Adverserial network augmentation and tabular data for a new covid-19 diagnostics approach

2022 8th International Conference on Control, Decision and Information Technologies (CoDIT), Год журнала: 2024, Номер 32, С. 2000 - 2005

Опубликована: Июль 1, 2024

Язык: Английский

---

Deep learning-based Covid-19 diagnosis: a thorough assessment with a focus on generalization capabilities

EURASIP Journal on Image and Video Processing, Год журнала: 2024, Номер 2024(1)

Опубликована: Ноя. 9, 2024

The Covid-19 pandemic has significantly spurred the development of deep learning (DL) models for pathology automatic diagnosis based on CT scan images. However, assumption about generalization proposed remains to be assessed and shown concrete clinical use. In this work, we have investigated real value widely used public datasets elaboration DL that are dedicated using scans. We collected various international from 13 countries. Different Convolutional Neural Networks (CNNs) been trained their performances carefully assessed. Two evaluations conducted: (1) an internal evaluation following a cross-validation procedure, (2) external patients coming new different sources. objective is assess capabilities considering real-world conditions: acquisition conditions, devices configurations. Three families most effective CNN selected (ResNet, DenseNet EfficientNet). These fine-tuned, evaluated within training methodology transfer learning. further customized in order create task at hand. improved performance.

Язык: Английский

---

CoSEF-DBP: Convolution scope expanding fusion network for identifying DNA-binding proteins through bilingual representations

Expert Systems with Applications, Год журнала: 2024, Номер 263, С. 125763 - 125763

Опубликована: Ноя. 10, 2024

Язык: Английский

---

Predicting the Severity of COVID-19 Pneumonia from Chest X-Ray Images: A Convolutional Neural Network Approach

EAI Endorsed Transactions on Industrial Networks and Intelligent Systems, Год журнала: 2024, Номер 12(1)

Опубликована: Ноя. 25, 2024

This study addresses significant limitations of previous works based on the Brixia and COVIDGR datasets, which primarily provided qualitative lung injury scores focused mainly detecting mild moderate cases. To bridge these critical gaps, we developed a unified comprehensive analytical framework that accurately assesses COVID-19-induced injuries across four levels: Normal, Mild, Moderate, Severe. approach’s core is meticulously curated, balanced dataset comprising 9,294 high-quality chest X-ray images. Notably, this has been made widely available to research community, fostering collaborative efforts enhancing precision classification at all severity levels. validate framework’s effectiveness, conducted an in-depth evaluation using advanced deep learning models, including VGG16, RegNet, DenseNet, MobileNet, EfficientNet, Vision Transformer (ViT), dataset. The top-performing model was further enhanced by optimizing additional fully connected layers adjusting weights, achieving outstanding sensitivity 94.38%. These results affirm accuracy reliability proposed solution demonstrate its potential for broad application in clinical practice. Our represents step forward developing AI-powered diagnostic tools, contributing timely precise diagnosis COVID-19 Furthermore, our methodological hold serve as foundation future research, paving way advancements detection respiratory diseases with higher efficiency.

Язык: Английский

---

Comprehensive analysis study of techniques in different domains for Turkish music genre classification task

Neural Computing and Applications, Год журнала: 2024, Номер 37(5), С. 3005 - 3021

Опубликована: Дек. 10, 2024

Язык: Английский

---