Efficient Guided Grad-CAM Tuned Patch Neural Network for Accurate Anomaly Detection in Full Images DOI Creative Commons
Rajeev Rajkumar,

D L Shanthi,

K. Manivannan

et al.

Information Technology And Control, Journal Year: 2024, Volume and Issue: 53(2), P. 355 - 371

Published: June 26, 2024

Deep learning-based anomaly detection in images has recently gained popularity as an investigative field with many global submissions. To simplify complex data analysis, researchers the deep learning subfield of machine employ Artificial Neural Networks (ANNs) hidden layers. Finding occurrences that significantly differ from generalizable to most sets is primary goal detection. Many medical imaging applications use convolutional neural networks (CNNs) examine anomalies automatically. While CNN structures are reliable feature extractors, they encounter challenges when simultaneously classifying and segmenting spots need removal scans. We suggest a separate integration system solve these issues, separated into two distinct departments: classification segmentation. Initially, network architecturesare taught independently for each abnormality, networks’ main components combined. A sharedcomponent branched structure functions all abnormalities. The final branches: onehas sub-networks, intended classify particular other various CNNs training directly on high-resolution necessitate input layer image compression, which results loss information necessary detecting guided GradCAM (GCAM) tuned patch applied full-size localization. Therefore, suggested approach merges pre-trained class activation mappings area suggestion systems construct abnormality sensors then fine-tunes picture patches, focusing abnormalities instead whole images. mammogram set was used test classifier, had 99% overall accuracy. Brain tumor integrateddetector’s ability detect abnormalities, it did so average precision 0.99.

Language: Английский

Design and assessment of improved Convolutional Neural Network based brain tumor segmentation and classification system DOI Creative Commons
Alok Singh Chauhan, Jyoti Prakash Singh, Sumit Kumar

et al.

Journal of Integrated Science and Technology, Journal Year: 2024, Volume and Issue: 12(4)

Published: Feb. 8, 2024

Deep learning techniques have recently demonstrated promising outcomes in the segmentation of brain tumors from MRI images. Due to its capability handle high-resolution images and segment entire tumor region, U-Net model is one them frequently utilized. For analysis planning treatments, accurate using multi-contrast essential. models including U-Net, PSPNet, DeepLabV3+, ResNet50 encouraging tumors. Using BraTS 2018 dataset, we compare these this research. We evaluate a variety measures, Hausdorff Distance (HD), Absolute Volume Difference (AVD), Dice Similarity Coefficient (DSC), look into how data augmentation transfer approaches affect models' performance. The findings demonstrate that 3D performed best, with DSC 0.90, HD 10.69mm, AVD 11.15%. PSPNet achieved comparable performance, 0.89, 11.37mm, 12.24%. DeepLabV3+ lower DSCs 0.85 0.83, respectively. Based on discoveries analysis, suggested for utilizing URN:NBN:sciencein.jist.2024.v12.793

Language: Английский

Citations

9
Enhancing brain tumor diagnosis: an optimized CNN hyperparameter model for improved accuracy and reliability DOI Creative Commons
Abdullah A. Asiri, Ahmad Shaf, Tariq Ali

et al.

PeerJ Computer Science, Journal Year: 2024, Volume and Issue: 10, P. e1878 - e1878

Published: March 14, 2024

Hyperparameter tuning plays a pivotal role in the accuracy and reliability of convolutional neural network (CNN) models used brain tumor diagnosis. These hyperparameters exert control over various aspects network, encompassing feature extraction, spatial resolution, non-linear mapping, convergence speed, model complexity. We propose meticulously refined CNN hyperparameter designed to optimize critical parameters, including filter number size, stride padding, pooling techniques, activation functions, learning rate, batch layers. Our approach leverages two publicly available MRI datasets for research purposes. The first dataset comprises total 7,023 human images, categorized into four classes: glioma, meningioma, no tumor, pituitary. second contains 253 images classified as “yes” “no.” delivers exceptional results, demonstrating an average 94.25% precision, recall, F1-score with 96% 1, while 87.5% F1-score, 88% 2. To affirm robustness our findings, we perform comprehensive comparison existing revealing that method consistently outperforms these approaches. By systematically fine-tuning hyperparameters, not only enhances its performance but also bolsters generalization capabilities. This optimized provides medical experts more precise efficient tool supporting their decision-making processes

Language: Английский

Citations

7
Enhanced TumorNet: Leveraging YOLOv8s and U-Net for Superior Brain Tumor Detection and Segmentation Utilizing MRI Scans DOI Creative Commons

Wisal Zafar,

Ghassan Husnain,

Abid Iqbal

et al.

Results in Engineering, Journal Year: 2024, Volume and Issue: unknown, P. 102994 - 102994

Published: Sept. 1, 2024

Language: Английский

Citations

7
A Novel Hybrid System of Detecting Brain Tumors in MRI DOI Creative Commons
Raghav Agarwal, Sagar Dhanraj Pande, Sachi Nandan Mohanty

et al.

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 118372 - 118385

Published: Jan. 1, 2023

The growth of irregular brain cells leads to a disease called tumor (BT).It is difficult predict patient's chance survival due the low rate and wide range shapes.Even though it possible manually detect cancer, doing so time-consuming runs risk producing false-positive results.This can be done via MRI, which necessary for locating cancer.It very reliably identify different illnesses from MRI images successful therapy computer-aided diagnostic system.In experiment, three openly accessible benchmark datasets were utilized.To perform feature extraction in our proposed method, CNN model was employed, followed by application five machine learning classifiers: Decision tree, Naive Bayes, Adaptive Boosting, K-nearest neighbor, support vector machine.The outcomes show that architecture with KNN classifier performs better than previous models outperforming other cutting-edge DL under various classification metrics.Finally, achieved F1-Score, precision, recall, accuracy values detection 99.58%, 99.59%, respectively.For comparison study, additional Transfer Learning are utilized.Experimental findings strength architecture, has rapidly accelerated improved classifications BTs.The designed method outperforms body existing knowledge, demonstrating quick precise classifying BTs.

Language: Английский

Citations

14
A Novel Lightweight CNN Architecture for the Diagnosis of Brain Tumors Using MR Images DOI Creative Commons
K. Rasool Reddy, Ravindra Dhuli

Diagnostics, Journal Year: 2023, Volume and Issue: 13(2), P. 312 - 312

Published: Jan. 14, 2023

Over the last few years, brain tumor-related clinical cases have increased substantially, particularly in adults, due to environmental and genetic factors. If they are unidentified early stages, there is a risk of severe medical complications, including death. So, diagnosis tumors plays vital role treatment planning improving patient’s condition. There different forms, properties, treatments tumors. Among them, manual identification classification complex, time-demanding, sensitive error. Based on these observations, we developed an automated methodology for detecting classifying using magnetic resonance (MR) imaging modality. The proposed work includes three phases: pre-processing, classification, segmentation. In started with skull-stripping process through morphological thresholding operations eliminate non-brain matters such as skin, muscle, fat, eyeballs. Then employed image data augmentation improve model accuracy by minimizing overfitting. Later phase, novel lightweight convolutional neural network (lightweight CNN) extract features from skull-free augmented MR images then classify them normal abnormal. Finally, obtained infected tumor regions segmentation phase fast-linking modified spiking cortical (FL-MSCM). this sequence operations, our framework achieved 99.58% 95.7% dice similarity coefficient (DSC). experimental results illustrate efficiency its appreciable performance compared existing techniques.

Language: Английский

Citations

13
Proposed CNN Model for Classification of Brain Tumor Disease DOI
Rahul Singh, Neha Sharma, Rupesh Gupta

et al.

Published: April 29, 2023

A brain tumor is a group of abnormal cells within the or surrounding tissues. Several variables, including family history, radiation exposure, and some genetic disorders, might increase likelihood developing tumor. The typical method for detecting tumors to perform MRI scans, which medical specialist then examines diagnosis. While time-consuming, this process fraught with possibility human error, especially when in its early stages. As result, diagnosis must be made properly as soon possible. With quick accurate identification, work aims prevent premature death, provide health resource-constrained conditions, promote patients' healthy lifestyles. CNN model created study detect cancers, dataset contains 251 scans. Because datasets are limited availability, data augmentation employed expand dataset's coverage. suggested model's outputs were evaluated using metrics Accuracy, F1-Score, Precision, Recall. In aggregate, has an accuracy 85%. deep-learning models have been demonstrated while spending no time resources effectively.

Language: Английский

Citations

10
Dual-method for semantic and instance brain tumor segmentation based on UNet and mask R-CNN using MRI DOI

Javaria Amin,

Nadia Gul,

Muhammad Irfan Sharif

et al.

Neural Computing and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 7, 2025

Language: Английский

Citations

0
Advanced brain tumor segmentation using DeepLabV3Plus with Xception encoder on a multi-class MR image dataset DOI
Shoffan Saifullah, Rafał Dreżewski, Anton Yudhana

et al.

Multimedia Tools and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 21, 2025

Language: Английский

Citations

0
Strategies to Improve the Robustness and Generalizability of Deep Learning Segmentation and Classification in Neuroimaging DOI Creative Commons
Tran Anh Tuan, Tal Zeevi, Seyedmehdi Payabvash

et al.

BioMedInformatics, Journal Year: 2025, Volume and Issue: 5(2), P. 20 - 20

Published: April 14, 2025

Artificial Intelligence (AI) and deep learning models have revolutionized diagnosis, prognostication, treatment planning by extracting complex patterns from medical images, enabling more accurate, personalized, timely clinical decisions. Despite its promise, challenges such as image heterogeneity across different centers, variability in acquisition protocols scanners, sensitivity to artifacts hinder the reliability integration of models. Addressing these issues is critical for ensuring accurate practical AI-powered neuroimaging applications. We reviewed summarized strategies improving robustness generalizability segmentation classification neuroimages. This review follows a structured protocol, comprehensively searching Google Scholar, PubMed, Scopus studies on neuroimaging, task-specific applications, model attributes. Peer-reviewed, English-language brain imaging were included. The extracted data analyzed evaluate implementation effectiveness techniques. study identifies key enhance including regularization, augmentation, transfer learning, uncertainty estimation. These approaches address major domain shifts, consistent performance diverse settings. technical this can improve their real-world practice.

Language: Английский

Citations

0
Enhanced Transfer Learning and CNN Approach for Brain Tumor Detection DOI

Kavinraj Srirangarayan Rangaraj,

Sowris Kumar Sripathy,

P. Swarnalatha

et al.

Studies in big data, Journal Year: 2025, Volume and Issue: unknown, P. 595 - 609

Published: Jan. 1, 2025

Language: Английский

Citations

0