[Paper Review] Detection and Classification of Glioblastoma Brain Tumor

Utkarsh Maurya, Appisetty Krishna Kalyan+2 more|arXiv (Cornell University)|Apr 18, 2023

Brain Tumor Detection and Classification4 citations

TL;DR

This paper proposes and evaluates UNet and Deeplabv3 deep learning models for detecting and segmenting glioblastoma brain tumors in preprocessed MRI scans. Deeplabv3 achieves higher accuracy than UNet but requires greater computational resources, demonstrating strong potential for improving early diagnosis and treatment planning in glioblastoma.

ABSTRACT

Glioblastoma brain tumors are highly malignant and often require early detection and accurate segmentation for effective treatment. We are proposing two deep learning models in this paper, namely UNet and Deeplabv3, for the detection and segmentation of glioblastoma brain tumors using preprocessed brain MRI images. The performance evaluation is done for these models in terms of accuracy and computational efficiency. Our experimental results demonstrate that both UNet and Deeplabv3 models achieve accurate detection and segmentation of glioblastoma brain tumors. However, Deeplabv3 outperforms UNet in terms of accuracy, albeit at the cost of requiring more computational resources. Our proposed models offer a promising approach for the early detection and segmentation of glioblastoma brain tumors, which can aid in effective treatment strategies. Further research can focus on optimizing the computational efficiency of the Deeplabv3 model while maintaining its high accuracy for real-world clinical applications. Overall, our approach works and contributes to the field of medical image analysis and deep learning-based approaches for brain tumor detection and segmentation. Our suggested models can have a major influence on the prognosis and treatment of people with glioblastoma, a fatal form of brain cancer. It is necessary to conduct more research to examine the practical use of these models in real-life healthcare settings.

Motivation & Objective

• To develop deep learning models for accurate detection and segmentation of glioblastoma brain tumors in MRI scans.
• To evaluate the performance of UNet and Deeplabv3 in terms of accuracy and computational efficiency.
• To support early diagnosis and effective treatment planning for glioblastoma, a highly malignant brain cancer.
• To identify trade-offs between model accuracy and computational cost for clinical deployment.

Proposed method

• The study employs UNet, a U-shaped encoder-decoder architecture, for semantic segmentation of brain tumor regions in MRI images.
• Deeplabv3, a dilated convolutional network with atrous spatial pyramid pooling, is used to capture multi-scale contextual information for improved segmentation.
• Preprocessing techniques are applied to MRI images to standardize input data and enhance model performance.
• Model training and evaluation are conducted using a dataset of preprocessed brain MRI scans with annotated tumor regions.
• Performance is quantitatively assessed using accuracy and computational efficiency metrics.
• The models are compared directly on the same dataset to evaluate relative strengths and trade-offs.

Experimental results

Research questions

• RQ1 How do UNet and Deeplabv3 perform in detecting and segmenting glioblastoma tumors in preprocessed MRI scans?
• RQ2 What is the trade-off between segmentation accuracy and computational resource requirements for these models?
• RQ3 Can Deeplabv3 achieve superior accuracy compared to UNet in glioblastoma tumor segmentation?
• RQ4 What are the practical implications of model efficiency for real-world clinical deployment?
• RQ5 How can model performance be optimized for both accuracy and computational feasibility in medical imaging?

Key findings

• Deeplabv3 outperforms UNet in terms of segmentation accuracy for glioblastoma brain tumors in MRI scans.
• UNet achieves good performance with lower computational resource demands compared to Deeplabv3.
• Both models demonstrate high accuracy in detecting and segmenting glioblastoma tumors from preprocessed MRI data.
• The computational cost of Deeplabv3 is significantly higher than that of UNet, limiting its real-time clinical use.
• The results suggest that Deeplabv3 is a promising candidate for high-accuracy tumor segmentation, pending optimization for efficiency.
• Further research is needed to adapt these models for practical deployment in clinical environments.