Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation

Brain tumor segmentation is a vital process in medical imaging, essential for accurate diagnosis, treatment planning, and monitoring of brain tumors. Over the years, segmentation techniques have evolved from manual methods to machine learning approaches and, more recently, to deep learning technique...

Full description

Saved in:
Bibliographic Details
Main Authors: Nisha Purohit, Chandi Prasad Bhatt
Format: Article
Language:English
Published: Wolters Kluwer Medknow Publications 2025-04-01
Series:Journal of Medical Physics
Subjects:
Online Access:https://journals.lww.com/10.4103/jmp.jmp_12_25
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850035133134929920
author Nisha Purohit
Chandi Prasad Bhatt
author_facet Nisha Purohit
Chandi Prasad Bhatt
author_sort Nisha Purohit
collection DOAJ
description Brain tumor segmentation is a vital process in medical imaging, essential for accurate diagnosis, treatment planning, and monitoring of brain tumors. Over the years, segmentation techniques have evolved from manual methods to machine learning approaches and, more recently, to deep learning techniques. The advent of deep learning, particularly convolutional neural networks, has revolutionized the field, allowing for end-to-end learning and eliminating the need for manual feature extraction. This review focuses on analyzing different deep learning architectures and explores their performance when optimized using different optimizers. While deep learning techniques have significantly improved segmentation accuracy and robustness, challenges remain, particularly in terms of computational complexity, dataset imbalance, and generalization across diverse clinical settings. Achieved high segmentation accuracy in brain tumor detection, outperforming traditional methods with improved Dice scores, precision, and computational efficiency. Different machine learning and deep learning-based architectures and optimized models yielded superior performance, with Dice scores up to 0.91 and validation accuracy of 98%. Achieved dice scores of 0.84, 0.85, and 0.91 for enhancing tumor, tumor core, and whole tumor, respectively, with a mean intersection over union of 0.8665. Future research directions include exploring transfer learning, improving dataset diversity, and developing explainable artificial intelligence techniques to enhance clinical adoption. The insights from this review emphasize the need for further research to overcome existing limitations and expand the applicability of deep learning models in brain tumor segmentation.
format Article
id doaj-art-1df41f08e830480ca4e49b87fc818e63
institution DOAJ
issn 0971-6203
1998-3913
language English
publishDate 2025-04-01
publisher Wolters Kluwer Medknow Publications
record_format Article
series Journal of Medical Physics
spelling doaj-art-1df41f08e830480ca4e49b87fc818e632025-08-20T02:57:34ZengWolters Kluwer Medknow PublicationsJournal of Medical Physics0971-62031998-39132025-04-0150218519710.4103/jmp.jmp_12_25Overview of Deep Learning Algorithms and Optimizers for Brain Tumor SegmentationNisha PurohitChandi Prasad BhattBrain tumor segmentation is a vital process in medical imaging, essential for accurate diagnosis, treatment planning, and monitoring of brain tumors. Over the years, segmentation techniques have evolved from manual methods to machine learning approaches and, more recently, to deep learning techniques. The advent of deep learning, particularly convolutional neural networks, has revolutionized the field, allowing for end-to-end learning and eliminating the need for manual feature extraction. This review focuses on analyzing different deep learning architectures and explores their performance when optimized using different optimizers. While deep learning techniques have significantly improved segmentation accuracy and robustness, challenges remain, particularly in terms of computational complexity, dataset imbalance, and generalization across diverse clinical settings. Achieved high segmentation accuracy in brain tumor detection, outperforming traditional methods with improved Dice scores, precision, and computational efficiency. Different machine learning and deep learning-based architectures and optimized models yielded superior performance, with Dice scores up to 0.91 and validation accuracy of 98%. Achieved dice scores of 0.84, 0.85, and 0.91 for enhancing tumor, tumor core, and whole tumor, respectively, with a mean intersection over union of 0.8665. Future research directions include exploring transfer learning, improving dataset diversity, and developing explainable artificial intelligence techniques to enhance clinical adoption. The insights from this review emphasize the need for further research to overcome existing limitations and expand the applicability of deep learning models in brain tumor segmentation.https://journals.lww.com/10.4103/jmp.jmp_12_25brain tumor segmentationdeep learningmedical imagingoptimizer
spellingShingle Nisha Purohit
Chandi Prasad Bhatt
Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation
Journal of Medical Physics
brain tumor segmentation
deep learning
medical imaging
optimizer
title Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation
title_full Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation
title_fullStr Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation
title_full_unstemmed Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation
title_short Overview of Deep Learning Algorithms and Optimizers for Brain Tumor Segmentation
title_sort overview of deep learning algorithms and optimizers for brain tumor segmentation
topic brain tumor segmentation
deep learning
medical imaging
optimizer
url https://journals.lww.com/10.4103/jmp.jmp_12_25
work_keys_str_mv AT nishapurohit overviewofdeeplearningalgorithmsandoptimizersforbraintumorsegmentation
AT chandiprasadbhatt overviewofdeeplearningalgorithmsandoptimizersforbraintumorsegmentation