. The data normalization is one of the most crucial steps of data processing and this process must be carefully considered as it has a profound effect on the results of the analysis. In this work, we focus on a comprehensive comparison of five normalization methods related to sequencing depth, widely used for transcriptome sequencing (RNA-seq) data, and their impact on the results of gene expression analysis. Based on this study, we suggest a universal workflow that can be applied for the selection of the optimal normalization procedure for any particular data set. The described workflow includes calculation of the bias and variance values for the control genes, sensitivity and specificity of the methods, and classification errors as well as generation of the diagnostic plots. Combining the above information facilitates the selection of the most appropriate normalization method for the studied data sets and determines which methods can be used interchangeably.

scholarly journals SEMANTIC3D.NET: A NEW LARGE-SCALE POINT CLOUD CLASSIFICATION BENCHMARK

2017 ◽  
Vol IV-1/W1 ◽  
pp. 91-98 ◽  
Author(s):  
T. Hackel ◽  
N. Savinov ◽  
L. Ladicky ◽  
J. D. Wegner ◽  
K. Schindler ◽  
...  
Keyword(s):  
Deep Learning ◽  
Point Cloud ◽  
Point Clouds ◽  
Full Potential ◽  
3D Point Cloud ◽  
Learning Methods ◽  
Data Set ◽  
Cloud Classification ◽  
Wide Range ◽  
Point Cloud Classification

This paper presents a new 3D point cloud classification benchmark data set with over four billion manually labelled points, meant as input for data-hungry (deep) learning methods. We also discuss first submissions to the benchmark that use deep convolutional neural networks (CNNs) as a work horse, which already show remarkable performance improvements over state-of-the-art. CNNs have become the de-facto standard for many tasks in computer vision and machine learning like semantic segmentation or object detection in images, but have no yet led to a true breakthrough for 3D point cloud labelling tasks due to lack of training data. With the massive data set presented in this paper, we aim at closing this data gap to help unleash the full potential of deep learning methods for 3D labelling tasks. Our semantic3D.net data set consists of dense point clouds acquired with static terrestrial laser scanners. It contains 8 semantic classes and covers a wide range of urban outdoor scenes: churches, streets, railroad tracks, squares, villages, soccer fields and castles. We describe our labelling interface and show that our data set provides more dense and complete point clouds with much higher overall number of labelled points compared to those already available to the research community. We further provide baseline method descriptions and comparison between methods submitted to our online system. We hope semantic3D.net will pave the way for deep learning methods in 3D point cloud labelling to learn richer, more general 3D representations, and first submissions after only a few months indicate that this might indeed be the case.

Human Activity Recognition using Fourier Transform Inspired Deep Learning Combination Model

2019 ◽  
Vol 9 (1) ◽  
pp. 16-31
Author(s):  
Kyungkoo Jun
Keyword(s):  
Fourier Transform ◽  
Deep Learning ◽  
Short Term Memory ◽  
Window Size ◽  
Sensor Data ◽  
Data Sets ◽  
Data Set ◽  
Proposed Model ◽  
Testing Data ◽  
Labeling Scheme

Background & Objective: This paper proposes a Fourier transform inspired method to classify human activities from time series sensor data. Methods: Our method begins by decomposing 1D input signal into 2D patterns, which is motivated by the Fourier conversion. The decomposition is helped by Long Short-Term Memory (LSTM) which captures the temporal dependency from the signal and then produces encoded sequences. The sequences, once arranged into the 2D array, can represent the fingerprints of the signals. The benefit of such transformation is that we can exploit the recent advances of the deep learning models for the image classification such as Convolutional Neural Network (CNN). Results: The proposed model, as a result, is the combination of LSTM and CNN. We evaluate the model over two data sets. For the first data set, which is more standardized than the other, our model outperforms previous works or at least equal. In the case of the second data set, we devise the schemes to generate training and testing data by changing the parameters of the window size, the sliding size, and the labeling scheme. Conclusion: The evaluation results show that the accuracy is over 95% for some cases. We also analyze the effect of the parameters on the performance.

scholarly journals Data augmentation for computed tomography angiography via synthetic image generation and neural domain adaptation

2020 ◽  
Vol 6 (1) ◽  
Author(s):  
Malte Seemann ◽  
Lennart Bargsten ◽  
Alexander Schlaefer
Keyword(s):  
Computed Tomography ◽  
Neural Networks ◽  
Deep Learning ◽  
Medical Imaging ◽  
Computed Tomography Angiography ◽  
Data Augmentation ◽  
Domain Adaptation ◽  
Synthetic Image ◽  
Wide Range ◽  
The Impact

AbstractDeep learning methods produce promising results when applied to a wide range of medical imaging tasks, including segmentation of artery lumen in computed tomography angiography (CTA) data. However, to perform sufficiently, neural networks have to be trained on large amounts of high quality annotated data. In the realm of medical imaging, annotations are not only quite scarce but also often not entirely reliable. To tackle both challenges, we developed a two-step approach for generating realistic synthetic CTA data for the purpose of data augmentation. In the first step moderately realistic images are generated in a purely numerical fashion. In the second step these images are improved by applying neural domain adaptation. We evaluated the impact of synthetic data on lumen segmentation via convolutional neural networks (CNNs) by comparing resulting performances. Improvements of up to 5% in terms of Dice coefficient and 20% for Hausdorff distance represent a proof of concept that the proposed augmentation procedure can be used to enhance deep learning-based segmentation for artery lumen in CTA images.

scholarly journals A Review of Computer-Aided Expert Systems for Breast Cancer Diagnosis

Cancers ◽  
2021 ◽  
Vol 13 (11) ◽  
pp. 2764
Author(s):  
Xin Yu Liew ◽  
Nazia Hameed ◽  
Jeremie Clos
Keyword(s):  
Breast Cancer ◽  
Machine Learning ◽  
Deep Learning ◽  
Main Process ◽  
Learning Approaches ◽  
Learning Methods ◽  
Advantages And Disadvantages ◽  
Computer Aided ◽  
Conventional Methods ◽  
The Impact

A computer-aided diagnosis (CAD) expert system is a powerful tool to efficiently assist a pathologist in achieving an early diagnosis of breast cancer. This process identifies the presence of cancer in breast tissue samples and the distinct type of cancer stages. In a standard CAD system, the main process involves image pre-processing, segmentation, feature extraction, feature selection, classification, and performance evaluation. In this review paper, we reviewed the existing state-of-the-art machine learning approaches applied at each stage involving conventional methods and deep learning methods, the comparisons within methods, and we provide technical details with advantages and disadvantages. The aims are to investigate the impact of CAD systems using histopathology images, investigate deep learning methods that outperform conventional methods, and provide a summary for future researchers to analyse and improve the existing techniques used. Lastly, we will discuss the research gaps of existing machine learning approaches for implementation and propose future direction guidelines for upcoming researchers.

scholarly journals A Visual and VAE Based Hierarchical Indoor Localization Method

Sensors ◽  
2021 ◽  
Vol 21 (10) ◽  
pp. 3406
Author(s):  
Jie Jiang ◽  
Yin Zou ◽  
Lidong Chen ◽  
Yujie Fang
Keyword(s):  
Image Retrieval ◽  
Indoor Localization ◽  
Data Sets ◽  
Indoor Environments ◽  
Global Features ◽  
Data Set ◽  
Data Annotation ◽  
Wide Range ◽  
Annotation Costs ◽  
Global And Local

Precise localization and pose estimation in indoor environments are commonly employed in a wide range of applications, including robotics, augmented reality, and navigation and positioning services. Such applications can be solved via visual-based localization using a pre-built 3D model. The increase in searching space associated with large scenes can be overcome by retrieving images in advance and subsequently estimating the pose. The majority of current deep learning-based image retrieval methods require labeled data, which increase data annotation costs and complicate the acquisition of data. In this paper, we propose an unsupervised hierarchical indoor localization framework that integrates an unsupervised network variational autoencoder (VAE) with a visual-based Structure-from-Motion (SfM) approach in order to extract global and local features. During the localization process, global features are applied for the image retrieval at the level of the scene map in order to obtain candidate images, and are subsequently used to estimate the pose from 2D-3D matches between query and candidate images. RGB images only are used as the input of the proposed localization system, which is both convenient and challenging. Experimental results reveal that the proposed method can localize images within 0.16 m and 4° in the 7-Scenes data sets and 32.8% within 5 m and 20° in the Baidu data set. Furthermore, our proposed method achieves a higher precision compared to advanced methods.

scholarly journals The Midlatitude Continental Convective Clouds Experiment (MC3E) sounding network: operations, processing and analysis

2015 ◽  
Vol 8 (1) ◽  
pp. 421-434 ◽  
Author(s):  
M. P. Jensen ◽  
T. Toto ◽  
D. Troyan ◽  
P. E. Ciesielski ◽  
D. Holdridge ◽  
...  
Keyword(s):  
Large Scale ◽  
Scale Model ◽  
Data Sets ◽  
Central Plains ◽  
Data Set ◽  
Convective Systems ◽  
Convective Clouds ◽  
Quality Checks ◽  
Network Operations ◽  
The Impact

Abstract. The Midlatitude Continental Convective Clouds Experiment (MC3E) took place during the spring of 2011 centered in north-central Oklahoma, USA. The main goal of this field campaign was to capture the dynamical and microphysical characteristics of precipitating convective systems in the US Central Plains. A major component of the campaign was a six-site radiosonde array designed to capture the large-scale variability of the atmospheric state with the intent of deriving model forcing data sets. Over the course of the 46-day MC3E campaign, a total of 1362 radiosondes were launched from the enhanced sonde network. This manuscript provides details on the instrumentation used as part of the sounding array, the data processing activities including quality checks and humidity bias corrections and an analysis of the impacts of bias correction and algorithm assumptions on the determination of convective levels and indices. It is found that corrections for known radiosonde humidity biases and assumptions regarding the characteristics of the surface convective parcel result in significant differences in the derived values of convective levels and indices in many soundings. In addition, the impact of including the humidity corrections and quality controls on the thermodynamic profiles that are used in the derivation of a large-scale model forcing data set are investigated. The results show a significant impact on the derived large-scale vertical velocity field illustrating the importance of addressing these humidity biases.

scholarly journals Considerations of the Scale of Radiocarbon Offsets in the East Mediterranean, and Considering a Case for the Latest (Most Recent) Likely Date for the Santorini Eruption

Radiocarbon ◽  
2012 ◽  
Vol 54 (3-4) ◽  
pp. 449-474 ◽  
Author(s):  
Sturt W Manning ◽  
Bernd Kromer
Keyword(s):  
Time Horizon ◽  
Weighted Average ◽  
17Th Century ◽  
Data Sets ◽  
East Mediterranean ◽  
Average Value ◽  
Accuracy And Precision ◽  
Wide Range ◽  
The Impact ◽  
East Mediterranean Region

The debate over the dating of the Santorini (Thera) volcanic eruption has seen sustained efforts to criticize or challenge the radiocarbon dating of this time horizon. We consider some of the relevant areas of possible movement in the14C dating—and, in particular, any plausible mechanisms to support as late (most recent) a date as possible. First, we report and analyze data investigating the scale of apparent possible14C offsets (growing season related) in the Aegean-Anatolia-east Mediterranean region (excluding the southern Levant and especially pre-modern, pre-dam Egypt, which is a distinct case), and find no evidence for more than very small possible offsets from several cases. This topic is thus not an explanation for current differences in dating in the Aegean and at best provides only a few years of latitude. Second, we consider some aspects of the accuracy and precision of14C dating with respect to the Santorini case. While the existing data appear robust, we nonetheless speculate that examination of the frequency distribution of the14C data on short-lived samples from the volcanic destruction level at Akrotiri on Santorini (Thera) may indicate that the average value of the overall data sets is not necessarily the most appropriate14C age to use for dating this time horizon. We note the recent paper of Soter (2011), which suggests that in such a volcanic context some (small) age increment may be possible from diffuse CO2emissions (the effect is hypothetical at this stage and hasnotbeen observed in the field), and that "if short-lived samples from the same stratigraphic horizon yield a wide range of14C ages, the lower values may be the least altered by old CO2." In this context, it might be argued that a substantive “low” grouping of14C ages observable within the overall14C data sets on short-lived samples from the Thera volcanic destruction level centered about 3326–3328 BP is perhaps more representative of the contemporary atmospheric14C age (without any volcanic CO2contamination). This is a subjective argument (since, in statistical terms, the existing studies using the weighted average remain valid) that looks to support as late a date as reasonable from the14C data. The impact of employing this revised14C age is discussed. In general, a late 17th century BC date range is found (to remain) to be most likelyeven ifsuch a late-dating strategy is followed—a late 17th century BC date range is thus a robust finding from the14C evidence even allowing for various possible variation factors. However, the possibility of a mid-16th century BC date (within ∼1593–1530 cal BC) is increased when compared against previous analyses if the Santorini data are considered in isolation.

Sign in / Sign up

Export Citation Format

Share Document