RESUMO
The outbreak of the corona virus disease (COVID-19) has changed the lives of most people on Earth. Given the high prevalence of this disease, its correct diagnosis in order to quarantine patients is of the utmost importance in the steps of fighting this pandemic. Among the various modalities used for diagnosis, medical imaging, especially computed tomography (CT) imaging, has been the focus of many previous studies due to its accuracy and availability. In addition, automation of diagnostic methods can be of great help to physicians. In this paper, a method based on pre-trained deep neural networks is presented, which, by taking advantage of a cyclic generative adversarial net (CycleGAN) model for data augmentation, has reached state-of-the-art performance for the task at hand, i.e., 99.60% accuracy. Also, in order to evaluate the method, a dataset containing 3163 images from 189 patients has been collected and labeled by physicians. Unlike prior datasets, normal data have been collected from people suspected of having COVID-19 disease and not from data from other diseases, and this database is made available publicly. Moreover, the method's reliability is further evaluated by calibration metrics, and its decision is interpreted by Grad-CAM also to find suspicious regions as another output of the method and make its decisions trustworthy and explainable.
RESUMO
BACKGROUND: Deep learning techniques have been successfully applied to bioimaging problems; however, these methods are highly data demanding. An approach to deal with the lack of data and avoid overfitting is the application of data augmentation, a technique that generates new training samples from the original dataset by applying different kinds of transformations. Several tools exist to apply data augmentation in the context of image classification, but it does not exist a similar tool for the problems of localization, detection, semantic segmentation or instance segmentation that works not only with 2 dimensional images but also with multi-dimensional images (such as stacks or videos). RESULTS: In this paper, we present a generic strategy that can be applied to automatically augment a dataset of images, or multi-dimensional images, devoted to classification, localization, detection, semantic segmentation or instance segmentation. The augmentation method presented in this paper has been implemented in the open-source package CLoDSA. To prove the benefits of using CLoDSA, we have employed this library to improve the accuracy of models for Malaria parasite classification, stomata detection, and automatic segmentation of neural structures. CONCLUSIONS: CLoDSA is the first, at least up to the best of our knowledge, image augmentation library for object classification, localization, detection, semantic segmentation, and instance segmentation that works not only with 2 dimensional images but also with multi-dimensional images.
Assuntos
Algoritmos , Processamento de Imagem Assistida por Computador , Semântica , Animais , Aprendizado Profundo , Humanos , Malária/parasitologia , Modelos Teóricos , Parasitos/classificaçãoRESUMO
BACKGROUND: Fungi have diverse biotechnological applications in, among others, agriculture, bioenergy generation, or remediation of polluted soil and water. In this context, culture media based on color change in response to degradation of dyes are particularly relevant; but measuring dye decolorisation of fungal strains mainly relies on a visual and semiquantitative classification of color intensity changes. Such a classification is a subjective, time-consuming and difficult to reproduce process. RESULTS: DecoFungi is the first, at least up to the best of our knowledge, application to automatically characterise dye decolorisation level of fungal strains from images of inoculated plates. In order to deal with this task, DecoFungi employs a deep-learning model, accessible through a user-friendly web interface, with an accuracy of 96.5%. CONCLUSIONS: DecoFungi is an easy to use system for characterising dye decolorisation level of fungal strains from images of inoculated plates.
Assuntos
Corantes/química , Fungos/metabolismo , Internet , Software , Modelos Teóricos , Interface Usuário-ComputadorRESUMO
DNA fingerprinting is a genetic typing technique that allows the analysis of the genomic relatedness between samples, and the comparison of DNA patterns. The analysis of DNA gel fingerprint images usually consists of five consecutive steps: image pre-processing, lane segmentation, band detection, normalization and fingerprint comparison. In this article, we firstly survey the main methods that have been applied in the literature in each of these stages. Secondly, we focus on lane-segmentation and band-detection algorithms-as they are the steps that usually require user-intervention-and detect the seven core algorithms used for both tasks. Subsequently, we present a benchmark that includes a data set of images, the gold standards associated with those images and the tools to measure the performance of lane-segmentation and band-detection algorithms. Finally, we implement the core algorithms used both for lane segmentation and band detection, and evaluate their performance using our benchmark. As a conclusion of that study, we obtain that the average profile algorithm is the best starting point for lane segmentation and band detection.
Assuntos
DNA/genética , Algoritmos , Benchmarking , Processamento de Imagem Assistida por ComputadorRESUMO
DNA fingerprinting is a genetic typing technique that allows the analysis of the genomic relatedness between samples, and the comparison of DNA patterns. This technique has multiple applications in different fields (medical diagnosis, forensic science, parentage testing, food industry, agriculture and many others). An important task in molecular epidemiology of infectious diseases is the analysis and comparison of pulsed-field gel electrophoresis (PFGE) patterns. This is applied to determine the clonal diversity of bacteria in the follow-up of outbreaks or for tracking specific clones of special relevance. The resulting images produced by DNA fingerprinting are sometimes difficult to interpret, and multiple tools have been developed to simplify this task. In this article, we present a survey of tools for analysing DNA fingerprints. In particular, we compare 33 tools using a set of predefined criteria. The comparison was carried out by hands-on experiences-whenever possible-and inspecting the documentation of the tools. As no system is preferred in all the possible scenarios, we have created a spreadsheet that can be customized by researchers to determine the best system for their needs.
Assuntos
Impressões Digitais de DNA , DNA Bacteriano , Eletroforese em Gel de Campo PulsadoRESUMO
BACKGROUND: DNA fingerprinting is a technique for comparing DNA patterns that has applications in a wide variety of contexts. Several commercial and freely-available tools can be used to analyze DNA fingerprint gel images; however, commercial tools are expensive and usually difficult to use; and, free tools support the basic functionality for DNA fingerprint analysis, but lack some instrumental features to obtain accurate results. RESULTS: In this paper, we present GelJ, a feather-weight, user-friendly, platform-independent, open-source and free tool for analyzing DNA fingerprint gel images. Some of the outstanding features of GelJ are mechanisms for accurate lane- and band-detection, several options for computing migration models, a number of band- and curve-based similarity methods, different techniques for generating dendrograms, comparison of banding patterns from different experiments, and database support. CONCLUSIONS: GelJ is an easy to use tool for analyzing DNA fingerprint gel images. It combines the best characteristics of both free and commercial tools: GelJ is light and simple to use (as free programs), but it also includes the necessary features to obtain precise results (as commercial programs). In addition, GelJ incorporates new functionality that is not supported by any other tool.
Assuntos
Impressões Digitais de DNA/métodos , Software , Análise por Conglomerados , DNA/análise , Eletroforese em Gel de Campo Pulsado , InternetRESUMO
Introduction: Drought is one of the biggest problems for crop production and also affects the survival and persistence of soil rhizobia, which limits the establishment of efficient symbiosis and endangers the productivity of legumes, the main source of plant protein worldwide. Aim: Since the biodiversity can be altered by several factors including abiotic stresses or cultural practices, the objective of this research was to evaluate the effect of water availability, plant genotype and agricultural management on the presence, nodulation capacity and genotypic diversity of rhizobia. Method: A field experiment was conducted with twelve common bean genotypes under irrigation and rain-fed conditions, both in conventional and organic management. Estimation of the number of viable rhizobia present in soils was performed before the crop establishment, whereas the crop yield, nodule number and the strain diversity of bacteria present in nodules were determined at postharvest. Results: Rainfed conditions reduced the number of nodules and of isolated bacteria and their genetic diversity, although to a lesser extent than the agrochemical inputs related to conventional management. In addition, the effect of water scarcity on the conventional management soil was greater than observed under organic conditions. Conclusions: The preservation of diversity will be a key factor to maintain crop production in the future, as problems caused by drought will be exacerbated by climate change and organic management can help to maintain the biodiversity of soil microbiota, a fundamental aspect for soil health and quality.
RESUMO
The early detection of glaucoma is essential in preventing visual impairment. Artificial intelligence (AI) can be used to analyze color fundus photographs (CFPs) in a cost-effective manner, making glaucoma screening more accessible. While AI models for glaucoma screening from CFPs have shown promising results in laboratory settings, their performance decreases significantly in real-world scenarios due to the presence of out-of-distribution and low-quality images. To address this issue, we propose the Artificial Intelligence for Robust Glaucoma Screening (AIROGS) challenge. This challenge includes a large dataset of around 113,000 images from about 60,000 patients and 500 different screening centers, and encourages the development of algorithms that are robust to ungradable and unexpected input data. We evaluated solutions from 14 teams in this paper and found that the best teams performed similarly to a set of 20 expert ophthalmologists and optometrists. The highest-scoring team achieved an area under the receiver operating characteristic curve of 0.99 (95% CI: 0.98-0.99) for detecting ungradable images on-the-fly. Additionally, many of the algorithms showed robust performance when tested on three other publicly available datasets. These results demonstrate the feasibility of robust AI-enabled glaucoma screening.
Assuntos
Inteligência Artificial , Glaucoma , Humanos , Glaucoma/diagnóstico por imagem , Fundo de Olho , Técnicas de Diagnóstico Oftalmológico , AlgoritmosRESUMO
BACKGROUND AND OBJECTIVE: Age-related macular degeneration (AMD) is an eye disease that happens when ageing causes damage to the macula, and it is the leading cause of blindness in developed countries. Screening retinal fundus images allows ophthalmologists to early detect, diagnose and treat this disease; however, the manual interpretation of images is a time-consuming task. In this paper, we aim to study different deep learning methods to diagnose AMD. METHODS: We have conducted a thorough study of two families of deep learning models based on convolutional neural networks (CNN) and transformer architectures to automatically diagnose referable/non-referable AMD, and grade AMD severity scales (no AMD, early AMD, intermediate AMD, and advanced AMD). In addition, we have analysed several progressive resizing strategies and ensemble methods for convolutional-based architectures to further improve the performance of the models. RESULTS: As a first result, we have shown that transformer-based architectures obtain considerably worse results than convolutional-based architectures for diagnosing AMD. Moreover, we have built a model for diagnosing referable AMD that yielded a mean F1-score (SD) of 92.60% (0.47), a mean AUROC (SD) of 97.53% (0.40), and a mean weighted kappa coefficient (SD) of 85.28% (0.91); and an ensemble of models for grading AMD severity scales with a mean accuracy (SD) of 82.55% (2.92), and a mean weighted kappa coefficient (SD) of 84.76% (2.45). CONCLUSIONS: This work shows that working with convolutional based architectures is more suitable than using transformer based models for classifying and grading AMD from retinal fundus images. Furthermore, convolutional models can be improved by means of progressive resizing strategies and ensemble methods.
Assuntos
Macula Lutea , Degeneração Macular , Humanos , Reprodutibilidade dos Testes , Degeneração Macular/diagnóstico por imagem , Redes Neurais de Computação , Fundo de OlhoRESUMO
Mechanical interactions between cells and their microenvironment play an important role in determining cell fate, which is particularly relevant in metastasis, a process where cells invade tissue matrices with different mechanical properties. In vitro, type I collagen hydrogels have been commonly used for modeling the microenvironment due to its ubiquity in the human body. In this work, the combined influence of the stiffness of these hydrogels and their ultrastructure on the migration patterns of HCT-116 and HT-29 spheroids are analyzed. For this, six different types of pure type I collagen hydrogels by changing the collagen concentration and the gelation temperature are prepared. The stiffness of each sample is measured and its ultrastructure is characterized. Cell migration studies are then performed by seeding the spheroids in three different spatial conditions. It is shown that changes in the aforementioned parameters lead to differences in the mechanical stiffness of the matrices as well as the ultrastructure. These differences, in turn, lead to distinct cell migration patterns of HCT-116 and HT-29 spheroids in either of the spatial conditions tested. Based on these results, it is concluded that the stiffness and the ultrastructural organization of the matrix can actively modulate cell migration behavior in colorectal cancer spheroids.
RESUMO
The accumulation of tau fibrils is associated with neurodegenerative diseases, which are collectively termed tauopathies. Cryo-EM studies have shown that the packed fibril core of tau adopts distinct structures in different tauopathies, such as Alzheimer's disease, corticobasal degeneration, and progressive supranuclear palsy. A subset of tauopathies are linked to missense mutations in the tau protein, but it is not clear whether these mutations impact the structure of tau fibrils. To answer this question, we developed a high-throughput protein purification platform and purified a panel of 37 tau variants using the full-length 0N4R splice isoform. Each of these variants was used to create fibrils in vitro, and their relative structures were studied using a high-throughput protease sensitivity platform. We find that a subset of the disease-associated mutations form fibrils that resemble wild-type tau, while others are strikingly different. The impact of mutations on tau structure was not clearly associated with either the location of the mutation or the relative kinetics of fibril assembly, suggesting that tau mutations alter the packed core structures through a complex molecular mechanism. Together, these studies show that single-point mutations can impact the assembly of tau into fibrils, providing insight into its association with pathology and disease.
Assuntos
Doença de Alzheimer , Tauopatias , Humanos , Proteínas tau/metabolismo , Tauopatias/metabolismo , Doença de Alzheimer/metabolismo , Mutação/genéticaRESUMO
This paper introduces the Human Action Multi-Modal Monitoring in Manufacturing (HA4M) dataset, a collection of multi-modal data relative to actions performed by different subjects building an Epicyclic Gear Train (EGT). In particular, 41 subjects executed several trials of the assembly task, which consists of 12 actions. Data were collected in a laboratory scenario using a Microsoft® Azure Kinect which integrates a depth camera, an RGB camera, and InfraRed (IR) emitters. To the best of authors' knowledge, the HA4M dataset is the first multi-modal dataset about an assembly task containing six types of data: RGB images, Depth maps, IR images, RGB-to-Depth-Aligned images, Point Clouds and Skeleton data. These data represent a good foundation to develop and test advanced action recognition systems in several fields, including Computer Vision and Machine Learning, and application domains such as smart manufacturing and human-robot collaboration.
RESUMO
Simplicial-map neural networks are a recent neural network architecture induced by simplicial maps defined between simplicial complexes. It has been proved that simplicial-map neural networks are universal approximators and that they can be refined to be robust to adversarial attacks. In this paper, the refinement toward robustness is optimized by reducing the number of simplices (i.e., nodes) needed. We have shown experimentally that such a refined neural network is equivalent to the original network as a classification tool but requires much less storage.
RESUMO
BACKGROUND AND OBJECTIVES: Infectious diseases produced by antimicrobial resistant microorganisms are a major threat to human, and animal health worldwide. This problem is increased by the virulence and spread of these bacteria. Surface motility has been regarded as a pathogenicity element because it is essential for many biological functions, but also for disease spreading; hence, investigations on the motility behaviour of bacteria are crucial to understand chemotaxis, biofilm formation and virulence in general. To identify a motile strain in the laboratory, the bacterial spread area is observed on media solidified with agar. Up to now, the task of measuring bacteria spread was a manual, and, therefore, tedious and time-consuming task. The aim of this work is the development of a set of tools for bacteria segmentation in motility images. METHODS: In this work, we address the problem of measuring bacteria spread on motility images by creating an automatic pipeline based on deep learning models. Such a pipeline consists of a classification model to determine whether the bacteria has spread to cover completely the Petri dish, and a segmentation model to determine the spread of those bacteria that do not fully cover the Petri dishes. In order to annotate enough images to train our deep learning models, a semi-automatic annotation procedure is presented. RESULTS: The classification model of our pipeline achieved a F1-score of 99.85%, and the segmentation model achieved a Dice coefficient of 95.66%. In addition, the segmentation model produces results that are indistinguishable, and in many cases preferred, from those produced manually by experts. Finally, we facilitate the dissemination of our pipeline with the development of MotilityJ, an open-source and user-friendly application for measuring bacteria spread on motility images. CONCLUSIONS: In this work, we have developed an algorithm and trained several models for measuring bacteria spread on motility images. Thanks to this work, the analysis of motility images will be faster and more reliable. The developed tools will help to advance our understanding of the behaviour and virulence of bacteria.
Assuntos
Bactérias , Fenômenos Fisiológicos Bacterianos , Transmissão de Doença Infecciosa , HumanosRESUMO
BACKGROUND AND OBJECTIVES: Spheroids are the most widely used 3D models for studying the effects of different micro-environmental characteristics on tumour behaviour, and for testing different preclinical and clinical treatments. In order to speed up the study of spheroids, imaging methods that automatically segment and measure spheroids are instrumental; and, several approaches for automatic segmentation of spheroid images exist in the literature. However, those methods fail to generalise to a diversity of experimental conditions. The aim of this work is the development of a set of tools for spheroid segmentation that works in a diversity of settings. METHODS: In this work, we have tackled the spheroid segmentation task by first developing a generic segmentation algorithm that can be easily adapted to different scenarios. This generic algorithm has been employed to reduce the burden of annotating a dataset of images that, in turn, has been employed to train several deep learning architectures for semantic segmentation. Both our generic algorithm and the constructed deep learning models have been tested with several datasets of spheroid images where the spheroids were grown under several experimental conditions, and the images acquired using different equipment. RESULTS: The developed generic algorithm can be particularised to different scenarios; however, those particular algorithms fail to generalise to different conditions. By contrast, the best deep learning model, constructed using the HRNet-Seg architecture, generalises properly to a diversity of scenarios. In order to facilitate the dissemination and use of our algorithms and models, we present SpheroidJ, a set of open-source tools for spheroid segmentation. CONCLUSIONS: In this work, we have developed an algorithm and trained several models for spheroid segmentation that can be employed with images acquired under different conditions. Thanks to this work, the analysis of spheroids acquired under different conditions will be more reliable and comparable; and, the developed tools will help to advance our understanding of tumour behaviour.
Assuntos
Algoritmos , SemânticaRESUMO
Schizophrenia (SZ) is a mental disorder whereby due to the secretion of specific chemicals in the brain, the function of some brain regions is out of balance, leading to the lack of coordination between thoughts, actions, and emotions. This study provides various intelligent deep learning (DL)-based methods for automated SZ diagnosis via electroencephalography (EEG) signals. The obtained results are compared with those of conventional intelligent methods. To implement the proposed methods, the dataset of the Institute of Psychiatry and Neurology in Warsaw, Poland, has been used. First, EEG signals were divided into 25 s time frames and then were normalized by z-score or norm L2. In the classification step, two different approaches were considered for SZ diagnosis via EEG signals. In this step, the classification of EEG signals was first carried out by conventional machine learning methods, e.g., support vector machine, k-nearest neighbors, decision tree, naïve Bayes, random forest, extremely randomized trees, and bagging. Various proposed DL models, namely, long short-term memories (LSTMs), one-dimensional convolutional networks (1D-CNNs), and 1D-CNN-LSTMs, were used in the following. In this step, the DL models were implemented and compared with different activation functions. Among the proposed DL models, the CNN-LSTM architecture has had the best performance. In this architecture, the ReLU activation function with the z-score and L2-combined normalization was used. The proposed CNN-LSTM model has achieved an accuracy percentage of 99.25%, better than the results of most former studies in this field. It is worth mentioning that to perform all simulations, the k-fold cross-validation method with k = 5 has been used.
RESUMO
Multiple Sclerosis (MS) is a type of brain disease which causes visual, sensory, and motor problems for people with a detrimental effect on the functioning of the nervous system. In order to diagnose MS, multiple screening methods have been proposed so far; among them, magnetic resonance imaging (MRI) has received considerable attention among physicians. MRI modalities provide physicians with fundamental information about the structure and function of the brain, which is crucial for the rapid diagnosis of MS lesions. Diagnosing MS using MRI is time-consuming, tedious, and prone to manual errors. Research on the implementation of computer aided diagnosis system (CADS) based on artificial intelligence (AI) to diagnose MS involves conventional machine learning and deep learning (DL) methods. In conventional machine learning, feature extraction, feature selection, and classification steps are carried out by using trial and error; on the contrary, these steps in DL are based on deep layers whose values are automatically learn. In this paper, a complete review of automated MS diagnosis methods performed using DL techniques with MRI neuroimaging modalities is provided. Initially, the steps involved in various CADS proposed using MRI modalities and DL techniques for MS diagnosis are investigated. The important preprocessing techniques employed in various works are analyzed. Most of the published papers on MS diagnosis using MRI modalities and DL are presented. The most significant challenges facing and future direction of automated diagnosis of MS using MRI modalities and DL techniques are also provided.
Assuntos
Aprendizado Profundo , Esclerose Múltipla , Inteligência Artificial , Humanos , Imageamento por Ressonância Magnética , Espectroscopia de Ressonância Magnética , Esclerose Múltipla/diagnóstico por imagemRESUMO
BACKGROUND AND OBJECTIVE: The effective processing of biomedical images usually requires the interoperability of diverse software tools that have different aims but are complementary. The goal of this work is to develop a bridge to connect two of those tools: ImageJ, a program for image analysis in life sciences, and OpenCV, a computer vision and machine learning library. METHODS: Based on a thorough analysis of ImageJ and OpenCV, we detected the features of these systems that could be enhanced, and developed a library to combine both tools, taking advantage of the strengths of each system. The library was implemented on top of the SciJava converter framework. We also provide a methodology to use this library. RESULTS: We have developed the publicly available library IJ-OpenCV that can be employed to create applications combining features from both ImageJ and OpenCV. From the perspective of ImageJ developers, they can use IJ-OpenCV to easily create plugins that use any functionality provided by the OpenCV library and explore different alternatives. From the perspective of OpenCV developers, this library provides a link to the ImageJ graphical user interface and all its features to handle regions of interest. CONCLUSIONS: The IJ-OpenCV library bridges the gap between ImageJ and OpenCV, allowing the connection and the cooperation of these two systems.
Assuntos
Biologia Computacional/métodos , Processamento de Imagem Assistida por Computador/métodos , Informática Médica/métodos , Software , Aprendizado de Máquina , Testes de Sensibilidade Microbiana , Interface Usuário-ComputadorRESUMO
BACKGROUND AND OBJECTIVE: The manual transformation of DNA fingerprints of dominant markers into the input of tools for population genetics analysis is a time-consuming and error-prone task; especially when the researcher deals with a large number of samples. In addition, when the researcher needs to use several tools for population genetics analysis, the situation worsens due to the incompatibility of data-formats across tools. The goal of this work consists in automating, from banding patterns of gel images, the input-generation for the great diversity of tools devoted to population genetics analysis. METHODS: After a thorough analysis of tools for population genetics analysis with dominant markers, and tools for working with phylogenetic trees; we have detected the input requirements of those systems. In the case of programs devoted to phylogenetic trees, the Newick and Nexus formats are widely employed; whereas, each population genetics analysis tool uses its own specific format. In order to handle such a diversity of formats in the latter case, we have developed a new XML format, called PopXML, that takes into account the variety of information required by each population genetics analysis tool. Moreover, the acquired knowledge has been incorporated into the pipeline of the GelJ system - a tool for analysing DNA fingerprint gel images - to reach our automatisation goal. RESULTS: We have implemented, in the GelJ system, a pipeline that automatically generates, from gel banding patterns, the input of tools for population genetics analysis and phylogenetic trees. Such a pipeline has been employed to successfully generate, from thousands of banding patterns, the input of 29 population genetics analysis tools and 32 tools for managing phylogenetic trees. CONCLUSIONS: GelJ has become the first tool that fills the gap between gel image processing software and population genetics analysis with dominant markers, phylogenetic reconstruction, and tree editing software. This has been achieved by automating the process of generating the input for the latter software from gel banding patterns processed by GelJ.