A Lightweight Multimodal Xception Network for Glioma Grading Using MRI Images DOI
Yu Liang, Dongjie Li, Jiaxin Ren

et al.

International Journal of Imaging Systems and Technology, Journal Year: 2024, Volume and Issue: 34(6)

Published: Nov. 1, 2024

ABSTRACT Gliomas are the most common type of primary brain tumors, classified into low‐grade gliomas (LGGs) and high‐grade (HGGs). There is a significant difference in survival rates between patients with different grades gliomas, making imaging‐based grading research hotspot. Current deep learning–based glioma algorithms face challenges, such as network complexity, low accuracy, difficulty large‐scale application. This paper proposes multimodal, lightweight Xception to address these issues. The introduces convolutional block attention modules employs dilated convolutions for spatial feature aggregation, reducing parameter count while maintaining same receptive field. By integrating channel squeeze‐and‐excitation modules, it achieves more accurate learning, alongside improvements residual connection critical information retention. Compared existing methods, proposed approach improves classification accuracy reduced count. was trained validated on 344 cases (261 HGGs 83 LGGs) tested 38 (29 9 LGGs). Experimental results demonstrate that an 92.67% AUC 0.9413 using fully connected layer classifier. features extracted improved achieved 93.42% when KNN RF classifiers. study aims provide diagnostic suggestions clinical use through simple, effective, noninvasive multimodal medical imaging method LGG/HGG grading, thereby accelerating treatment decision‐making.

Language: Английский

Alzheimer’s disease diagnosis by 3D-SEConvNeXt DOI Creative Commons

Zhongyi Hu,

Yuhang Wang, Lei Xiao

et al.

Journal Of Big Data, Journal Year: 2025, Volume and Issue: 12(1)

Published: Jan. 28, 2025

Language: Английский

Citations

0
An experimental study of U-net variants on liver segmentation from CT scans DOI Creative Commons
Akash Halder, Arup Sau, Surya Majumder

et al.

Journal of Intelligent Systems, Journal Year: 2025, Volume and Issue: 34(1)

Published: Jan. 1, 2025

Abstract The liver, a complex and important organ in the human body, is crucial to many physiological processes. For diagnosis ongoing monitoring of wide spectrum liver diseases, an accurate segmentation from medical imaging essential. importance clinical practice examined this research, along with difficulties attaining masks, particularly when working small structures precise details. This study investigates performance ten well-known U-Net models, including Vanilla U-Net, Attention V-Net, 3+, R2U-Net, U 2 {{\rm{U}}}^{2} -Net, U-Net++, Res Swin-U-Net, Trans-U-Net. These variations have become optimal approaches segmentation, each providing certain benefits addressing particular difficulties. We conducted research on computed tomography scan images three standard datasets, namely, 3DIRCADb, CHAOS, LiTS datasets. architecture has mainstay contemporary picture due its success preserving contextual information capturing fine features. structural functional characteristics that help it perform well tasks even scant annotated data are highlighted study. code additional results can be found Github https://github.com/akalder/ComparativeStudyLiverSegmentation .

Language: Английский

Citations

0
Ensemble ResDenseNet: Alzheimer’s disease staging from brain MRI using deep weighted ensemble transfer learning DOI
Md. Rabiul Hasan, A. B. M. Aowlad Hossain, Shah Muhammad Azmat Ullah

et al.

International Journal of Computers and Applications, Journal Year: 2024, Volume and Issue: 46(7), P. 539 - 554

Published: July 2, 2024

Language: Английский

Citations

2
Logistic Regression based Sentiment Analysis System: Rectify DOI
Harsh Pratap Singh,

Nagendra Singh,

Anuprita Mishra

et al.

Published: Feb. 24, 2024

Language: Английский

Citations

1
A new adoption model for quality of experience assessed by radiologists using AI medical imaging technology DOI Creative Commons
Anuchit Nirapai, Adisorn Leelasantitham

Journal of Open Innovation Technology Market and Complexity, Journal Year: 2024, Volume and Issue: 10(3), P. 100369 - 100369

Published: Aug. 25, 2024

This study introduces a new adoption model for assessing the quality of experience (QoE) radiologists using AI-based medical imaging technology. While AI has increasingly been used by screening, diagnosis, and classification images, previous investigations have primarily focused on metrics such as effectiveness, efficiency, satisfaction. research expands evaluation criteria to include user interface (UX/UI) factors, integrating them within broader QoE. QoE is conceptualized multifaceted construct influenced both human system which affect cognitive perception, including hedonic pragmatic aspects. Data were collected from 159 hospital with prior in technology systems through structured questionnaire. The data then analyzed structural equation modeling principles. findings suggest that contextual content, characteristics significantly influence turn affects utilization imaging. captures radiologists' integration throughout various stages radiological procedures, scheduling, scanning, acquisition, interpretation, reporting, communication. also highlights importance collection, storage, sharing practices compliance privacy policies.

Language: Английский

Citations

1
Blockchain Cloud Computing: Comparative study on DDoS, MITM and SQL Injection Attack DOI

Nagendra Singh,

Harsh Pratap Singh,

Anuprita Mishra

et al.

Published: Feb. 24, 2024

Language: Английский

Citations

0
Classification of Alzheimer's disease using advanced deep learning and ensemble techniques DOI

Viraj Desai,

Sucharitha Shetty,

T. Sujithra

et al.

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 18, 2024

Abstract Alzheimer's disease (AD), a principal contributor to dementia, poses critical challenge within the domain of neurology, particularly in achieving precise diagnoses and prognoses. Traditional techniques, including basic deep learning machine methods, often fall short terms classification accuracy robustness. This study capitalizes on capabilities advanced via application ensemble methodology refine image-based AD classification. Focusing Deep Convolutional Neural Networks (DCNNs) with help Mish ReLU activation functions, this research explores implementation models from Visual Geometry Group (VGG) experiments sophisticated architectures such as ResNet 50V2 101V2 along additional convolutional layers. The introduced model, which employs ResNet101V2, VGG19, customized CNN, uses soft voting judiciously assigned weights maximize efficacy achieves an 95.125%. validation our findings across various metrics, precision, recall, AUC, illustrates significant impact state-of-the-art methods accurate stages. implications contribute markedly advancement diagnostic prognostic practices, signifying considerable progression realms medical imaging neurology.

Language: Английский

Citations

0
A Lightweight Multimodal Xception Network for Glioma Grading Using MRI Images DOI
Yu Liang, Dongjie Li, Jiaxin Ren

et al.

International Journal of Imaging Systems and Technology, Journal Year: 2024, Volume and Issue: 34(6)

Published: Nov. 1, 2024

ABSTRACT Gliomas are the most common type of primary brain tumors, classified into low‐grade gliomas (LGGs) and high‐grade (HGGs). There is a significant difference in survival rates between patients with different grades gliomas, making imaging‐based grading research hotspot. Current deep learning–based glioma algorithms face challenges, such as network complexity, low accuracy, difficulty large‐scale application. This paper proposes multimodal, lightweight Xception to address these issues. The introduces convolutional block attention modules employs dilated convolutions for spatial feature aggregation, reducing parameter count while maintaining same receptive field. By integrating channel squeeze‐and‐excitation modules, it achieves more accurate learning, alongside improvements residual connection critical information retention. Compared existing methods, proposed approach improves classification accuracy reduced count. was trained validated on 344 cases (261 HGGs 83 LGGs) tested 38 (29 9 LGGs). Experimental results demonstrate that an 92.67% AUC 0.9413 using fully connected layer classifier. features extracted improved achieved 93.42% when KNN RF classifiers. study aims provide diagnostic suggestions clinical use through simple, effective, noninvasive multimodal medical imaging method LGG/HGG grading, thereby accelerating treatment decision‐making.

Language: Английский

Citations

0