Transformers for Neuroimage Segmentation: Scoping Review (Preprint) DOI
Maya Iratni,

A. Sheik Abdullah,

Mariam Aldhaheri

et al.

Published: Feb. 25, 2024

BACKGROUND Neuroimaging segmentation is increasingly important for diagnosing and planning treatments neurological diseases. Manual time-consuming, apart from being prone to human error variability. Transformers are a promising deep learning approach automated medical image segmentation. OBJECTIVE This scoping review will synthesize current literature assess the use of various transformer models neuroimaging METHODS A systematic search in major databases, including Scopus, IEEE Xplore, PubMed, ACM Digital Library, was carried out studies applying transformers problems 2019 through 2023. The inclusion criteria allow only peer-reviewed journal papers conference focused on transformer-based brain imaging data. Excluded dealing with nonneuroimaging data or raw signals electroencephalogram Data extraction performed identify key study details, modalities, datasets, conditions, models, evaluation metrics. Results were synthesized using narrative approach. RESULTS Of 1246 publications identified, 67 (5.38%) met criteria. Half all included published 2022, more than two-thirds used segmenting tumors. most common modality magnetic resonance (n=59, 88.06%), while frequently dataset tumor (n=39, 58.21%). 3D (n=42, 62.69%) prevalent their 2D counterparts. developed those hybrid convolutional neural network-transformer architectures (n=57, 85.07%), where vision type (n=37, 55.22%). frequent metric Dice score (n=63, 94.03%). Studies generally reported increased accuracy ability model both local global features images. CONCLUSIONS represents recent increase adoption segmentation, particularly detection. Currently, achieve state-of-the-art performances benchmark datasets over standalone models. Nevertheless, applicability remains highly limited by high computational costs potential overfitting small datasets. heavy reliance field hints at diverse set validate variety Further research needed define optimal training methods clinical applications. Continuing development may make fast, accurate, reliable which could lead improved tools evaluating disorders.

Language: Английский

Transformers for Neuroimage Segmentation: Scoping Review DOI Creative Commons
Maya Iratni, Ahmad Shahidan Abdullah,

Mariam Aldhaheri

et al.

Journal of Medical Internet Research, Journal Year: 2025, Volume and Issue: 27, P. e57723 - e57723

Published: Jan. 29, 2025

Background Neuroimaging segmentation is increasingly important for diagnosing and planning treatments neurological diseases. Manual time-consuming, apart from being prone to human error variability. Transformers are a promising deep learning approach automated medical image segmentation. Objective This scoping review will synthesize current literature assess the use of various transformer models neuroimaging Methods A systematic search in major databases, including Scopus, IEEE Xplore, PubMed, ACM Digital Library, was carried out studies applying transformers problems 2019 through 2023. The inclusion criteria allow only peer-reviewed journal papers conference focused on transformer-based brain imaging data. Excluded dealing with nonneuroimaging data or raw signals electroencephalogram Data extraction performed identify key study details, modalities, datasets, conditions, models, evaluation metrics. Results were synthesized using narrative approach. Of 1246 publications identified, 67 (5.38%) met criteria. Half all included published 2022, more than two-thirds used segmenting tumors. most common modality magnetic resonance (n=59, 88.06%), while frequently dataset tumor (n=39, 58.21%). 3D (n=42, 62.69%) prevalent their 2D counterparts. developed those hybrid convolutional neural network-transformer architectures (n=57, 85.07%), where vision type (n=37, 55.22%). frequent metric Dice score (n=63, 94.03%). Studies generally reported increased accuracy ability model both local global features images. Conclusions represents recent increase adoption segmentation, particularly detection. Currently, achieve state-of-the-art performances benchmark datasets over standalone models. Nevertheless, applicability remains highly limited by high computational costs potential overfitting small datasets. heavy reliance field hints at diverse set validate variety Further research needed define optimal training methods clinical applications. Continuing development may make fast, accurate, reliable which could lead improved tools evaluating disorders.

Language: Английский

Citations

2
Advancements in deep learning techniques for brain tumor segmentation: A survey DOI Creative Commons

C. Umarani,

Shantappa G. Gollagi, Shridhar Allagi

et al.

Informatics in Medicine Unlocked, Journal Year: 2024, Volume and Issue: 50, P. 101576 - 101576

Published: Jan. 1, 2024

Citations

4
Exploring Deep Learning Techniques for MRI Brain Tumor Image Segmentation: A Survey DOI

R Rohith,

Muthu Dayalan,

M Meena

et al.

2022 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI), Journal Year: 2024, Volume and Issue: unknown, P. 1 - 5

Published: May 9, 2024

Language: Английский

Citations

1
Dual Vision Transformer-DSUNET With Feature Fusion for Brain Tumor Segmentation DOI Creative Commons
Mohammed Zakariah, Muna Al‐Razgan, Taha Alfakih

et al.

Heliyon, Journal Year: 2024, Volume and Issue: 10(18), P. e37804 - e37804

Published: Sept. 1, 2024

Language: Английский

Citations

1
On Undesired Emergent Behaviors in Compound Prostate Cancer Detection Systems DOI

Erlend Sortland Rolfsnes,

Philip Thangngat,

Trygve Eftestøl

et al.

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 73 - 82

Published: Oct. 8, 2024

Language: Английский

Citations

1
Impact of uncertainty quantification through conformal prediction on volume assessment from deep learning-based MRI prostate segmentation DOI Creative Commons

Marie Gade,

Kevin Mekhaphan Nguyen,

Sol Gedde

et al.

Insights into Imaging, Journal Year: 2024, Volume and Issue: 15(1)

Published: Nov. 29, 2024

Abstract Objectives To estimate the uncertainty of a deep learning (DL)-based prostate segmentation algorithm through conformal prediction (CP) and to assess its effect on calculation volume (PV) in patients at risk cancer (PC). Methods Three-hundred seventy-seven multi-center 3-Tesla axial T2-weighted exams from biopsied males (66.64 $$\pm$$ ± 7.47 years) PC were retrospectively included study. Assessment PV based PI-RADS 2.1 ellipsoid formula ( $${{{\rm{PV}}}}_{{ref}}$$ PV r e f ) was available for patients. Prostate segmentations obtained DL model used calculate $${{{\rm{PV}}}}_{{DL}}$$ D L ). CP applied confidence level 85% flag unreliable pixel model. Subsequently, $${{{\rm{PV}}}}_{{CP}}$$ C P calculated when disregarding uncertain segmentations. Agreement between evaluated against reference standard . Intraclass correlation coefficient (ICC) Bland–Altman plots agreement. The relative difference (RVD) evaluate accuracy, Wilcoxon Signed-Rank Test statistical differences. A p -value < 0.05 considered statistically significant. Results Conformal significantly reduced RVD compared (RVD = − 2.81 8.85 −8.01 11.50). showed larger agreement than using (mean (95% limits agreement) : 1.27 mL (− 13.64; 16.17 mL) 6.07 14.29; 26.42 mL)), with an excellent ICC 0.97 CI: 0.98)). Conclusion Uncertainty quantification increases accuracy reliability DL-based assessment PC. Critical relevance statement can predictions MRI desired level, increasing safety cancer. Key Points user-defined level. Deep shows high volumetric assessment. automatic ellipsoid-derived prediction. Graphical

Language: Английский

Citations

1
Efficient Brain Tumor Detection Using VGG-16 and ResNet50 Transfer Learning Models DOI

T. Kujani,

S. Alex David,

T. Sathya

et al.

Advances in intelligent systems and computing, Journal Year: 2023, Volume and Issue: unknown, P. 455 - 467

Published: Jan. 1, 2023

Language: Английский

Citations

2
Review of deep learning-driven MRI brain tumor detection and segmentation methods DOI Open Access
Rong Zhang, Hongliang Luo, Weijie Chen

et al.

Advances in Computer Signals and Systems, Journal Year: 2023, Volume and Issue: 7(8), P. 17 - 28

Published: Sept. 1, 2023

The application of deep learning in the field medical imaging has become increasingly widespread, greatly promoting advancement and development Magnetic Resonance Imaging (MRI) brain tumor detection segmentation techniques. Therefore, a comprehensive review learning-based methods for MRI was conducted. This introduces basic concepts tumors segmentation, discusses specific applications typical analyzes compares performance advantages disadvantages different methods. Additionally, representative tu-mor dataset (BraTS) its evaluation metrics are introduced, upon which various on BraTS 2019-2022 is compared. Lastly, challenges future trends summarized anticipated.

Language: Английский

Citations

2
Dual Vision Transformer-DSUNET With Feature Fusion for Brain Tumor Segmentation DOI Creative Commons
Mohammed Zakariah, Muna Al‐Razgan, Taha Alfakih

et al.

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

Published: March 15, 2024

Abstract Brain tumors are one of the leading causes cancer death; screening early is best strategy to diagnose and treat brain tumors. Magnetic Resonance Imaging (MRI) extensively utilized for tumor diagnosis; nevertheless, achieving improved accuracy performance, which a critical challenge in most previously reported automated medical diagnostics, difficult problem. The study introduces Dual Vision Transformer-DSUNET model, incorporates feature fusion techniques provide precise efficient differentiation between other regions by leveraging multi-modal MRI data. impetus this arises from necessity automating segmentation process imaging, component realms diagnosis therapy strategy. To tackle issue BRATS 2020 dataset employed, an This encompasses images, including T1-weighted, T2-weighted, T1Gd (contrast-enhanced), FLAIR modalities. proposed model dual vision idea comprehensively capture heterogeneous properties across several imaging Moreover, utilization implemented augment amalgamation data originating modalities, hence enhancing dependability segmentation. evaluation model's performance conducted employing Dice Coefficient as prevalent metric quantifying accuracy. results obtained experiment exhibit remarkable with values 91.47% enhanced tumors, 92.38% core 90.88% edema. cumulative score entirety classes 91.29%. In addition, has high level accuracy, roughly 99.93%, underscores its durability efficacy task segmenting Experimental findings demonstrate integrity suggested architecture, quickly detection many diseases.

Language: Английский

Citations

0
OPTIMIZED U-NET SEGMENTATION MODEL AND DEEP MAXOUT CLASSIFIER FOR BRAIN TUMOR CLASSIFICATION DOI

Subha Thomas,

R. Sudarmani

Biomedical Engineering Applications Basis and Communications, Journal Year: 2024, Volume and Issue: 36(05)

Published: July 20, 2024

The most serious nervous system ailment, a brain tumor impairs one’s health seriously and ultimately results in death. MRI, one of the frequently used medical imaging modalities for tumors, has emerged as main diagnostic treatment study tumors. It was challenging to segment classify many kinds swarm intelligence approach potential more efficiently effectively tackle number issues. Therefore, this work develops novel model classification tumors that includes various phases. Primarily, input image is preprocessed via proposed median filtering aids removing noises. Subsequently, segmentation done optimal U-Net. For precise segmentation, weights are tuned optimally by battle royale optimization with Bernoulli randomization (BROBR) algorithm. Then, features like local Gabor XOR pattern (PLGXP), texton features, gray level co-occurrence matrix (GLCM), correlation extracted. Finally, BTC using deep maxout network (DMO) provides final output on absence or presence tumor.

Language: Английский

Citations

0