Classification of COVID-19 via Homology of CT-SCAN

Computers in Biology and Medicine, Год журнала: 2025, Номер 193, С. 110226 - 110226

Опубликована: Май 27, 2025

Язык: Английский

---

Hybrid Support Vector Machine‐Convolutional Neural Networks Multi‐Classification Models for Detection of Kidney Stones

International Journal of Imaging Systems and Technology, Год журнала: 2025, Номер 35(4)

Опубликована: Июнь 2, 2025

ABSTRACT The accurate and early detection of kidney stones is crucial for effective treatment patient management. This study presents a hybrid machine learning approach combining Support Vector Machines (SVM) Convolutional Neural Networks (CNN) the multi‐classification stones. proposed model leverages feature extraction capabilities CNNs with robust classification performance SVMs to improve diagnostic accuracy. methodology validated on publicly available stone dataset, experimental results demonstrate superiority over standalone CNN SVM models. Different techniques, such as enhancing contrast images, gray conversion train one channel, Gaussian filter blur noise data augmentation, SMOTE balance using 5‐fold cross‐validation prevent overfitting. Features that we extracted from were optimized classified SVM, KNN, RF. All classifiers incorporated showed high overall accuracy 98%. Among these classifiers, Hybrid CNN‐SVM outperformed other models higher test 98.49%. At same time, CNN‐KNN, CNN‐RF, achieved an 98.46%, 98.01%, 97.62%, respectively. These show effectiveness in reducing training time improving compared single

Язык: Английский

---

Classification of Thoracic X-Ray Images of COVID-19 Patients Using the Convolutional Neutral Network (CNN) Method

Journal of Computer Science, Год журнала: 2024, Номер 20(4), С. 357 - 364

Опубликована: Фев. 15, 2024

Recently, radiology modalities have been widely used to detect COVID-19. Thoracic X-rays and CT scans are the primary radiological tools utilized in diagnosis treatment of individuals with In addition, chest more accurate sensitive early COVID-19 identification. A new problem arises diagnosing results scan images by radiologists or specialists where is difficult distinguish from pneumonia caused other viruses bacteria, so misdiagnosis can occur. Many researchers worldwide developed computer-aided detection schemes based on medical image processing machine learning overcome this challenge. This research focuses development previous studies, use Convolutional Neural Network (CNN) method classify X-ray Images Patients compared model Roboflow. Image manipulation techniques applied study pseudo color program Python. employs technique uses data patients confirmed at Andalas University Hospital 2022. Based study's results, a very good CNN Specificity score 93% was obtained perfect Sensitivity value produced using Roboflow model, which 100%. However, Kappa for both methods below expected threshold 36-38%. ROC value, calculating normal patients.

Язык: Английский

---

MRI-Based Brain Tumor Classification Using Dilated Parallel Deep Convolutional Neural Network with Ensemble of Machine Learning Classifiers

Опубликована: Май 7, 2024

Brain tumors are frequently classified with high accuracy using convolutional neural networks (CNNs) and better comprehend the spatial connections among pixels in complex pictures. Due to their tiny receptive fields, majority of deep network (DCNN)-based techniques overfit unable extract global context information from more significant regions. While dilated convolution retains data resolution at output layer increases field without adding computation, stacking several convolutions has drawback producing a grid effect. To handle gridding artifacts both coarse fine features images, this research suggests parallel (PDCNN) architecture that preserves wide field. reduce complexity, initially, input images resized then grayscale transformed. Data augmentation since been used expand number datasets. Dilated PDCNN makes use lower computational overhead contributes reduction artifacts. By contrasting various dilation rates, path uses low rate (2,1,1), while local (4,2,1) for decremental even numbers tackle two paths. Using three different types MRI datasets, suggested average ensemble method performs better. The provided by Multiclass Kaggle dataset-III, Figshare dataset-II, Binary tumor identification dataset-I is 98.35%, 98.13%, 98.67%, respectively. In comparison state-of-the-art techniques, structure improves results extracting features, making it efficient.

Язык: Английский

---

Deep learned features selection algorithm: Removal operation of anomaly feature maps (RO-AFM)

Applied Soft Computing, Год журнала: 2024, Номер 162, С. 111809 - 111809

Опубликована: Май 26, 2024

Язык: Английский

---