Computer-aided Diagnosis Research Articles

Inspired by “Focus, Fusion, Collaboration”: A multi-level ensemble network for automatic pneumonia diagnosis from full slice CT images

Inspired by “Focus, Fusion, Collaboration”: A multi-level ensemble network for automatic pneumonia diagnosis from full slice CT images

Depression level prediction via textual and acoustic analysis.

Depression level prediction via textual and acoustic analysis.

A NOVEL METHOD TO DETECT HUMAN DENTAL CAVITIES USING YOLOv11

One of the most important facets of oral healthcare is the spotting of dental cavities, or caries, which are often done manually by qualified specialists utilizing radiography, visual inspection, or tactile methods. Despite their effectiveness, these traditional approaches are frequently arbitrary, labor-intensive, and prone to human mistake. In this study, we present a unique method for detecting dental cavities in humans by utilizing the You Only Look Once (YOLO) v11 architecture in conjunction with a deep learning-based object identification framework. A large dataset’s of dental images is used to train YOLOv11, a cutting-edge real-time object detection model, to detect cavities in their early stages, allowing for automatic and precise diagnosis. Our method makes use of YOLOv11’s high precision and low latency cavity detection and localization capabilities, even in complicated dental pictures. Training data for the suggested model comes from annotated images.

Cheetah optimized CNN: A bio-inspired neural network for automated diabetic retinopathy detection

The escalating global prevalence of diabetes has underscored the critical need for effective screening and diagnosis of diabetic retinopathy (DR), a common complication of diabetes that can lead to irreversible vision loss. In this study, we propose a novel algorithm for automated DR detection in retinal fundus images using deep learning techniques. The algorithm incorporates a cheetah optimized convolutional neural network (CO-CNN) that draws inspiration from cheetah hunting behavior for efficient image processing, segmentation, feature extraction, and classification. Preprocessing steps involve median filter and contrast limited adaptive histogram equalization to enhance image quality. The segmented output is clustered using the cascaded fuzzy C-means algorithm and features are extracted with the speeded-up robust features algorithm. The experimental results on the Indian Diabetic Retinopathy Image Dataset demonstrate an accuracy of 98.64% in predicting various stages of DR. The proposed CO-CNN approach shows superior performance compared to that of state-of-the-art methods, offering potential applications in telemedicine, treatment planning, early detection, screening, and patient education. Integrating fuzzy logic enhances the model’s interpretability and robustness, paving the way for improved healthcare outcomes in diabetic retinopathy management.

Fusion-Brain-Net: A Novel Deep Fusion Model for Brain Tumor Classification.

Brain tumors are among the most prevalent and lethal diseases. Early diagnosis and precise treatment are crucial. However, the manual classification of brain tumors is a laborious and complex task. This study aimed to develop a fusion model to address certain limitations of previous works, such as covering diverse image modalities in various datasets. We presented a hybrid transfer learning model, Fusion-Brain-Net, aimed at automatic brain tumor classification. The proposed method included four stages: preprocessing and data augmentation, fusion of deep feature extractions, fine-tuning, and classification. Integrating the pre-trained CNN models, VGG16, ResNet50, and MobileNetV2, the model enhanced comprehensive feature extraction while mitigating overfitting issues, improving the model's performance. The proposed model was rigorously tested and verified on four public datasets: Br35H, Figshare, Nickparvar, and Sartaj. It achieved remarkable accuracy rates of 99.66%, 97.56%, 97.08%, and 93.74%, respectively. The numerical results highlight that the model should be further investigated for potential use in computer-aided diagnoses to improve clinical decision-making.

Development and Validation of an Algorithm for Segmentation of the Prostate and its Zones from Three-dimensional Transrectal Multiparametric Ultrasound Images.

Development and Validation of an Algorithm for Segmentation of the Prostate and its Zones from Three-dimensional Transrectal Multiparametric Ultrasound Images.

Segmentation of skin layers on HFUS images using the attention mechanism.

Segmentation of skin layers on HFUS images using the attention mechanism.

Automatic melanoma and non-melanoma skin cancer diagnosis using advanced adaptive fine-tuned convolution neural networks

Skin Cancer is an extensive and possibly dangerous disorder that requires early detection for effective treatment. Add specific global statistics on skin cancer prevalence and mortality to emphasize the importance of early detection. Example: “Skin cancer accounts for 1 in 5 diagnosed cancers globally, with melanoma causing over 60,000 deaths annually. Manual skin cancer screening is both time-intensive and expensive. Deep learning (DL) techniques have shown exceptional performance in various applications and have been applied to systematize skin cancer diagnosis. However, training DL models for skin cancer diagnosis is challenging due to limited available data and the risk of overfitting. Traditionally approaches have High computational costs, a lack of interpretability, deal with numerous hyperparameters and spatial variation have always been problems with machine learning (ML) and DL. An innovative method called adaptive learning has been developed to overcome these problems. In this research, we advise an intelligent computer-aided system for automatic skin cancer diagnosis using a two-stage transfer learning approach and Pre-trained Convolutional Neural Networks (CNNs). CNNs are well-suited for learning hierarchical features from images. Annotated skin cancer photographs are utilized to detect ROIs and reset the initial layer of the pre-trained CNN. The lower-level layers learn about the characteristics and patterns of lesions and unaffected areas by fine-tuning the model. To capture high-level, global features specific to skin cancer, we replace the fully connected (FC) layers, responsible for encoding such features, with a new FC layer based on principal component analysis (PCA). This unsupervised technique enables the mining of discriminative features from the skin cancer images, effectively mitigating overfitting concerns and letting the model adjust structural features of skin cancer images, facilitating effective detection of skin cancer features. The system shows great potential in facilitating the initial screening of skin cancer patients, empowering healthcare professionals to make timely decisions regarding patient referrals to dermatologists or specialists for further diagnosis and appropriate treatment. Our advanced adaptive fine-tuned CNN approach for automatic skin cancer diagnosis offers a valuable tool for efficient and accurate early detection. By leveraging DL and transfer learning techniques, the system has the possible to transform skin cancer diagnosis and improve patient outcomes.

Computer-aided diagnosis tool utilizing a deep learning model for preoperative T-staging of rectal cancer based on three-dimensional endorectal ultrasound.

The prognosis and treatment outcomes for patients with rectal cancer are critically dependent on an accurate and comprehensive preoperative evaluation.Three-dimensional endorectal ultrasound (3D-ERUS) has demonstrated high accuracy in the T staging of rectal cancer. Thus, we aimed to develop a computer-aided diagnosis (CAD) tool using a deep learning model for the preoperative T-staging of rectal cancer with 3D-ERUS. We retrospectively analyzed the data of 216 rectal cancer patients who underwent 3D-ERUS. The patients were randomly assigned to a training cohort (n = 156) or a testing cohort (n = 60). Radiologists interpreted the 3D-ERUS images of the testing cohort with and without the CAD tool. The diagnostic performance of the CAD tool and its impact on the radiologists' interpretations were evaluated. The CAD tool demonstrated high diagnostic efficacy for rectal cancer tumors of all T stages, with the best diagnostic performance achieved for T1-stage tumors (AUC, 0.85; 95% CI, 0.73-0.93). With assistance from the CAD tool, the AUC for T1 tumors improved from 0.76 (95% CI, 0.63-0.86) to 0.80 (95% CI, 0.68-0.94) (P = 0.020) for junior radiologist 2. For junior radiologist 1, the AUC improved from 0.61 (95% CI, 0.48-0.73) to 0.79 (95% CI, 0.66-0.88) (P = 0.013) for T2 tumors and from 0.73 (95% CI, 0.60-0.84) to 0.84 (95% CI, 0.72-0.92) (P = 0.038) for T3 tumors. The diagnostic consistency (κ value) also improved from 0.31 to 0.64 (P = 0.005) for the junior radiologists and from 0.52 to 0.66 (P = 0.005) for the senior radiologists. A CAD tool utilizing a deep learning model based on 3D-ERUS images showed strong performance in T staging rectal cancer. This tool could improve the performance of and consistency between radiologists in preoperatively assessing rectal cancer patients.

Deep learning-based decision support system for cervical cancer identification in liquid-based cytology pap smears.

BackgroundCervical cancer is the fourth most common cause of women cancer deaths worldwide. The primary etiology of cervical cancer is the persistent infection of specific high-risk strains of the human papillomavirus. Liquid-based cytology is the established method for early detection of cervical cancer. The evaluation of cellular abnormalities at a microscopic level allows for the identification of malignant or precancerous features in liquid-based cytology pap smears. This technique is characterized by its time-consuming nature and susceptibility to both inter- and intra-observer variability. Hence, the utilization of Artificial Intelligence in computer-assisted diagnosis can reduce the duration needed for diagnosing this ailment, thereby eliminating delayed diagnosis and facilitating the implementation of an efficient treatment.ObjectiveThis research presents a new deep learning-based cervical cancer identification decision support system in liquid-based cytology smear images.MethodsThe proposed diagnosis support system incorporates a novel hybrid feature reduction and optimization module, which integrates a sparse Autoencoder with the Binary Harris Hawk metaheuristic optimization algorithm to select the most informative features from a supplemented feature set of the input images. The supplemented feature set is retrieved by three pretrained Convolutional Neural Networks. The module utilizes an improved feature set to conduct a Bayesian-optimized K Nearest Neighbors machine learning classification of cervical cancer in input Pap smears.ResultsThe introduced approach achieves a classification accuracy of 99.9% and demonstrates an improved ability to detect the stages of cervical cancer, with a sensitivity of 99.8%. In addition, the system has the ability to identify the lack of cervical cancer stages with a specificity rate of 99.9%.ConclusionThe proposed system outpaces recent deep learning-based cervical cancer identification systems.

Automatic Detection of Genetic Diseases in Pediatric Age Using Pupillometry

Abstract— This paper introduces a comprehensive framework for automated disease detection using pupillometry data. Our approach establishes a robust pipeline that includes data preprocessing, feature extraction, and machine learning-based classification of patients based on their pupillary responses. We extract key features from both left and right pupil diameter measurements, such as maximum and minimum values, delta, channel height (CH), latency, and mean change velocity (MCV).To enhance classification accuracy, we train and evaluate multiple machine learning models, including Support Vector Machines (SVMs), an ensemble classifier, Extreme Learning Machines (ELM), Multi-Layer Perceptrons (MLPs), and Random Forests. Additionally, we propose a novel hybrid model that integrates the strengths of multiple algorithms, outperforming individual models in accuracy. Our experimental results highlight the effectiveness of this hybrid approach, demonstrating its potential for improving non-invasive and efficient disease diagnosis. This research contributes to advancements in clinical ophthalmology and neurology by leveraging pupillometry and machine learning for more precise and accessible diagnostic tools. Keywords: Pupillometry data, Channel height, Latency, Mean Change Velocity, Support Vector Machines, Extreme Learning Machines, Multi Layer Perceptrons, Random Forest.

AI-Driven Alzheimer’s Detection Performance Analysis of Pretrained CNN Models on MRI Data

Alzheimer's disease (AD) is a memory and cognitive function neurodegenerative disease. Early diagnosis is important for appropriate treatment, but conventional diagnosing techniques are not very effective. In this paper, automatic detection of Alzheimer's disease from MRI scans using deep learning techniques is discussed. We trained and compared six pre-trained Convolutional Neural Network (CNN) models—VGG19, ResNet101, EfficientNetB3, MobileNetV2, InceptionV3, and DenseNet121—to differentiate MRI scans as Alzheimer and Non-Alzheimer. In experiments, InceptionV3 is the optimal model for classification of Alzheimer's disease and all others are inferior with an accuracy of 99.2%. Confusion matrices, classification reports, and accuracy metrics were utilized to test performance. We also created a web-based real-time MRI classification tool with a user-friendly interface for clinicians. This work showcases the potential of deep learning in medical imaging and the detection of early disease, and opens the door to AI-augmented improvement in Alzheimer's diagnosis

Improving Diabetic Retinopathy Detection by using Image Processing and Machine Learning

The rapid development and proliferation of medical imaging technologies is revolutionizing medicine. Medical imaging allows scientists and physicians to glean potentially life-saving information by peering noninvasively into the human body. Diabetic retinopathy (DR) is an irreversible fundus retinopathy. A deep learning-based auto-mated DR diagnosis system can save diagnostic time. This paper reviews and analyses state-of-the-art deep learning methods in supervised, self-supervised and Vision Transformer setups, proposing retinal fundus image classification and detection. For instance, referable, nonreferable and proliferative classifications of Diabetic Retinopathy are reviewed and summarized. Moreover, this paper discusses the available retinal fundus datasets for Diabetic Retinopathy that are used for tasks such as detection, classification and segmentation. The paper also assesses research gaps in the area of DR detection/classification and addresses various challenges that need further study and investigation

A hybrid deep learning framework for early detection of diabetic retinopathy using retinal fundus images

Recent advancements in deep learning have significantly impacted medical image processing domain, enabling sophisticated and accurate diagnostic tools. This paper presents a novel hybrid deep learning framework that combines convolutional neural networks (CNNs) and recurrent neural networks (RNNs) for diabetic retinopathy (DR) early detection and progression monitoring using retinal fundus images. Utilizing the sequential nature of disease progression, the proposed method integrates temporal information across multiple retinal scans to enhance detection accuracy. The proposed model utilizes publicly available DRIVE and Kaggle diabetic retinopathy datasets to evaluate the performance. The benchmark datasets provide a diverse set of annotated retinal images and the proposed hybrid model employs a CNN to extract spatial features from retinal images. The spatial feature extraction is enhanced by multi-scale feature extraction to capture fine details and broader patterns. These enriched spatial features are then fed into an RNN with attention mechanism to capture temporal dependencies so that most relevant data aspects can be considered for analysis. This combined approach enables the model to consider both current and previous states of the retina, improving its ability to detect subtle changes indicative of early-stage DR. Proposed model experimental evaluation demonstrate the superior performance over traditional deep learning models like CNN, RNN, InceptionV3, VGG19 and LSTM in terms of both sensitivity and specificity, achieving 97.5% accuracy on the DRIVE dataset, 94.04% on the Kaggle dataset, 96.9% on the Eyepacs Dataset. This research work not only advances the field of automated DR detection but also provides a framework for utilizing temporal information in medical image analysis.

Agritech-Plant Disease Detection and Classification

Abstract - The Plant Disease Detection and Classification project is designed to help farmers identify crop diseases at an early stage using images of plant leaves. By applying advanced machine learning (ML) and deep learning (DL) techniques, this system aims to automatically detect and classify various plant diseases with high accuracy. Early identification of plant diseases plays a vital role in improving crop yield, reducing unnecessary pesticide usage, and encouraging sustainable farming practices. Once a disease is detected, the system not only identifies it but also provides a brief description and recommends appropriate prevention and treatment methods. To enhance performance, the project uses data augmentation and normalization techniques—helping the model handle variations in lighting, angles, and environmental conditions. The system is trained on a diverse dataset containing images of both healthy and diseased leaves from different plant species, categorized into 38 different classes. The project incorporates various machine learning algorithms such as Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Decision Trees, and Random Forests, as well as a Convolutional Neural Network (CNN) for deep learning. Among these, CNNs have shown outstanding performance due to their ability to learn spatial features from images effectively. While existing research has shown promising results, real-world challenges such as varying image quality, lighting conditions, and background clutter still affect accuracy. This project aims to overcome these issues by training on large, diverse datasets and fine-tuning model parameters. Ultimately, this project hopes to empower farmers with an intelligent tool to monitor crop health, make informed decisions, and contribute to food security through precision agriculture. Keywords: Plant Disease Detection, Automated Disease Detection System, Machine Learning, Deep Learning, Data Augmentation, Data Normalization, Support Vector Machine, KNN Classifier, Convolutional Neural Network, Decision Trees, Random Forest

Rice Leaf Diseases Classification Using CNN with Transfer Learning

Rice is a staple crop in India, significantly contributing to food security and agricultural economies. However, rice cultivation faces substantial challenges due to various diseases that affect crop yield, particularly during the Kharif season (June-October), when warm and humid conditions favor pathogen proliferation. Manual disease identification by farmers is often inaccurate due to limited expertise, leading to improper disease management and yield losses. Recent advancements in deep learning, particularly Convolutional Neural Networks (CNNs), offer promising solutions for automated disease detection. This study proposes a CNN-based model leveraging Transfer Learning (TL) with a VGG16 architecture to classify rice leaf diseases efficiently. Due to the scarcity of publicly available datasets, we curated a custom dataset comprising images collected from rice fields and online sources. Despite the dataset’s limited size, transfer learning enabled robust feature extraction and improved classification performance. The model was trained and evaluated on real-world images, demonstrating high accuracy in identifying common rice leaf diseases. Our findings highlight the potential of deep learning in precision agriculture, providing farmers with an accessible tool for early disease detection and mitigation. The proposed approach can be extended to other crops, contributing to sustainable agricultural practices

Enhancing Cassava Disease Detection: Leveraging Deep Convolutional Neural Networks and Data Augmentation for Accurate Diagnosis

Cassava, a widely cultivated staple food crop in the tropics, is frequently afflicted by various diseases that significantly decrease its yield. Cassava leaf disease diagnosis with four common cassava leaf diseases: cassava brown streak disease, cassava green mite, cassava bacterial blight, and cassava mosaic disease using state-of-the-art image classification techniques of deep CNN was implemented. In the proposed method data augmentation approaches have been applied to enlarge the dataset for better model performance. The methodology uses machine learning methods to surpass the conventional manual identification methods for recognizing affected leaf images, potentially increasing agricultural productivity through faster and more accurate disease detection that will lead to improved yield and better resource management in cassava cultivation. Evaluated the performance of various Deep Learning architectures, including LeNet, VGG-16, ResNet-50 and Efficient Net, as well as used various callback techniques. These include early stopping to prevent overfitting and model checkpointing, which saves the best-performing model, along with a learning rate reduction for fine-tuning these models and further optimizing their performance. The research, therefore, opens the possibilities of automated disease detection, greatly helping farmers in timely and accurate diagnosis for better management practices to ensure increased food security in cassava-growing regions, especially when mobile technology increases access to diagnostic tools and supports the field applicability of machine learning models. Further, deploying such models in the field has another advantage in embedding disease surveillance into the day-to-day activities of farmers, which again becomes an effective way to improve crop management strategies and responses in different agricultural contexts.

No Effect of Computer Aided Diagnosis on Colonoscopic Adenoma Detection in a Large Pragmatic Multicenter Randomized Study.

No Effect of Computer Aided Diagnosis on Colonoscopic Adenoma Detection in a Large Pragmatic Multicenter Randomized Study.

Single-View Contrastive Learning for Laryngeal Leukoplakia Classification With NBI Laryngoscopy Images.

Laryngeal cancer is the second most common upper respiratory tract cancer. Early and accurate diagnosis can improve the cure rate of patients. Laryngoscopy with NBI is a commonly used tool that can help endoscopists diagnose laryngeal diseases. However, the fine classification of laryngeal leukoplakia using NBI images is challenging for computer-aided diagnosis. In this article, we propose a single-view contrastive learning network to locate lesion regions, construct sample pairs for contrastive learning, and provide pseudo-labels to unlabeled data in order to achieve fine classification under small samples. Firstly, we pretrain the backbone network using the original NBI images. Secondly, in order to augment the number of samples for contrastive learning, we design different patch generation methods based on an attention-guided network. The original NBI images are cropped into small patches for the purpose of generating lesion-related regions and complementary samples. The pseudo-labels of these small patches are obtained by applying the pre-trained backbone network. Finally, we combine the contrastive loss function and the cross-entropy loss function for jointly training the backbone network and contrastive learning network. Our NBI dataset is classified into six categories: normal tissue, inflammatory keratosis, mild dysplasia, moderate dysplasia, severe dysplasia, and squamous cell carcinoma. Experimental results demonstrate that our model achieves an accuracy of 96.12%, which is higher than the current mainstream models. Our model also achieves high specificity and sensitivity. The code is available at https://github.com/hans-bbt/single-view-contrastive-learning.

Role of Artificial Intelligence in Improving Quality of Colonoscopy

Colorectal cancer is a common malignancy and a major health concern in Korea. Although colonoscopy is an effective tool for screening and preventing colorectal cancer through the early detection of pre-cancerous lesions, many factors influence the quality of colonoscopy, including fatigue, experience, inter-observer variation, and human error. Minimizing errors and providing consistent performance improves the quality of colonoscopy, which can lower cancer-related mortality. Advances in artificial intelligence (AI) have led to the application of computer-aided detection (CADe) and computer-aided diagnosis (CADx) of neoplastic polyps, such as adenomas, and computer-aided quality assessment (CAQ), which involves monitoring withdrawal time, assessing cecal insertion, and ensuring sufficient colonic surface observation. Many AI models have been developed, and some CADe and CADx systems have become commercially available, demonstrating their usefulness in detection of adenomas and characterization of polyps. Additionally, clinical studies on the usefulness of CAQ have been published. This innovative technology holds great potential to assist endoscopists and benefit the general population. In the future, an evaluation of the practical benefits and cost-effectiveness of applying AI models to colonoscopy in clinical practice seems necessary.