Disease Segmentation Research Articles

Retinal disease projection conditioning by biological traits

Fundus image captures rear of an eye which has been studied for disease identification, classification, segmentation, generation, and biological traits association using handcrafted, conventional, and deep learning methods. In biological traits estimation, most of the studies have been carried out for the age prediction and gender classification with convincing results. The current study utilizes the cutting-edge deep learning (DL) algorithms to estimate biological traits in terms of age and gender together with associating traits to retinal visuals. For the trait’s association, we embed aging as the label information into the proposed DL model to learn knowledge about the effected regions with aging. Our proposed DL models named FAG-Net and FGC-Net, which correspondingly estimates biological traits (age and gender) and generates fundus images. FAG-Net can generate multiple variants of an input fundus image given a list of ages as conditions. In this study, we analyzed fundus images and their corresponding association in terms of aging and gender. Our proposed models outperform randomly selected state-of-the-art DL models.

Detection and Segmentation of Rice Diseases Using Deep Convolutional Neural Networks

Detection and Segmentation of Rice Diseases Using Deep Convolutional Neural Networks

Radial basis function Neural Network optimized with Salp Swarm algorithm espoused paddy leaf disease classification

Diseases in rice often cause yield losses of 20–40% in crop production and are closely linked globally. Disease detection is difficult to quickly plan treatments and decrease crop loss. Diagnosis of rice diseases till done manually. Attain AI-assisted detection of disease, here proposed radial function Neural Network optimized with Salp-Swarm approach (PLDC-RBFNN-SSA) for rice leaf disease segmentation. First, we extract rice leaf images from a dataset containing rice leaf disease image samples. Then the preprocessing augmentation process is performed using a random transformation method such as: Simple image rotation as well as mirror functions implemented to every images. B. Rotate Right 90°, Rotate Left 90°, Flip Vertically, Flip Horizontally, Rotate 180°. These preprocessed output images are then passed to Black Widow's k-means clustering method to segment regions of interest (ROI) for rice leaf disease. The segmented output image is then sent to an adaptive grayscale co-occurrence matrix windowing algorithm (GLCMWAA) to extract radiation features. Extracted features then input into a radial function Neural Network optimized with Salp-Swarm algorithm to classify rice leaf images into bacterial blight, blast, brown spot and tungro. The proposed method is implemented in MATLAB and the performance of the proposed PLDC-RBFNN-SSA approach achieves 15.25%, 18.98%, 20.5% and 24.85% higher accuracy. Computation time is reduced by 50.2%, 48.2%, 38.26%, 20.2% compared with existing methods. Finally, the simulation results demonstrate that the proposed PLDC-RBFNN-SSA method is more efficient and accurate to obtain the optimal global solution for detecting and classifying leaf diseases in rice breeding is shown.

MAC-Leonets-A novel optimized hybrid convolutional neural networks for the segmentation and diagnosis of Edema diseases using retinal OCT images

An intelligent segmentation and identification of edemas diseases constitutes a most important crucial ophthalmological issues since they provide important information for the diagnosis process in accordance to the disease severity. But diagnosing the different edema diseases using the OCT-images are considered to be daunting challenge among the researchers. The implementation of computational intelligence techniques such as machine learning, deep learning, bio inspired algorithms and image processing techniques may help the doctors for some extent in improving the automatic extraction and diagnosis process consequently improving patients’ life quality. But, these are liable to more errors and less performance, which requires further improvisation in designing the intelligent systems for an effective classification of edema diseases. In this context, this paper proposes the hybrid intelligent framework for the identification, segmentation and classification of three types of edemas such as using the retinal optical coherence tomography (OCT) Images. In this process, Single Feed Forward Training networks (SLFTN) are integrated with Convolutional Layers whose hyperparameters are tuned by using Lion Optimization algorithm. An intensive experimentation is carried out using the Kaggle Retinal OCT Image datasets-2020 with Tensor flow and the proposed framework is trained with the different set of 84,494 images in which performance metrics such as accuracy, sensitivity, specificity, recall and f1score are calculated. Results shows the proposed system has provided satisfactory performance, reaching the average highest accuracy of 99.9% in identifying and classifying the respectively.

Local Reversible Transformer for semantic segmentation of grape leaf diseases

Grape leaf diseases segmentation is an essential basis for achieving precise diagnosis and identification of diseases. However, the complex background renders it difficult for small disease areas to be precisely segmented. The existing Transformer mainly focuses on utilizing key and value downsampling to improve model performance while neglecting that downsampling is irreversible with the loss of contextual information. To this end, this paper proposed a novel Locally Reversible Transformer (LRT) segmentation model for grape leaf diseases in natural scene images, whose representation is learned in a reversible downsampling manner. Specifically, a Local Learning Bottleneck (LLB) is developed to enhance local perception and extract richer semantic information of grape leaf diseases via inverted residual convolution. Furthermore, motivated by the wavelet theory, the Reversible Attention (RA) is designed to replace the original downsampling operation by introducing wavelet transform into the multi-headed attention and solving the problem of difficult detection and segmentation of small disease targets with complex backgrounds. Extensive experiments demonstrate that the segmentation performance of LRT outperforms state-of-the-art models with comparable GFLOPs and parameters. Moreover, LRT can retain more multi-grain information and can increase the receptive field to focus on small disease regions with complex backgrounds.

Soybean Disease Detection and Segmentation Based on Mask-RCNN Algorithm

Anthracnose, frogeye leaf spot (FLS), rhizoctonia aerial blight (RAB), soybean mosaic virus (SMV), and yellow mosaic virus (YMV) of soybean are major common soybean leaf diseases that seriously affect soybean yield in India. However, the existing system needs a real-time detection method for soybean leaf diseases, which will help to take appropriate action for disease cure with minimum losses. This study studied a real-time detector for soybean leaf diseases based on deep convolutional neural networks. The 3,127 RGB images of disease-free leaves, anthracnose, FLS, RAB, SMV, and YMV-affected leaves of soybean were collected from the agriculture fields. The Mask R-CNN detection algorithm was used for the detection of soybean leaf diseases by introducing the Res Net 50 module. The pre-processed images (512×512 pixels) were used as input in Mask R-CNN. The model was trained at a number of epochs, training step per epoch training & validation, and learning rate were 80, 500, 50, 8, and 0.001, respectively. The detection accuracy was calculated at three levels of minimum detection confidence i.e. 0.80, 0.85, and 0.90. The results indicate that the maximum detection accuracy i.e. greater than 85% at 0.90 level of minimum detection confidence. This research indicates that the real-time detector based on deep learning provides a feasible solution for diagnosing soybean leaf diseases and provides guidance for the detection of other plant diseases. In addition to that the application of pesticide in the early stage reduce the use of pesticide resulting in less environmental pollution.

Beyin Bilgisayarlı Tomografi Görüntülerinde Derin Öğrenme Tabanlı İskemik İnme Hastalığı Segmentasyonu

Stroke is brain cell death because of either lack of blood flow (ischemic) or bleeding (hemorrhagic) that prevents the brain from functioning properly in both conditions. Ischemic stroke is a common type of stroke caused by a blockage in the cerebrovascular system that prevents blood from flowing to brain regions and directly blocks blood vessels. Computed tomography (CT) scanning is frequently used in the evaluation of stroke, and rapid and accurate diagnosis of ischemic stroke with CT images is critical for determining the appropriate treatment. The manual diagnosis of ischemic stroke can be error-prone due to several factors, such as the busy schedules of specialists and the large number of patients admitted to healthcare facilities. Therefore, in this paper, a deep learning-based interface was developed to automatically diagnose the ischemic stroke through segmentation on CT images leading to a reduction on the diagnosis time and workload of specialists. Convolutional Neural Networks (CNNs) allow automatic feature extraction in ischemic stroke segmentation, utilized to mark the disease regions from CT images. CNN-based architectures, such as U-Net, U-Net VGG16, U-Net VGG19, Attention U-Net, and ResU-Net, were used to benchmark the ischemic stroke disease segmentation. To further improve the segmentation performance, ResU-Net was modified, adding a dilation convolution layer after the last layer of the architecture. In addition, data augmentation was performed to increase the number of images in the dataset, including the ground truths for the ischemic stroke disease region. Based on the experimental results, our modified ResU-Net with a dilation convolution provides the highest performance for ischemic stroke segmentation in dice similarity coefficient (DSC) and intersection over union (IoU) with 98.45 % and 96.95 %, respectively. The experimental results show that our modified ResU-Net outperforms the state-of-the-art approaches for ischemic stroke disease segmentation. Moreover, the modified architecture has been deployed into a new desktop application called BrainSeg, which can support specialists during the diagnosis of the disease by segmenting ischemic stroke.

HMT-Net: Transformer and MLP Hybrid Encoder for Skin Disease Segmentation.

At present, convolutional neural networks (CNNs) have been widely applied to the task of skin disease image segmentation due to the fact of their powerful information discrimination abilities and have achieved good results. However, it is difficult for CNNs to capture the connection between long-range contexts when extracting deep semantic features of lesion images, and the resulting semantic gap leads to the problem of segmentation blur in skin lesion image segmentation. In order to solve the above problems, we designed a hybrid encoder network based on transformer and fully connected neural network (MLP) architecture, and we call this approach HMT-Net. In the HMT-Net network, we use the attention mechanism of the CTrans module to learn the global relevance of the feature map to improve the network's ability to understand the overall foreground information of the lesion. On the other hand, we use the TokMLP module to effectively enhance the network's ability to learn the boundary features of lesion images. In the TokMLP module, the tokenized MLP axial displacement operation strengthens the connection between pixels to facilitate the extraction of local feature information by our network. In order to verify the superiority of our network in segmentation tasks, we conducted extensive experiments on the proposed HMT-Net network and several newly proposed Transformer and MLP networks on three public datasets (ISIC2018, ISBI2017, and ISBI2016) and obtained the following results. Our method achieves 82.39%, 75.53%, and 83.98% on the Dice index and 89.35%, 84.93%, and 91.33% on the IOU. Compared with the latest skin disease segmentation network, FAC-Net, our method improves the Dice index by 1.99%, 1.68%, and 1.6%, respectively. In addition, the IOU indicators have increased by 0.45%, 2.36%, and 1.13%, respectively. The experimental results show that our designed HMT-Net achieves state-of-the-art performance superior to other segmentation methods.

A Weakly Supervised Approach for Disease Segmentation of Maize Northern Leaf Blight from UAV Images

The segmentation of crop disease zones is an important task of image processing since the knowledge of the growth status of crops is critical for agricultural management. Nowadays, images taken by unmanned aerial vehicles (UAVs) have been widely used in the segmentation of crop diseases, and almost all current studies use the study paradigm of full supervision, which needs a large amount of manually labelled data. In this study, a weakly supervised method for disease segmentation of UAV images is proposed. In this method, auxiliary branch block (ABB) and feature reuse module (FRM) were developed. The method was tested using UAV images of maize northern leaf blight (NLB) based on image-level labels only, i.e., only the information as to whether NBL occurs is given. The quality (intersection over union (IoU) values) of the pseudo-labels in the validation dataset achieved 43% and the F1 score reached 58%. In addition, the new method took 0.08 s to generate one pseudo-label, which is highly efficient in generating pseudo-labels. When pseudo-labels from the train dataset were used in the training of segmentation models, the IoU values of disease in the test dataset reached 50%. These accuracies outperformed the benchmarks of the ACoL (45.5%), RCA (36.5%), and MDC (34.0%) models. The segmented NLB zones from the proposed method were more complete and the boundaries were more clear. The effectiveness of ABB and FRM was also explored. This study is the first time supervised segmentation of UAV images of maize NLB using only image-level data was applied, and the above test results confirm the effectiveness of the proposed method.

Development of weighted ensemble transfer learning for tomato leaf disease classification solving low resolution problems

ABSTRACT The manual prediction of Tomato Leaf Diseases (TLD) is more time-consuming, and the implementation cost is high. The various natural characteristic and the low resolution of images make a difficult task for plant disease recognition. The main intention of this paper is to develop a novel tomato leaf disease classification using deep learning techniques, especially for solving the low-resolution issues in the images. The pre-processing is done by the contrast enhancement technique and then the low-resolution problem in pre-processed images is done through Deep Convolutional Neural Networks (DCNN). The segmentation of diseases is carried out through Mask Region-Based Convolutional Neural Networks (Mask R-CNN). The weight of each classifier score is optimized by Controlling Parameter-based Artificial Gorilla Troops Optimization (CP-AGTO). The accuracy and sensitivity of the recommended technique attain 92% and 93%. Thus, the efficacy of the proposed model has been examined with several performance metrics over various recent approaches.

Performance and Usability of Code-Free Deep Learning for Chest Radiograph Classification, Object Detection, and Segmentation.

To evaluate the performance and usability of code-free deep learning (CFDL) platforms in creating DL models for disease classification, object detection, and segmentation on chest radiographs. Six CFDL platforms were evaluated in this retrospective study (September 2021). Single- and multilabel classifiers were trained for thoracic pathologic conditions using Guangzhou pediatric and NIH-CXR14 (ie, National Institutes of Health ChestX-ray14) datasets, and external testing was performed using subsets of NIH-CXR14 and Stanford CheXpert datasets, respectively. Pneumonia detection and pneumothorax segmentation models were trained using the Radiological Society of North America (RSNA) Pneumonia and Society for Imaging Informatics in Medicine (SIIM) Pneumothorax datasets, respectively. Model performance was evaluated using F1 scores. Usability was evaluated based on feasibility of image uploading and model training, ease of use, and cost. NIH-CXR14 and CheXpert datasets contained 112 120 (mean age, 47 years ± 17 [SD]; 63 340 male patients) and 151 522 images (mean age, 61 years ± 18; 88 931 male patients), respectively. The other datasets did not report demographics (Guangzhou, 5826 images; RSNA, 26 683 images; SIIM, 15 301 images). Six platforms offered single-label classifiers, four multilabel classifiers, five object detection models, and one segmentation model. Guangzhou pneumonia classifiers demonstrated good internal (F1, 0.93-0.99) and poor external (F1, 0.39-0.44) performance. Multilabel NIH-CXR14 classifiers showed poor internal and external performance (F1, 0.00-0.36 and 0.00-0.76, respectively). NIH-CXR14 single-label classifiers performed poorly (F1, 0.00, all). The single successfully trained pneumonia detection model had an F1 score of 0.48. No segmentation model was successfully trained. Platform usability was limited, with all requiring some type of coded solution. CFDL platforms demonstrated limited performance and usability for chest radiograph analysis.Keywords: Artificial Intelligence, Automated Machine Learning, Chest Radiographs, Deep Learning, Code-Free Deep Learning, Pneumonia, Pneumothorax, Radiology Supplemental material is available for this article. © RSNA, 2023.

Detecting High-risk Area for Lumpy Skin Disease in Cattle Using Deep Learning Feature

Cattle’s lumpy skin disease is a viral disease that transmits by blood-feeding insects like mosquitoes. The disease mostly affects animals that have not previously been exposed to the virus. Cattle lumpy skin disease impacts milk, beef, and national and international livestock trade. Traditional lumpy skin disease diagnosis is very difficult due to, the lack of materials, experts, and time-consuming. Due to this, it is crucial to use deep learning algorithms with the ability to classify the disease with high accuracy performance results. Therefore, Deep learning-based segmentation and classification are proposed for disease segmentation and classification by using deep features. For this, 10 layers of Convolutional Neural Networks have been chosen. The developed framework is initially trained on a collected Cattle’s lumpy Skin Disease (CLSD) dataset. The features are extracted from input images; hence the color of the skin is very important to identify the affected area during disease representation we used a color histogram. This segmented area of affected skin color is used for feature extraction by a deep pre-trained CNN. Then the generated result is converted into a binary using a threshold. The Extreme learning machine (ELM) classifier is used for classification. The classification performance of the proposed methodology achieved an accuracy of 0.9012% on CLSD To prove the effectiveness of the proposed methods, we present a comparison with the state-of-the-art techniques.

Tomato Leaf Disease Segmentation Using Clustering Method Based on FATPSO with Multi Features

Early blight is one of diseases that infects tomato leaves. This disease causes a decrease in the production of tomato plants. The early detection of this diseases is very important to maintain the tomato production. Monitoring tomato leaves health manually in large area is very time-consuming and inefficient. The drones and computer vision technology give an alternative in solving this problem. One of the important steps in detecting the tomato leaf disease based on computer vision is the segmentation area of the tomato leaf into the healthy and diseased tomato leaf. The K-means clustering offers an image segmentation method that is simple, fast and works unsupervised. However, the solutions of the K-means clustering often be trapped into the local optimum. The Particle Swarm Optimization (PSO) offers a solution of this problem. However, the performance of PSO depends on the particle velocity of the PSO, if the particle velocity is not determined precisely then the PSO will converge prematurely. Fuzzy Adaptive Turbulence Particle Swarm Optimization (FATPSO) is able to control minimum velocity the PSO particles adaptively for overcoming the premature convergence problem in PSO. The good features from image will increase the accuracy of machine learning algorithm. For this reason, these papers the tomato leaf segmentation based on the FATPSO clustering algorithm with multi features. The fitness function of FATPSO uses an objective function of K-means. The experiments use the image taken manually from garden tomatoes. The images have good quality but they have many varieties in size and color. The next research should be considered to use the image taken by drone to guarantee a robust method of image quality produced by drones. The experimental results show that the FATPSO clustering algorithm with multi features has a better performance than the PSO algorithm with multi feature in the tomato leaf disease segmentation

Applications of smart agriculture for environmental protection using deep learning techniques

DL, short for Deep Learning, is a cutting-edge approach that merges advanced techniques in image processing and data analysis with the power of big data analysis. Its potential is enormous and has already found practical applications in several fields, including autonomous driving, automatic speech recognition, medical research, image restoration, natural language processing, and, among others. DL has been recently introduced in agriculture showing promising results in solving various farming problems like disease detection, automated plant and fruit identification, and counting. This study presents a comprehensive review of research using DL techniques in farming, including crop monitoring, crop mapping, weed and pest detection and management, irrigation, fruit grading, reorganizations of species and herbicide identification. Furthermore, different DL techniques applied in various fields are analyzed and compared with existing techniques. It was found that DL outperforms traditional image processing technology in terms of accuracy, both in classification and regression. Additionally, the study suggests that DL can be applied beyond detections, classification tasks to yield production, and disease segmentation in agriculture.

An Effective Image-Based Tomato Leaf Disease Segmentation Method Using MC-UNet.

Tomato disease control is an urgent requirement in the field of intellectual agriculture, and one of the keys to it is quantitative identification and precise segmentation of tomato leaf diseases. Some diseased areas on tomato leaves are tiny and may go unnoticed during segmentation. Blurred edge also makes the segmentation accuracy poor. Based on UNet, we propose an effective image-based tomato leaf disease segmentation method called Cross-layer Attention Fusion Mechanism combined with Multi-scale Convolution Module (MC-UNet). First, a Multi-scale Convolution Module is proposed. This module obtains multiscale information about tomato disease by employing 3 convolution kernels of different sizes, and it highlights the edge feature information of tomato disease using the Squeeze-and-Excitation Module. Second, a Cross-layer Attention Fusion Mechanism is proposed. This mechanism highlights tomato leaf disease locations via gating structure and fusion operation. Then, we employ SoftPool rather than MaxPool to retain valid information on tomato leaves. Finally, we use the SeLU function appropriately to avoid network neuron dropout. We compared MC-UNet to the existing segmentation network on our self-built tomato leaf disease segmentation dataset and MC-UNet achieved 91.32% accuracy and 6.67M parameters. Our method achieves good results for tomato leaf disease segmentation, which demonstrates the effectiveness of the proposed methods.

A survey of artificial intelligence in tongue image for disease diagnosis and syndrome differentiation.

The rapid development of artificial intelligence technology has gradually extended from the general field to all walks of life, and intelligent tongue diagnosis is the product of a miraculous connection between this new discipline and traditional disciplines. We reviewed the deep learning methods and machine learning applied in tongue image analysis that have been studied in the last 5 years, focusing on tongue image calibration, detection, segmentation, and classification of diseases, syndromes, and symptoms/signs. Introducing technical evolutions or emerging technologies were applied in tongue image analysis; as we have noticed, attention mechanism, multiscale features, and prior knowledge were successfully applied in it, and we emphasized the value of combining deep learning with traditional methods. We also pointed out two major problems concerned with data set construction and the low reliability of performance evaluation that exist in this field based on the basic essence of tongue diagnosis in traditional Chinese medicine. Finally, a perspective on the future of intelligent tongue diagnosis was presented; we believe that the self-supervised method, multimodal information fusion, and the study of tongue pathology will have great research significance.

Cascading Autoencoder With Attention Residual U-Net for Multi-Class Plant Leaf Disease Segmentation and Classification

Cascading Autoencoder With Attention Residual U-Net for Multi-Class Plant Leaf Disease Segmentation and Classification

Skin Disease Segmentation Method Based on Network Feature Aggregation Module and Edge Enhanced Attention Mechanism

Skin Disease Segmentation Method Based on Network Feature Aggregation Module and Edge Enhanced Attention Mechanism

AFU-Net: A Novel U-Net Network for Rice Leaf Disease Segmentation

Highlights The attention mechanism enhances the ability of the model to learn specific semantic information in encoder. The redesigned residual structure deepens the network while reducing the number of parameters. The feature extraction module and feature fusion module obtain richer boundary feature information and effectively integrate output results from different levels. The mIoU, mPA, and Precision values of AFU-Net in the self-built dataset are 87.25%, 92.23%, and 99.67%, respectively. Abstract. Rice diseases adversely affect rice growth and yield. Precise spot segmentation helps to assess the severity of the disease so that appropriate control measures can be taken. In this article, we propose a segmentation method called AFU-Net for rice leaf diseases, and its performance is verified through experiments. Based on the traditional UNet, this method incorporates an attention mechanism, a residual module and a feature fusion module (FFM). The attention mechanism is embedded in skip connections, which enhances the learning of particular semantic features in the encoder layer. In addition, the residual module is integrated into the decoder layer, which deepens the network and enables it to extract richer semantic information. The proposed FFM structure effectively enhances the learning of boundary information and local detail features. The experimental results show that the mean intersection over union (mIoU), mean pixel accuracy (mPA) and Precision of the proposed model on the self-built rice leaf disease segmentation dataset are 87.25%, 92.23%, and 99.67%, respectively. All three evaluation indexes were improved over the control group, while the proposed model had the lowest number of parameters and displayed a good segmentation effect for smaller disease points and disease parts with less obvious characteristics. Keywords: Attention mechanism, Feature fusion module, Residual module, Rice leaves, UNet model.

Enhancing Ocular Healthcare: Deep Learning-Based Multi-Class Diabetic Eye Disease Segmentation and Classification

Enhancing Ocular Healthcare: Deep Learning-Based Multi-Class Diabetic Eye Disease Segmentation and Classification