ABSTRACT: The tomato (Solanum lycopersicum L.) plays a vital role in global agriculture and is a model organism in genetic studies. Visual classification of tomatoes for genetic improvement programs faces challenges due to variety diversity, uneven ripening, external damages, and evaluator subjectivity. Recent advances in the field of computational resources, such as image phenotyping have enabled pre- and post-harvest assessments that are both fast and precise. This study aimed to classify tomato fruits based on shape, group, color, and defects using Convolutional Neural Networks (CNNs). The performance of five architectures - VGG16, InceptionV3, ResNet50, EfficientNetB3, and InceptionResNetV2 was evaluated to identify and determine the most efficient one for this classification. The research considered ten hybrids and their five parental lines. The experiment was conducted in the field, and images of ripe fruits were acquired using a portable mini studio. The ExpImage package in R software was used for fruit individualization by image and to aid in creating a synthetic database for network training. Images were grouped according to their classifications in terms of shape, color, groups, and defects. The InceptionResNetV2 architecture was the most efficient, achieving metrics such as precision and recall exceeding 93 % for most analyzed variables, and shorter classification times. This study advances the understanding of CNN applications in agriculture and research and provides valuable guidelines for optimizing classification tasks in distinct types of fruits. ABSTRACT Solanum L. L studies diversity ripening damages subjectivity resources pre postharvest post harvest precise shape group color CNNs. CNNs . (CNNs) VGG16 VGG InceptionV3 InceptionV ResNet50 ResNet EfficientNetB3 EfficientNetB InceptionResNetV lines studio training groups 9 variables times (CNNs VGG1 ResNet5

The inhibition of the fatty acid amide hydrolase (FAAH), an endocannabinoid system component, emerged as a potentially new therapeutic target for a range of clinical disorders such as acute and chronic pain. Some α-ketoheterocycle derivatives demonstrated interesting analgesic and anti-inflammatory activities in vitro. Ligand-Based Drug Design techniques such as knowledge graph convolutional networks (kGCN) and hologram quantitative structure-activity relationship (HQSAR) using α-ketoheterocycle derivatives from five different datasets were generated to discover the relation between the chemical structures and the inhibition activity. Meanwhile, structure-based drug design simulations as interaction fields (MIF), molecular docking, and ligand sites studies (LSI) from FAAH were performed using Autogrid software and FTmap/FTsite servers. The results of both studies were merged to propose predictive models. The resulting kGCN model demonstrated adequate accuracy area under the curve by receiver operating characteristic (AUC-ROC 0.7922). From contribution maps of the Ligand-Based Drug Design (LBDD) models and the generated probes using MIF and LSI, it was observed that the oxazole ring, the ketone group, and the apolar chain present in the structures of the inhibitors are important, besides the evidence of the Cys269 and Val270 residues importance for the potential interaction, confirmed by carried docking studies. These fragments and structural information can be used to carry out new FAAH potential inhibitors studies and report kGCN as an accurate classification technique. FAAH, , (FAAH) component pain αketoheterocycle α ketoheterocycle antiinflammatory anti inflammatory vitro LigandBased Ligand Based (kGCN structureactivity structure activity HQSAR (HQSAR Meanwhile structurebased based MIF, (MIF) LSI (LSI FTmapFTsite FTmap FTsite servers AUCROC AUC ROC 0.7922. 07922 0.7922 . 0 7922 0.7922) LBDD (LBDD ring group important Cys Cys26 Val Val27 technique (FAAH (MIF 0792 0.792 792 Cys2 Val2 079 0.79 79 07 0.7 7 0.

Abstract: In actuality, the financial investment in Enterprises of the world is common. This investment is performed using internet platforms and value markets. This can generate a loss for many investors due to the uncertainty of the future financial health of the Enterprise. A comparison for the prediction of financial health based on algorithms of machine learning, particularly Support Vector Machine (SVM), Artificial Neural Networks (ANN), K-Nearest Neighbors (KNN), and Adaptive Neuro-Fuzzy Networks (ANFIS) is presented. The database of the Taiwan Economic Journal from 1999 to 2009 is used, with 95 financial ratios of enterprises financially healthiest and with bankruptcy problems. In the ANN, the epoch number, numbers of neurons, activation functions in each layer, loss function, and learning rate are tested; also, an architecture of a Convolutional Neural Network (CNN) is implemented. In SVM, the experiments are performed using different kernels, polynomial, RBF, linear. Besides, the size of C, size of gamma, and size of the polynomial are varied. In KNN, experiments with different numbers of neighbors, types of weight, and values of P are realized. In ANFIS, experiments with variants of the numbers of fuzzy rules, quantity and type of membership functions, number of epochs, and input dimensions are performed. Optimization using Genetic Algorithms (GA) and Particle Swarm Optimization (PSO) of the three models with the best results are performed; the optimization is based on the search for the best hyperparameters that would provide a higher accuracy. The neural network models presented the best average for all the proposed tests.

Abstract: Vision problems are common in patients with diabetes mellitus (DM) because they may suffer from diabetic retinopathy (DR). Because the symptoms of this condition are not easy to detect without the intervention of an expert technician, the use of convolutional neural networks (CNN) has been implemented to speed up the process of analyzing retina images. Due to the good results of this technology, efforts have been made to combine it with other technologies. In this paper, we present the use of an intelligent hybrid system that uses CNNs and Fuzzy Logic with the aim of improving the accuracy obtained. The implementation of fuzzy logic to adjust the hyperparameters of the network allowed us to obtain a mean of 0.9526 with a standard deviation of 0.008521158 in the binary case study, while in the multiclass case study we obtained a mean of 0.7299 and a standard deviation of 0.015614013, offering better results when fuzzy logic is combined compared to when not.

Abstract: Launching and landing rockets on the Earth and in space have had intensive research and development in the last years. The idea of reducing costs is related regularly to the reusability of some mission stages. Though the launch has attracted attention due to including the main engines fundamental to boosting spacecraft onto an orbital or interplanetary trajectory, the landing takes relevance in space missions. While the rocket’s landing has been carried out on Earth using an autonomous spaceport drone ship, it is challenging to design intelligent model-free systems that can continuously learn and compensate for slight deviations until they meet the target. This paper focuses on studying vertical rocket landing using convolutional neural networks. Assuming that the rocket is near the landing area, an attitude rocket control is proposed using a vision system to recognize it and drive -the nozzle TVC. Experimental results show the attitude control commanded by a nozzle TVC of an experimental rocket under different conditions.

Resumen La detección y el diagnóstico de diabetes e hipertensión presentan un escenario desafiante, principalmente en etapas tempranas. Estos dos son degenerativos y de graves consecuencias, provocando daños en múltiples órganos, incluidos los ojos. El oftalmólogo examina el fondo de la retina para discriminar el tipo de enfermedad. En el caso de la retinopatía, como enfermedad crónica, se evidencia en varias clases de progresión. La retinopatía diabética se clasifica en cuatro, que incluyen etapas leve, moderada, grave y proliferativa. En la etapa proliferativa, la retina produce factores vasoproliferativos que estimulan el crecimiento de nuevos vasos sanguíneos, aumentando el riesgo de pérdida de visión. En cuanto a la retinopatía hipertensiva, el método de clasificación más utilizado es la clasificación de Keith-Wagener-Barker. Similar a la retinopatía diabética, también consta de cuatro clases, que van de leves a graves. En la etapa grave, existe un mayor riesgo de desarrollar papiledema o edema retiniano. El objetivo principal de esta investigación es realizar la segmentación y clasificación de imágenes de fondo de retina con retinopatía diabética e hipertensiva. Se propuso una combinación de una red convolucional UNet y una ConvNet para la segmentación de máscara de vasos y la clasificación de retinopatía, respectivamente. El proceso de clasificación se basa en diez clases definidas, donde los valores que van del 0 al 4 representan la retinopatía diabética y los valores del 5 al 9 corresponden a la retinopatía hipertensiva. Los resultados aproximados en la segmentación fueron índices Jaccard de 74%, F1 de 85% y un Accuracy de 96%, y en la clasificación un Accuracy de 80%.

Abstract The detection and diagnosis of diabetes and hypertension present a challenging scenario, mainly in the early stages. These two are degenerative and with severe consequences, causing damage to multiple organs, including the eyes. The ophthalmologist examines the retinal fundus to discriminate between the type of disease. In the case of retinopathy, as a chronic disease, evidence in various classes of progression. Diabetic retinopathy is categorized into four, which include mild, moderate, severe, and proliferative stages. In the proliferative stage, the retina produces vasoproliferative factors that stimulate the growth of new blood vessels, increasing the risk of vision loss. As for the hypertensive retinopathy the most widely used method for classifying is the Keith-Wagener-Barker classification. Like diabetic retinopathy, it consists also of four classes, ranging from mild to severe. In the severe stage, there is an increased risk of developing papilledema or retinal edema. The main objective of this research is to perform segmentation and classification of diabetes and hypertensive retinopathy retinal fundus images. Using a combination of a convolutional network UNet and a ConvNet was proposed for vessel mask segmentation and retinopathy classification, respectively. The classification process relies on ten defined classes, where values ranging from 0 to 4 represent diabetic retinopathy, and values ranging from 5 to 9 correspond to hypertensive retinopathy. The approximate results in the segmentation were Jaccard index of 74%, F1 of 85%, and an Accuracy of 96%, and in the classification was Accuracy of 80%.

Abstract: The accuracy obtained in Art Media Classification (AMC) using CNN is lower compared to other image classification problems, where the acceptable accuracy ranges from 90 to 99%. This article presents an analysis of the performance of three different CNNs with transfer learning for AMC, to answer the question of what challenges arise in this application. We proposed the Art Media Dataset (ArtMD) to train three CNNs. ArtMD contains five classes of art: Drawing, Engraving, Iconography, Painting, and Sculpture. The analysis of the results demonstrates that all the tested CNNs exhibit similar behavior. Drawing, Engraving, and Painting had the highest relationship, showing a strong relationship between Drawing and Engraving. We implemented two more experiments, removing first Drawing and then Engraving. The best performance with 86% accuracy was achieved by removing Drawing. Analysis of the confusion matrix of the three experiments for each CNN confirms that Drawing and Painting have the lowest accuracy, showing a strong misclassification with the other classes. This analysis presents the degree of relationship between the three CNN models and details the challenges of AMC.

Abstract Weeds negatively affect crop yield and quality by competing with the crop for resources. Detecting them in time allows us to optimize control through the precise application of herbicides and reduce their environmental impact. However, its detection and classification are challenging due to the great diversity of species and similarities with crops. Deep learning methods have allowed the development of convolutional neural networks (CNN) for the classification of weed species. However, CNNs require large and varied data sets for training. This study proposes an innovative methodology using CNN, YOLOv8l and RetinaNet classifiers augmented with Stable Diffusion data, implemented artificial images of weeds. Three data sets were configured for training CNNs (real, artificial and mixed) in commercial tomato crops infested by Solanum nigrum L., Portulaca oleracea L. and Setaria verticillata L. The results showed a high degree of agreement with artificial images, reaching a maximum mean Average Precision (mAP) of 0.93 in both networks for the mixed data set. Stable Diffusion made it possible to generate large, high-quality data sets from a limited set of images, reducing the need to obtain a large number of images in the field and their subsequent manual labeling. Furthermore, the method can be adapted to other crops and weed species, thus contributing to the advancement of automated weed management systems.

Resumen Las malas hierbas afectan negativamente el rendimiento y calidad de las cosechas al competir con el cultivo por los recursos. Detectarlas a tiempo permite optimizar el control mediante la aplicación precisa de herbicidas y reducir su impacto ambiental. Sin embargo, su detección y clasificación es un desafío debido a la gran diversidad de especies y similitudes con los cultivos. Los métodos tradicionales de aprendizaje profundo (Deep Learning) han permitido el desarrollo de redes neuronales convolucionales (CNN) para la clasificación de especies de mala hierba. Sin embargo, las CNN requieren amplios y variados conjuntos de datos para su entrenamiento. Este estudio propone una metodología innovadora utilizando clasificadores de CNN, YOLOv8l y RetinaNet aumentados con datos de Stable Diffusion (SD), implementado imágenes de malas hierbas generadas artificialmente. Para ello, se configuraron tres conjuntos de datos para el entrenamiento de las CNNs (real, artificial y mixto) en cultivos de tomate comercial infestados por Solanum nigrum L., Portulaca oleracea L. y Setaria verticillata L. Los resultados mostraron un alto grado de acuerdo con imágenes artificiales, alcanzando un mean Average Precision (mAP) máximo de 0,93 en ambas redes para el conjunto de datos mixto. SD permitió generar grandes conjuntos de datos de alta calidad a partir de un conjunto limitado de imágenes, reduciendo la necesidad de obtener un gran número de imágenes en campo y su posterior etiquetado manual. Además, el método puede adaptarse a otros cultivos y especies de mala hierba, contribuyendo así al avance de los sistemas automatizados de gestión de malas hierbas.

Abstract Accurate detection and identification of weeds are essential in the implementation of Precision Agriculture (PA). In this study, we addressed the detection and classification of weeds in maize and tomato in their early growth stages by integrating images acquired from UAVs and analysis based on advanced convolutional neural networks. Subsequently, the objective of this research was the creation of geo-referenced maps that would allow localized and selective post-emergence management of weed species. Our results indicate that the combination of UAV-captured imagery and deep learning algorithm-based image analysis provides an effective solution for this purpose. The accuracy and efficiency achieved in the identification of weeds in their early stage were promising, both in the set of species present in the crop and in individual species. This advance is of great relevance in the field of PA, as it would allow a more efficient management of weeds by selecting the herbicide according to the type of weed present, reducing the use of herbicides and ultimately contributing to the sustainability and profitability of agriculture. In addition, the generation of geo-referenced maps facilitates real-time decision making. In summary, this study suggests that the combination of emerging technologies, such as drones and deep neural networks, may be applicable tools in localized weed management in the context of PA.

Resumen La detección precisa y la identificación de malas hierbas son esenciales en la implementación de la Agricultura de Precisión (AP). En este estudio, se abordó la detección y la clasificación de malas hierbas en maíz y tomate en sus etapas tempranas de crecimiento mediante la integración de imágenes adquiridas desde drones y análisis basado en avanzadas redes neuronales convolucionales. Posteriormente, el objetivo de esta investigación fue la creación de mapas georreferenciados que permitieran una gestión en post-emergencia localizada y selectiva de las especies de malas hierbas. Nuestros resultados indican que la combinación de imágenes captadas por drones y un análisis de las mismas basado en algoritmos de aprendizaje profundo proporciona una solución efectiva para este propósito. La precisión y eficiencia alcanzadas en la identificación de malas hierbas en su etapa temprana fueron prometedoras, tanto en el conjunto de especies presentes en el cultivo como en especies individuales. Este avance es de gran relevancia en el ámbito de la AP, ya que permitiría una gestión más eficiente de las malas hierbas mediante la selección del herbicida según el tipo de mala hierba presente, reduciendo el uso de herbicidas y, en última instancia, contribuyendo a la sostenibilidad y la rentabilidad de la agricultura. Además, la generación de mapas georreferenciados facilita la toma de decisiones en tiempo real. En resumen, este estudio sugiere que la combinación de tecnologías emergentes, como drones y redes neuronales profundas, pueden ser herramientas aplicables en el manejo localizado de malas hierbas en el contexto de la AP.

Resumo Os veículos aéreos não tripulados (UAVs), muitas vezes chamados de drones, ganharam prevalência progressiva por sua rápida capacidade operacional, bem como por sua ampla aplicabilidade em diversas situações do mundo real. Ultimamente, o uso de UAV na agricultura de precisão tem atraído muito interesse da comunidade científica. Este estudo analisará a ajuda de drones na agricultura de precisão. O big data tem a capacidade de analisar enormes quantidades de dados. Por isso é uma das diversas tecnologias cruciais de tecnologia da informação e comunicação (TIC) que foram aplicadas na agricultura de precisão para a abstração de informações críticas, bem como para auxiliar os praticantes agrícolas na compreensão das práticas agrícolas mais viáveis, e também para uma melhor tomada de decisão. Este trabalho analisa protocolos de comunicação, bem como sua aplicação no desafio de comandar uma frota de drones para proteção de lavouras contra infestações de parasitas. Para tarefas de visão computacional, bem como para aplicações com uso intensivo de dados, o método de aprendizagem profunda mostrou muito potencial. Devido ao seu vasto potencial, também pode ser utilizado na área agrícola. Esta pesquisa empregará vários esquemas para avaliar a eficácia de modelos, incluindo o Grupo de Geometria Visual (VGG-16), a Rede Neural Convolucional (CNN), bem como a Rede Totalmente Convolucional (FCN) na detecção de doenças de plantas. Os métodos de sistemas imunológicos artificiais (AIS) podem ser utilizados para adaptar redes neurais profundas à situação imediata. Os resultados simulados demonstram que o método proposto oferece desempenho superior em relação a vários outros métodos tecnologicamente avançados.

Abstract Unmanned Aerial Vehicles (UAVs), often called drones, have gained progressive prevalence for their swift operational ability as well as their extensive applicability in diverse real-world situations. Of late, UAV usage in precision agriculture has attracted much interest from scientific community. This study will look at drone aid in precise farming. Big data has the ability to analyze enormous amounts of data. Due to this, it is one of the diverse crucial technologies of Information and Communication Technology (ICT) which had applied in precision agriculture for the abstraction of critical information as well as for assisting agricultural practitioners in the comprehension of the most feasible farming practices, and also for better decision-making. This work analyses communication protocols, as well as their application toward the challenge of commanding a drone fleet for protecting crops from infestations of parasites. For computer-vision tasks as well as data-intensive applications, the method of deep learning has shown much potential. Due to its vast potential, it can also be used in the field of agriculture. This research will employ several schemes to assess the efficacy of models includes Visual Geometry Group (VGG-16), the Convolutional Neural Network (CNN) as well as the Fully-Convolutional Network (FCN) in plant disease detection. The methods of Artificial Immune Systems (AIS) can be used in order to adapt deep neural networks to the immediate situation. Simulated outcomes demonstrate that the proposed method is providing superior performance over various other technologically-advanced methods.

RESUMO: A previsão multivariada de séries temporais tem um papel importante em muitos domínios do mundo real. Especialmente, a previsão de preços sempre esteve no foco dos pesquisadores. No entanto, é uma tarefa desafiadora que requer a captura de correlações intra-séries e inter-séries. A maioria dos modelos na literatura foca apenas a correlação no domínio temporal. Neste artigo, selecionamos um novo conjunto de dados do site oficial do Ministério do Comércio Turco. O conjunto de dados consiste em preços diários e volume comercial de vegetais e abrange 1.791 dias entre 1º de janeiro de 2018 e 26 de novembro de 2022. Uma Rede Neural de Gráfico Temporal Espectral é empregada no conjunto de dados curado e os resultados são fornecidos em comparação com CNN, LSTM e Modelos de Floresta Aleatória. A arquitetura GNN alcançou um resultado de ponta (MAE: 1,37, RMSE: 1,94). Até onde sabemos, este é um dos poucos estudos que investiga GNN para análise de séries temporais e o primeiro estudo na área de arquitetura. RESUMO real Especialmente pesquisadores entanto intraséries intra interséries. interséries inter séries. inter-séries temporal artigo Turco 1791 1 791 1.79 º 201 2 2022 CNN Aleatória MAE (MAE 137 37 1,37 RMSE 1,94. 194 1,94 . 94 1,94) sabemos 179 79 1.7 20 202 13 3 1,3 19 1,9 9 17 7 1. 1,

ABSTRACT: Multivariate time series forecasting has an important role in many real-world domains. Especially, price prediction has always been on the focus of researchers. Yet, it is a challenging task that requires the capturing of intra-series and inter-series correlations. Most of the models in literature focus only on the correlation in temporal domain. In this paper, we have curated a new dataset from the official website of Turkish Ministry of Commerce. The dataset consists of daily prices and trade volume of vegetables and covers 1791 days between January 1, 2018 and November 26, 2022. A Spectral Temporal Graph Neural Network (StemGNN) is employed on the curated dataset and the results are given in comparison to Convolutional neural networks (CNN), Long short-term memory (LSTM) and Random Forest models. GNN architecture achieved a state-of-the-art result such as mean absolute error (MAE): 1,37 and root mean squared error (RMSE): 1.94). To our knowledge, this is one of the few studies that investigates GNN for time series analysis and the first study in architecture field. ABSTRACT realworld real world domains Especially researchers Yet intraseries intra interseries inter correlations domain paper Commerce 179 1 201 26 2022 StemGNN (StemGNN CNN, CNN , (CNN) shortterm short term LSTM (LSTM stateoftheart state art MAE (MAE) 137 37 1,3 RMSE (RMSE) 1.94. 194 1.94 . 94 1.94) knowledge field 17 20 2 202 (CNN (MAE 13 3 (RMSE 19 1.9 9 1.

Abstract The detection and segmentation of tumor regions in brain Magnetic Resonance Imaging (MRI) plays important aspects for last two decades. In this paper, Efficient Visual Geometry Group Convolutional Neural Networks Architecture is proposed for brain tumor detection which consists of pre-processing module, DWT module, and Classification module along with segmentation. The scaling based data augmentation methods are used as the pre-processing technique and the data augmented scaled images are decomposed using DWT Transform. The Wavelet feature maps are constructed from these decomposed coefficients and they are classified using EVGG-CNN architecture. This proposed architecture identifies the brain image into either Glioma or healthy brain image. The effectiveness of these developed methods stated in this work is verified by applying the proposed method on the brain images from BRATS 2021 and Kaggle dataset. The performance of proposed EVGG-CNN architecture is analyzed in terms of sensitivity, specificity and accuracy. MRI (MRI decades paper preprocessing pre processing Transform EVGGCNN EVGG CNN 202 dataset sensitivity accuracy 20 2

ABSTRACT Maize yields are important for human food security, and the issue of how to quickly and accurately segment areas of maize disease is an important one in the field of smart agriculture. To address the problem of irregular and multi-area clustering of regions of maize leaf lesions, which can lead to inaccurate segmentation, this paper proposes an improved Swin-Unet model called squeeze-and-excitation Swin-Unet (SE-Swin Unet). Our model applies Swin Transformer modules and skip connection structures for global and local learning. At each skip connection, a SENet module is incorporated to focus on global target features through channel-wise attention, with the aims of highlighting significant regions of disease on maize leaves and suppressing irrelevant background areas. The improved loss function in SE-Swin Unet is based on a combination of the binary cross entropy and Dice loss functions, which form the semantic segmentation model. Compared to other traditional convolutional neural networks on the same dataset, SE-Swin Unet achieves higher mean results for the intersection over union, accuracy, and F1-score, with values of 84.61%, 92.98%, and 89.91%, respectively. The SE-Swin Unet model proposed in this paper is therefore better able to extract information on maize leaf disease, and can provide a reference for the realisation of the complex task of corn leaf disease segmentation. security agriculture multiarea multi area lesions SwinUnet squeezeandexcitation squeeze excitation SESwin SE Unet. . Unet) learning channelwise channel wise attention functions dataset union accuracy F1score, F1score Fscore F1 score, score F F1-score 8461 84 61 84.61% 9298 92 98 92.98% 8991 89 91 89.91% respectively 846 8 6 84.61 929 9 92.98 899 89.91 84.6 92.9 89.9 84. 92. 89.

ABSTRACT Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive and lethal form of cancer with limited prognostic accuracy using traditional factors. This has led to the exploration of innovative prognostic models, including convolutional neural networks (CNNs), in PDAC. CNNs, a type of artificial intelligence algorithm, have shown promise in various medical applications, including image analysis and pattern recognition. Their ability to extract complex features from medical images makes them suitable for improving prognostication in PDAC. However, implementing CNNs in clinical practice poses challenges, such as data availability and interpretability. Future research should focus on multi-center studies, integrating multiple data modalities, and combining CNN outputs with biomarker panels. Collaborative efforts and patient autonomy should be considered to ensure the ethical implementation of CNN-based prognostic models. Further validation and optimisation of CNN-based models are necessary to enhance their reliability and clinical utility in PDAC prognostication. (PDAC factors , (CNNs) algorithm applications recognition However challenges interpretability multicenter multi center studies modalities panels CNNbased based (CNNs

RESUMO Contexto O adenocarcinoma ductal pancreático (ACDP) é uma forma de câncer altamente agressiva e letal com precisão prognóstica limitada usando fatores tradicionais. Isso levou à exploração de modelos prognósticos inovadores, incluindo redes neurais convolucionais (CNNs), no ACDP. As CNNs, um tipo de algoritmo de inteligência artificial, mostraram promessa em várias aplicações médicas, incluindo análise de imagem e reconhecimento de padrões. Sua capacidade de extrair características complexas de imagens médicas as torna adequadas para melhorar o prognóstico no ACDP. No entanto, a implementação de CNNs na prática clínica apresenta desafios, como a disponibilidade de dados e a interpretabilidade. Pesquisas futuras devem se concentrar em estudos multicêntricos, integrando múltiplas modalidades de dados e combinando saídas de CNN com painéis de biomarcadores. Esforços colaborativos e autonomia do paciente devem ser considerados para garantir a implementação ética de modelos prognósticos baseados em CNN. Mais validação e otimização de modelos baseados em CNN são necessárias para aumentar sua confiabilidade e utilidade clínica na prognostico do ACDP. ACDP (ACDP tradicionais inovadores , (CNNs) artificial padrões entanto desafios interpretabilidade multicêntricos biomarcadores (CNNs

ABSTRACT Purpose: This study aimed to evaluate the classification performance of pretrained convolutional neural network models or architectures using fundus image dataset containing eight disease labels. Methods: A publicly available ocular disease intelligent recognition database has been used for the diagnosis of eight diseases. This ocular disease intelligent recognition database has a total of 10,000 fundus images from both eyes of 5,000 patients for the following eight diseases: healthy, diabetic retinopathy, glaucoma, cataract, age-related macular degeneration, hypertension, myopia, and others. Ocular disease classification performances were investigated by constructing three pretrained convolutional neural network architectures including VGG16, Inceptionv3, and ResNet50 models with adaptive moment optimizer. These models were implemented in Google Colab, which made the task straight-forward without spending hours installing the environment and supporting libraries. To evaluate the effectiveness of the models, the dataset was divided into 70%, 10%, and 20% for training, validation, and testing, respectively. For each classification, the training images were augmented to 10,000 fundus images. Results: ResNet50 achieved an accuracy of 97.1%; sensitivity, 78.5%; specificity, 98.5%; and precision, 79.7%, and had the best area under the curve and final score to classify cataract (area under the curve = 0.964, final score = 0.903). By contrast, VGG16 achieved an accuracy of 96.2%; sensitivity, 56.9%; specificity, 99.2%; precision, 84.1%; area under the curve, 0.949; and final score, 0.857. Conclusions: These results demonstrate the ability of the pretrained convolutional neural network architectures to identify ophthalmological diseases from fundus images. ResNet50 can be a good architecture to solve problems in disease detection and classification of glaucoma, cataract, hypertension, and myopia; Inceptionv3 for age-related macular degeneration, and other disease; and VGG16 for normal and diabetic retinopathy. Purpose labels Methods 10000 10 000 10,00 5000 5 5,00 healthy retinopathy glaucoma agerelated age related degeneration hypertension myopia others VGG Inceptionv ResNet ResNet5 optimizer Colab straightforward straight forward libraries 70 70% 10% 20 validation testing respectively Results 97.1% 971 97 1 sensitivity 78.5% 785 78 specificity 98.5% 985 98 precision 797 79 7 79.7% 0964 0 964 0.964 0.903. 0903 0.903 . 903 0.903) contrast VGG1 96.2% 962 96 2 56.9% 569 56 9 99.2% 992 99 84.1% 841 84 0.949 0949 949 0857 857 0.857 Conclusions 1000 00 10,0 500 5,0 97.1 78.5 98.5 79.7 096 0.96 090 0.90 90 96.2 56.9 99.2 84.1 8 0.94 094 94 085 85 0.85 100 10, 50 5, 97. 78. 98. 79. 09 0.9 96. 56. 99. 84. 08 0.8 0.

RESUMO Objetivo: Avaliar o desempenho de classificação de modelos ou arquiteturas de rede neural convolucional pré--treinadas usando um conjunto de dados de imagem de fundo de olho contendo oito rótulos de doenças diferentes. Métodos: Neste artigo, o conjunto de dados de reconhecimento inteligente de doenças oculares publicamente disponível foi usado para o diagnóstico de oito rótulos de doenças diferentes. O banco de dados de reconhecimento inteligente de doenças oculares tem um total de 10.000 imagens de fundo de olho de ambos os olhos de 5.000 pacientes para oito categorias que contêm rótulos saudáveis, retinopatia diabética, glaucoma, catarata, degeneração macular relacionada à idade, hipertensão, miopia, outros. Investigamos o desempenho da classificação de doenças oculares construindo três arquiteturas de rede neural convolucional pré-treinadas diferentes, incluindo os modelos VGG16, Inceptionv3 e ResNet50 com otimizador de Momento Adaptativo. Esses modelos foram implementados no Google Colab o que facilitou a tarefa sem gastar horas instalando o ambiente e suportando bibliotecas. Para avaliar a eficácia dos modelos, o conjunto de dados é dividido em 70% para treinamento, 10% para validação e os 20% restantes utilizados para teste. As imagens de treinamento foram expandidas para 10.000 imagens de fundo de olho para cada tal. Resultados: Observou-se que o modelo ResNet50 alcançou acurácia de 97,1%, sensibilidade de 78,5%, especificidade de 98,5% e precisão de 79,7% e teve a melhor área sob a curva e pontuação final para classificar a categoria da catarata (área sob a curva=0,964, final=0,903). Em contraste, o modelo VGG16 alcançou uma precisão de 96,2%, sensibilidade de 56,9%, especificidade de 99,2% e precisão de 84,1%, área sob a curva 0,949 e pontuação final de 0,857. Conclusão: Esses resultados demonstram a capacidade das arquiteturas de rede neural convolucional pré-treinadas em identificar doenças oftalmológicas a partir de imagens de fundo de olho. ResNet50 pode ser uma boa solução para resolver problemas na detecção e classificação de doenças como glaucoma, catarata, hipertensão e miopia; Inceptionv3 para degeneração macular relacionada à idade e outras doenças; e VGG16 para retinopatia normal e diabética. Objetivo prétreinadas pré treinadas diferentes Métodos artigo 10000 10 000 10.00 5000 5 5.00 saudáveis diabética glaucoma miopia outros VGG Inceptionv ResNet ResNet5 Adaptativo bibliotecas 70 20 teste tal Resultados Observouse Observou se 971 97 1 97,1% 785 78 78,5% 985 98 98,5 797 79 7 79,7 curva0964 0 964 curva=0,964 final=0,903. final0903 final=0,903 . 903 final=0,903) contraste VGG1 962 96 2 96,2% 569 56 9 56,9% 992 99 99,2 841 84 84,1% 0949 949 0,94 0857 857 0,857 Conclusão 1000 00 10.0 500 5.0 97,1 78,5 98, 79, curva096 curva=0,96 final090 final=0,90 90 96,2 56,9 99, 8 84,1 094 94 0,9 085 85 0,85 100 10. 50 5. 97, 78, curva09 curva=0,9 final09 final=0,9 96, 56, 84, 09 0, 08 0,8 curva0 curva=0, final0 final=0, curva=0 final=0 curva= final=