Image Analysis Classification Research Articles

Chicken swarm algorithm with deep convolutional neural network based tongue image analysis for gastric cancer classification

Tongue images (the size, shape, and colour of tongue and the thickness, colour, and moisture content of tongue coating), reflecting the medical condition of entire body based on the model of traditional Chinese medicine (TCM) are extremely utilized in China for millions of years. Gastric cancer (GC) is great lethal kind of cancer in countries and societies. The screening and analysis of GC yet depend on gastroscopy, however its application was significantly restricted due to its invasive, maximum rate and the requirement for expert endoscopists. Early recognition in GC patients and direct treatment contribute significantly to safety for health. Consequently, this study introduces a Chicken Swarm Algorithm with Deep learningbased Tongue Image Analysis for Gastric Cancer Classification (CSADL-TIAGCC) system. The projected CSADL-TIAGCC approach studies the input tongue images for the identification and classification of GC. To accomplish this, the CSADL-TIAGCC system uses improved U-Net segmentation approach. Besides, residual network (ResNet-34) model-based feature extractor is used. Furthermore, long short term memory (LSTM) approach was exploited for GC classification and its hyperparameters are selected by the CSA. The simulation outcome of the CSADL-TIAGCC algorithm was examined under tongue image database. The experimental outcomes illustrate the enhanced results of the CSADL-TIAGCC technique with respect of different evaluation measures.

An efficient colorectal cancer detection network using atrous convolution with coordinate attention transformer and histopathological images

The second most common type of malignant tumor worldwide is colorectal cancer. Histopathology image analysis offers crucial data for the clinical diagnosis of colorectal cancer. Currently, deep learning techniques are applied to enhance cancer classification and tumor localization in histopathological image analysis. Moreover, traditional deep learning techniques might loss integrated information in the image while evaluating thousands of patches recovered from whole slide images (WSIs). This research proposes a novel colorectal cancer detection network (CCDNet) that combines coordinate attention transformer with atrous convolution. CCDNet first denoises the input histopathological image using a Wiener based Midpoint weighted non-local means filter (WMW-NLM) for guaranteeing precise diagnoses and maintain image features. Also, a novel atrous convolution with coordinate attention transformer (AConvCAT) is introduced, which successfully combines the advantages of two networks to classify colorectal tissue at various scales by capturing local and global information. Further, coordinate attention model is integrated with a Cross-shaped window (CrSWin) transformer for capturing tiny changes in colorectal tissue from multiple angles. The proposed CCDNet achieved accuracy rates of 98.61% and 98.96%, on the colorectal histological image and NCT-CRC-HE-100 K datasets correspondingly. The comparison analysis demonstrates that the suggested framework performed better than the most advanced methods already in use. In hospitals, clinicians can use the proposed CCDNet to verify the diagnosis.

Synergistic approach for land use and land cover dynamics prediction in Uttarakhand using cellular automata and Artificial neural network

Alterations in Land use and Land cover (LULC) stand out as a key catalyst for shifts in global climate patterns, environmental conditions, and ecological dynamics. In order to further enhance our comprehension of the effects of variability and climate shifts on the environment, Remote sensing and GIS analytical approaches have been thoroughly explored and are reflected in an imperative vision. Thus, the objective of this study is to model Uttarakhand’s LULC pattern for 2032 and analyse changes in Uttarakhand's LULC trend between 1992 and 2022. LULC change mapping was conducted utilizing a semi-automated hybrid classification approach for high level of accuracy which integrates both Maximum likelihood and Object based image analysis classification techniques on Landsat datasets. The machine learning techniques of Cellular automata and Artificial neural networks (CA-ANN) within MOLUSCE plugin of QGIS have been applied to model future LULC patterns. The accuracy assessment results showed that the overall accuracy of LULC for the years 1992, 2002, 2012, and 2022 was 96.94 %, 97.77 %, 98.61 % and 98.87 % respectively, and the overall kappa statistics coefficient was 0.92, 0.95, 0.94 and 0.95 respectively. The simulated and projected LULC map for 2022 implies a substantially high level of accuracy, with an overall Kappa coefficient value of 0.77 % and 85.39 % of overall correctness. Then, LULC changes for the year 2032 are predicted using CA-ANN. The observed alterations between 1992 and 2032 are significant, characterized by an augmentation in built-up areas, open land, and water bodies, alongside a decline in snow-covered regions, vegetation cover. Whereas, a slight increase is seen in Forested areas. Planners and policy makers aiming to accomplish a more sustainable future and efficient management of the environment will find the changes in LULC over a prolonged period of time to be a useful asset for optimal land use planning.

An ensemble deep learning models approach using image analysis for cotton crop classification in AI-enabled smart agriculture

BackgroundAgriculture is one of the most crucial assets of any country, as it brings prosperity by alleviating poverty, food shortages, unemployment, and economic instability. The entire process of agriculture comprises many sectors, such as crop cultivation, water irrigation, the supply chain, and many more. During the cultivation process, the plant is exposed to many challenges, among which pesticide attacks and disease in the plant are the main threats. Diseases affect yield production, which affects the country’s economy. Over the past decade, there have been significant advancements in agriculture; nevertheless, a substantial portion of crop yields continues to be compromised by diseases and pests. Early detection and prevention are crucial for successful crop management.MethodsTo address this, we propose a framework that utilizes state-of-the-art computer vision (CV) and artificial intelligence (AI) techniques, specifically deep learning (DL), for detecting healthy and unhealthy cotton plants. Our approach combines DL with feature extraction methods such as continuous wavelet transform (CWT) and fast Fourier transform (FFT). The detection process involved employing pre-trained models such as AlexNet, GoogLeNet, InceptionV3, and VGG-19. Implemented models performance was analysed based on metrics such as accuracy, precision, recall, F1-Score, and Confusion matrices. Moreover, the proposed framework employed ensemble learning framework which uses averaging method to fuse the classification score of individual DL model, thereby improving the overall classification accuracy.ResultsDuring the training process, the framework achieved better performance when features extracted from CWT were used as inputs to the DL model compared to features extracted from FFT. Among the learning models, GoogleNet obtained a remarkable accuracy of 93.4% and a notable F1-score of 0.953 when trained on features extracted by CWT in comparison to FFT-extracted features. It was closely followed by AlexNet and InceptionV3 with an accuracy of 93.4% and 91.8% respectively. To further improve the classification accuracy, ensemble learning framework achieved 98.4% on the features extracted from CWT as compared to feature extracted from FFT.ConclusionThe results show that the features extracted as scalograms more accurately detect each plant condition using DL models, facilitating the early detection of diseases in cotton plants. This early detection leads to better yield and profit which positively affects the economy.

Retinal Image Analysis for Simultaneous Classification and Severity Grading of Attention-Deficit Hyperactivity Disorder and Autism Spectrum Disorder using Deep Learning

Over the last 20 years, the number of children born with Autism Spectrum Disorder (ASD) and Attention-Deficit Hyperactivity Disorder (ADHD) has significantly increased. According to the U.S Centers for Disease Control and Prevention, the number of children with ASD in the United States had increased from 1 in 150 in 2000 to 1 in 36 in 2020, which is an increase of 316.7% and the number of children with ADHD has been around 10% of Population. As the number of children with ASD and ADHD is increasing rapidly, an early diagnosis of both mental disorders is essential, which can contribute to the improvement of the condition of children significantly. By employing retinal images, the early detection of ASD and ADHD in the developmental stages of childhood becomes feasible. Consequently, the development of a diagnostic system that identifies the presence of ASD and ADHD is essential for healthcare interventions and the enhancement of children's well-being when potential remediation is attainable. This research aims to develop a system for early diagnosis of ASD and ADHD utilizing retinal images and deep learning through convolutional neural networks. The proposed approach classifies the retinal images into disorder categories and severity levels. Experimental results demonstrate the viability of the proposed approach as a biometric for the early diagnosis of ASD and ADHD.

A hybrid approach of simultaneous segmentation and classification for medical image analysis

A hybrid approach of simultaneous segmentation and classification for medical image analysis

Accuracy Study on Deep Learning-Based CT Image Analysis for Lung Nodule Detection and Classification

Accuracy Study on Deep Learning-Based CT Image Analysis for Lung Nodule Detection and Classification

Integrated image and location analysis for wound classification: a deep learning approach

The global burden of acute and chronic wounds presents a compelling case for enhancing wound classification methods, a vital step in diagnosing and determining optimal treatments. Recognizing this need, we introduce an innovative multi-modal network based on a deep convolutional neural network for categorizing wounds into four categories: diabetic, pressure, surgical, and venous ulcers. Our multi-modal network uses wound images and their corresponding body locations for more precise classification. A unique aspect of our methodology is incorporating a body map system that facilitates accurate wound location tagging, improving upon traditional wound image classification techniques. A distinctive feature of our approach is the integration of models such as VGG16, ResNet152, and EfficientNet within a novel architecture. This architecture includes elements like spatial and channel-wise Squeeze-and-Excitation modules, Axial Attention, and an Adaptive Gated Multi-Layer Perceptron, providing a robust foundation for classification. Our multi-modal network was trained and evaluated on two distinct datasets comprising relevant images and corresponding location information. Notably, our proposed network outperformed traditional methods, reaching an accuracy range of 74.79–100% for Region of Interest (ROI) without location classifications, 73.98–100% for ROI with location classifications, and 78.10–100% for whole image classifications. This marks a significant enhancement over previously reported performance metrics in the literature. Our results indicate the potential of our multi-modal network as an effective decision-support tool for wound image classification, paving the way for its application in various clinical contexts.

Application of satellite images and GIS to assess the site situation and propose solutions for developing pineapple cultivation in U Minh Thuong, Kien Giang, Vietnam

The study was conducted to assess the site situation and propose solutions to develop the cropping area of pineapples in U Minh Thuong (UMT), Kien Giang province. The study used the Object-Based Image Analysis (OBIA) classification method to create a map of pineapple cropping in the U Minh Thuong area in 2016. Data on rainfall, soil type, temperature, and water reserves are inherited from the Department of Agriculture and Rural Development of Kien Giang Province and analyzed by the GIS method. Water irrigation in pineapple was estimated based on the formula of Mladen Todorovic, 2016 comparing freshwater reserves in the research region and irrigation water on pineapple, researched spatial distribution map of water response capacity for pineapple. The research results showed that the agricultural area in U Minh Thuong has natural conditions suitable for pineapple cultivation. The size of pineapple cultivation in this area was 4,387.29 ha, the largest in the Vinh Thuan district. The total freshwater reserve in the region varies, where the rainfall is, with the highest volume being 4,546.2 m3/ha (in September 2015) and the lowest 67.7 m3/ha (in March 2016). However, the U Minh Thuong region lacks water in the dry months from January 2016 to April 2016. The amount of water during this period is starved, with watering demand from 2,385 to 2,969 million m3 on the total cultivated area of pineapple. In contrast, the surplus water was quite high in the rainy season, about 2,252 - 27,257 million m3 in Vinh Thuan district in September 2015. The study proposes a solution to store water during the rainy season and use them effectively during the dry season. This solution is compatible with the main pineapple farming habit and model in U Minh Thuong, Kien Giang Province.

Cocoa plant mapping in Aceh Tamiang regency, Aceh Indonesia: an object-based image analysis approach

Cocoa is one of the leading commodities of plantations that has a vital role in the Indonesian economy. Due to the unavailability of detailed cocoa area data in Aceh Tamiang Regency, mapping was carried out by utilizing remote sensing technology. This study aims to map and determine the distribution of cocoa plants in Aceh Tamiang Regency using the object-based image analysis (OBIA) method and the accuracy test of the OBIA method to map cocoa plants. The object-based image analysis (OBIA) classification results for cocoa plants covered 946.87 ha and cocoa plants spread across 12 districts in Aceh Tamiang Regency. The widest distribution of cocoa is in Karang Baru District, 139.22 ha with a percentage of 14.70%, and the smallest area in Kota Kualasimpang District, covering an area of 6.30 ha with a rate of 0.67%. The results of the accuracy test to map cocoa plants using the OBIA method obtained an overall accuracy value of 95% and a kappa value of 86%.

ANALISIS TINGGI TERBANG UNMANNED AERIAL VEHICLE TERHADAP KLASIFIKASI TUTUPAN LAHAN MENGGUNAKAN OBJECT BASED IMAGE ANALYSIS

The development of forest and land cover classification methods was currently increasing rapidly, one of which is by utilizing remote sensing applications, namely UAV technology. One method with high accuracy that can be used to classify forest and land cover is the object-based image analysis (OBIA) classification method. There is an assumption that the results of orthophoto shooting will look the same at a glance, but each orthophoto result also has a different spatial resolution and affects the detail of each OBIA segmentation. This makes this research important to do how to classify forest and land cover in the Ghimbo Pomuan customary forest, Kenegerian Kampa, Kampar Riau at different flying heights of the UAV. This study aims to map forest and land cover classes based on the UAV flight height and analyze the UAV flight altitude on the resulting land cover classification. The method used is exploration and processing of UAV imagery. Starting with the preparation and determination of observation points, forest and land cover classification using the OBIA method at several UAV flight heights and data analysis. In the Ghimbo Pomuan customary forest area, several land cover classifications were produced using OBIA. The land cover classes include water bodies, forests, oil palm plantations and shrubs. Classification of land cover in the water body class has a tendency of under segmentation at UAV flight height of 120 m. Object details in the oil palm and shrub land cover class were higher at UAV flight heights of 100 m than at 120 m and 130 m, whereas in the forest class there was no difference.

Deep learning image analysis for filamentous fungi taxonomic classification: Dealing with small datasets with class imbalance and hierarchical grouping.

Deep learning applications in taxonomic classification for animals and plants from images have become popular, while those for microorganisms are still lagging behind. Our study investigated the potential of deep learning for the taxonomic classification of hundreds of filamentous fungi from colony images, which is typically a task that requires specialized knowledge. We isolated soil fungi, annotated their taxonomy using standard molecular barcode techniques, and took images of the fungal colonies grown in petri dishes (n = 606). We applied a convolutional neural network with multiple training approaches and model architectures to deal with some common issues in ecological datasets: small amounts of data, class imbalance, and hierarchically structured grouping. Model performance was overall low, mainly due to the relatively small dataset, class imbalance, and the high morphological plasticity exhibited by fungal colonies. However, our approach indicates that morphological features like color, patchiness, and colony extension rate could be used for the recognition of fungal colonies at higher taxonomic ranks (i.e. phylum, class, and order). Model explanation implies that image recognition characters appear at different positions within the colony (e.g. outer or inner hyphae) depending on the taxonomic resolution. Our study suggests the potential of deep learning applications for a better understanding of the taxonomy and ecology of filamentous fungi amenable to axenic culturing. Meanwhile, our study also highlights some technical challenges in deep learning image analysis in ecology, highlighting that the domain of applicability of these methods needs to be carefully considered.

Implementation and Optimization Of Inception Resnet-v2 with Data Balancing (Case Study Of Lung Disease Classification)

Lungs are vital organs in humans because the process of breathing in humans occurs through the lung organs. However, there are diseases of the lungs, such as COVID-19, Pneumonia, and Tuberculosis that can disrupt the respiratory system in humans. Early detection is required by chest x-ray examination. The development of artificial intelligence technology can help classify chest x-rays with image analysis. In classification for image analysis, there are problems where the data is not balanced, which can cause errors in classification. Thus, data balancing is needed to balance the data. This study conducted training using InceptionResnet-v2 with data balancing. The best model performance results were obtained by training using random oversampling on the model using epoch 20, batch size 64, and learning rate 0.0001 with an accuracy value of 89.23%, loss 0.28, precision 90.05%, recall 89.88%, F1-score 89.74%, and AUC 98%. The accuracy value increased by 7.52% compared to the imbalanced dataset and by 1.29% compared to the random undersampling dataset. Abstract.

Rapid wildfire damage estimation using integrated object-based classification with auto-generated training samples from Sentinel-2 imagery on Google Earth Engine

Satellite data are essential during wildfires for understanding its adverse effects and improving the effectiveness of rapid disaster management. However, existing techniques used for damage assessments are inaccurate and lack automation. In this study, we propose an integrated machine learning approach with auto-generated training samples for a rapid wildfire disaster response framework using Sentinel-2 imagery at 10 m resolution from Google Earth Engine (GEE). First, training samples of burned areas were obtained by utilizing textural data based on features that had changed because of the wildfire, and samples of unburned areas were obtained using the normalized difference vegetation index (NDVI). The images were categorized as burned and unburned images using the object-based image analysis (OBIA) classification method. Finally, using the classified maps, burn severity maps and estimated pixel counts for each severity class were generated and compared. The proposed method was implemented to put out a wildfire that broke out in Uljin, Gyeongsangbuk-do, South Korea in March 2022 and the transferability of the model was evaluated in Gangneung, Gangwon-do, South Korea. The study findings indicate that the random forest (RF) classifier acquired the greatest overall accuracy (OA) of 97.6 % in Uljin; additionally, the model transferability performed well in Gangneung with an OA of 93.8 %. The RF also generated the fewest pixels of the unchanged class when the burn severity map was evaluated. Overall, our study proposes a quick and automated approach for estimating wildfire damage that could be used for immediate mitigation actions.

A Novel Cluster Matching-Based Improved Kernel Fisher Criterion for Image Classification in Unsupervised Domain Adaptation

Unsupervised domain adaptation (UDA) is a popular approach to reducing distributional discrepancies between labeled source and the unlabeled target domain (TD) in machine learning. However, current UDA approaches often align feature distributions between two domains explicitly without considering the target distribution and intra-domain category information, potentially leading to reduced classifier efficiency when the distribution between training and test sets differs. To address this limitation, we propose a novel approach called Cluster Matching-based Improved Kernel Fisher criterion (CM-IKFC) for object classification in image analysis using machine learning techniques. CM-IKFC generates accurate pseudo-labels for each target sample by considering both domain distributions. Our approach employs K-means clustering to cluster samples in the latent subspace in both domains and then conducts cluster matching in the TD. During the model component training stage, the Improved Kernel Fisher Criterion (IKFC) is presented to extend cluster matching and preserve the semantic structure and class transitions. To further enhance the performance of the Kernel Fisher criterion, we use a normalized parameter, due to the difficulty in solving the characteristic equation that draws inspiration from symmetry theory. The proposed CM-IKFC method minimizes intra-class variability while boosting inter-class variants in all domains. We evaluated our approach on benchmark datasets for UDA tasks and our experimental findings show that CM-IKFC is superior to current state-of-the-art methods.

Toward the automatic detection of effective gerbil holes in desert grasslands through unmanned aerial vehicle imagery

Rapidly determining the density of effective gerbil holes is ecologically important but technically challenging. However, unmanned aerial vehicles (UAVs) offer new methods to identify effective gerbil holes and understand the spatial relationships between gerbils and grass coverage. In this study, we focused on Meriones unguiculatus, which live in desert grasslands of Ordos, and gathered UAV imagery to explore the performance of rule-based classification in object-based image analysis (OBIA) for identifying effective gerbil holes. To improve the classification accuracy and reduce the time cost of repeated ‘trial and error’, we propose a novel method to build rule sets. We adopted the estimation of scale parameter_2 (ESP2) method during the segmentation stage to enable fast and objective segmentation parameterization. For the classification stage, a correlation-based feature selection (CFS) algorithm was applied for feature selection and threshold prediction. Our analysis determined the following universally recommended rule sets for identifying effective gerbil holes using OBIA: scale parameter of 107, shape factor of 0.2, compactness value of 0.3, excess green value less than −5, brightness feature value in the 143–149 range and roundness feature less than 0.9. With these rule sets, OBIA exhibited higher classification accuracy than maximum likelihood (ML) classification, with an overall classification accuracy of 88.25%. We also found that the relationship between the area of effective gerbil holes (y) and the grass coverage (x) satisfied a quadratic function: This study provides practical guidance for grassland management and rodent infestation control.

Classification of COVID-19 Patient Based on Multilayer Perceptron Neural Networks Optimized with Garra Rufa Fish Optimization Using CT Scan Images

COVID-19 is known in recent times as a severe syndrome of respiratory organ (Lungs) and has gradually produced pneumonia, a lung disorder all around the world. As coronavirus is continually spreading rapidly globally, the computed tomography (CT) technique has been made important and essential for quick diagnosis of this dangerous syndrome. Hence, it is necessitated to develop a precise computer-based technique for assisting medical clinicians in identifying the COVID-19 influenced patients with the help of CT scan images. Therefore, the multilayer perceptron neural networks optimized with Garra Rufa Fish optimization using images of CT scan is proposed in this paper for the classification of COVID-19 patients (COV-19-MPNN-GRF-CTI). The input images are taken from SARS-COV-2 CT-scan dataset. Initially, the input images are pre-processed utilizing convolutional auto-encoder (CAE) to enhance the quality of the input images by eliminating noises. The pre-processed images are fed to Residual Network (ResNet-50) for extracting the global and statistical features. The extraction over the features of CT scan images is made through ResNet-50 and subsequently input to multilayer perceptron neural networks (MPNN) for CT images classification as COVID-19 and Non-COVID-19 patients. Here, the layer of Batch Normalization of the MPNN is separated and added with ResNet-50 layer. Generally, MPNN classifier does not divulge any adoption of optimization approach for calculating the optimal parameters and accurately classifying the extracted features of CT images. The Garra Rufa Fish (GRF) optimization algorithm performs to optimize the weight parameters of MPNN classifiers. The proposed approach is executed in MATLAB. The performance metrics, such as sensitivity, precision, specificity, [Formula: see text]-measure, accuracy and error rate, are examined. Then the performance of the proposed COV-19-MPNN-GRF-CTI method provides 22.08%, 24.03%, 34.76% higher accuracy, 23.34%, 26.45%, 34.44% higher precision, 33.98%, 21.95%, 34.78% lower error rate compared with the existing methods, like multi-task deep learning using CT image analysis for COVID-19 pneumonia classification and segmentation (COV-19-MDP-CTI), COVID-19 classification utilizing CT scan depending on meta-classifier approach (COV-19-SEMC-CTI) and deep learning-based COVID-19 prediction utilizing CT scan images (COV-19-CNN-CTI), respectively.

Geomorphological evolution of the Blanco Este River after recent eruptions of the Calbuco volcano, Chile

In this paper, we retrace the geomorphological evolution of a 6.5 km-long segment of the Blanco Este River, which drains the Calbuco volcano in southern Chile, highlighting the geomorphic signatures of the disturbances caused by the 1961 and 2015 eruptions. In particular, we track the geomorphological changes and the evolution of the geomorphological units and discuss the underlying processes. In addition, we use the NDVI (Normalize Difference Vegetation Index) to characterise the vegetation recovery at the basin and the active channel scales. The reconstruction of the geomorphological changes is performed by means of photointerpretation and supervised classification image analysis. Moreover, we quantify braiding indices and active channel widths to characterise the fluvial corridor and to describe its evolution after the two eruptive events. It is found that the evolutionary trajectories of the Blanco River immediately after the eruptive events of 1961 and 2015 exhibited strong similarities. In fact, the major geomorphological adjustment of the Blanco Este River occurred during the first period of observation (19 years) after the 1961 eruption. Subsequently the river progressed gradually towards an advanced relaxation stage in 2014. For 53 years (between 1961 and 2014), the reduction of the active channel width was the dominant process, followed by erosion, which favoured the regeneration and consequent expansion of vegetation cover within the river corridor. The 2015 eruption reset the fluvial landform and a new cycle of sedimentation, braiding and formation of central bars began; the highly dynamic pattern is still observed today.

Sparrow Search Algorithm With Stacked Deep Learning Based Medical Image Analysis for Pancreatic Cancer Detection and Classification

Sparrow Search Algorithm With Stacked Deep Learning Based Medical Image Analysis for Pancreatic Cancer Detection and Classification

Use of machine learning models-based image analysis for classification of haploid and diploid maize

Use of machine learning models-based image analysis for classification of haploid and diploid maize