COViT: Convolutions and ViT based Deep Learning Model for Covid19 and Viral Pneumonia Classification using X-ray Datasets DOI
Athar Shahzad Fazal, Somaiya Khan, Ali Khan

et al.

Published: Dec. 11, 2023

Artificial Intelligence based Covid19 through X-ray scans has revolutionized early diagnosis and treatment since the outbreak. There have been remarkable achievements in research of from Normal or other Pneumonia image classification using a convolutional neural network (CNN). CNN alone face problems describing low-level features can miss important information. Moreover, accurate is medical field with minimum false alarms. To answer issue, researchers this paper turned to self-attention mechanism inspired by ViT, which displayed state-of-the-art performance task. The proposed COViT method uses convolutions 3 × instead patch embedding as then alternate MLP hardswish function are added, finally, head average pooling, fully connected (FC) layer ReLU kernel L2 classifier improves accuracy. Exhaustive experiments carried out on three datasets. We only considered Viral classes for our problem. model achieved 98.98% accuracy dataset1, 99.50% dataset2 99.18% dataset3, validates efficiency shows superiority over SOTA models better than methods literature.

Language: Английский

On the Analyses of Medical Images Using Traditional Machine Learning Techniques and Convolutional Neural Networks DOI Creative Commons
Saeed Iqbal, Adnan N. Qureshi, Jianqiang Li

et al.

Archives of Computational Methods in Engineering, Journal Year: 2023, Volume and Issue: 30(5), P. 3173 - 3233

Published: April 4, 2023

Convolutional neural network (CNN) has shown dissuasive accomplishment on different areas especially Object Detection, Segmentation, Reconstruction (2D and 3D), Information Retrieval, Medical Image Registration, Multi-lingual translation, Local language Processing, Anomaly Detection video Speech Recognition. CNN is a special type of Neural Network, which compelling effective learning ability to learn features at several steps during augmentation the data. Recently, interesting inspiring ideas Deep Learning (DL) such as activation functions, hyperparameter optimization, regularization, momentum loss functions improved performance, operation execution Different internal architecture innovation representational style significantly performance. This survey focuses taxonomy deep learning, models vonvolutional network, depth width in addition components, applications current challenges learning.

Language: Английский

Citations

86
Deep Learning Based Alzheimer Disease Diagnosis: A Comprehensive Review DOI

S. Suganyadevi,

A. Shiny Pershiya,

K. Balasamy

et al.

SN Computer Science, Journal Year: 2024, Volume and Issue: 5(4)

Published: April 4, 2024

Language: Английский

Citations

8
Multi-objective optimization-driven machine learning for charging and V2G pattern for plug-in hybrid vehicles: Balancing battery aging and power management DOI

Zohre M. Mosammam,

Pouria Ahmadi, Ehsan Houshfar

et al.

Journal of Power Sources, Journal Year: 2024, Volume and Issue: 608, P. 234639 - 234639

Published: May 9, 2024

Language: Английский

Citations

6
Uncertainty-guided and cross-modality attention network for liver tumor segmentation and quantification via integrating dynamic MRI DOI Creative Commons
Jianfeng Zhao, Shuo Li

Knowledge-Based Systems, Journal Year: 2025, Volume and Issue: unknown, P. 113021 - 113021

Published: Jan. 1, 2025

Language: Английский

Citations

0
Automatic placement of simulated dental implants within CBCT images in optimum positions: a deep learning model DOI

Shahd Alotaibi,

Mona Alsomali,

Shatha Alghamdi

et al.

Medical & Biological Engineering & Computing, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 26, 2025

Language: Английский

Citations

0
EGFR gene mutation detection method using multi-path dual-layer routing attention network and multi-domain standardization GAN DOI
Pengtao Zhang,

Jinrun Guo,

Wei Zhou

et al.

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 107, P. 107765 - 107765

Published: March 13, 2025

Language: Английский

Citations

0
Tackling the small data problem in medical image classification with artificial intelligence: a systematic review DOI Creative Commons
Stefano Piffer, Leonardo Ubaldi, Sabina Tangaro

et al.

Progress in Biomedical Engineering, Journal Year: 2024, Volume and Issue: 6(3), P. 032001 - 032001

Published: May 30, 2024

Abstract Though medical imaging has seen a growing interest in AI research, training models require large amount of data. In this domain, there are limited sets data available as collecting new is either not feasible or requires burdensome resources. Researchers facing with the problem small datasets and have to apply tricks fight overfitting. 147 peer-reviewed articles were retrieved from PubMed, published English, up until 31 July 2022 assessed by two independent reviewers. We followed Preferred Reporting Items for Systematic reviews Meta-Analyse (PRISMA) guidelines paper selection 77 studies regarded eligible scope review. Adherence reporting standards was using TRIPOD statement (transparent multivariable prediction model individual prognosis diagnosis). To solve issue transfer learning technique, basic augmentation generative adversarial network applied 75%, 69% 14% cases, respectively. More than 60% authors performed binary classification given scarcity difficulty tasks. Concerning generalizability, only four explicitly stated an external validation developed carried out. Full access all code severely (unavailable more 80% studies). suboptimal (<50% adherence 13 37 items). The goal review provide comprehensive survey recent advancements dealing images samples size. Transparency improve quality publications well follow existing also supported.

Language: Английский

Citations

3
Classification of EEG signals using Machine learning algorithms DOI

S. Suganyadevi,

S. Shanmuga Priya,

B. Kiruba

et al.

2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon), Journal Year: 2022, Volume and Issue: unknown, P. 1 - 6

Published: Oct. 16, 2022

An alternative to human expert-performed manual identification is automatic detection of epilepsy using electroencephalogram (EEG) data. Automatic from EEG data need high classification performance in order eliminate false positives. A strategy for automated being proposed this work. The signals generated form the device were transformed DWT before feature extraction was carried out. Based on various statistical parameters and crossing frequency features, an algorithm dubbed GBMs fusion developed identify As added bonus, significant traits first selected a genetic algorithm. University Bonn has been used test suggested method's ability distinguish between normal ictal patterns. Experimentation shown that may increase performance. It also possible with 100% accuracy fusion.

Language: Английский

Citations

14
Smart Healthcare in IoT using Convolutional Based Cyber Physical System DOI

S. Suganyadevi,

S. Shanmuga Priya,

R. Menaha

et al.

2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon), Journal Year: 2022, Volume and Issue: unknown, P. 1 - 6

Published: Oct. 16, 2022

The intelligent Internet of Things (IoT) through infinite networking possibilities for medical data investigation is elevating the interaction between technology and healthcare society. Recent years have seen fruitful transformations in deep networks widespread use health wearables. IoT enabled by Deep Neural Networks brought about novel societal advances medicine new to study data. Despite improvements, there are still certain problems that need be addressed terms service quality. In this research, we present Grey Filter Bayesian Convolution Network (GFB-CNN), a Network-driven smart approach makes real-time Here, suggested comprehensive AI-driven eHealth architecture using GFB-CNN improve accuracy efficiency across critical quality criteria. order evaluate method's viability, large-scale Mobile HEALTH (MHEALTH) dataset analysed. From design ideas matching accuracy, overheads, time related state-of-the-art approaches, instructive example examines addresses all relevant elements method. has assessed beside methods multiplicity simulated settings. We demonstrate our successfully analyses information heart signs efficiently differentiating among good sick signals with low cost required sensing collecting.

Language: Английский

Citations

13
Internet-of-Things-Assisted Wireless Body Area Network-Enabled Biosensor Framework for Detecting Ventilator and Hospital-Acquired Pneumonia DOI

K M Abubeker,

S. Baskar,

Poonam Yadav

et al.

IEEE Sensors Journal, Journal Year: 2024, Volume and Issue: 24(7), P. 11354 - 11361

Published: Feb. 14, 2024

Ventilator-associated pneumonia (VAP) and hospital-acquired (HAP) are the leading cause of death in intensive care units (ICUs) developed two days after endotracheal intubation hospitalization or ICU admission. Hospital-acquired affects ventilated patients twice as frequently nonvented patients. Detecting volatile organic compounds (VOCs) exhaled breath can differentiate between sick healthy people. A noninvasive biosensor framework is necessary to detect VOC-induced from reducing mortality rates ICUs. To identify symptoms pneumonia, researchers have a portable wearable arrays machine learning frameworks examine VOCs air. Wireless body area networks (WBANs) allow ubiquitous devices internet-enabled monitoring for health tracking. These findings suggest that system built by biosensors Internet Things (IoT) recognize contracted hospitals ventilators. 128-core NVIDIA Jetson Nano graphics processing unit (GPU) enables seamless transmission VOC data other patient biological characteristics Amazon Web Service (AWS) IoT Core. The support vector (SVM) k-nearest neighbor (KNN) deployed Nano, SVM model outperforms KNNs terms accuracy (92.35%), sensitivity (92.67%), precision (93.38%), receiver operating characteristic (ROC, 93.11%).

Language: Английский

Citations

2