A Hybrid Deep Learning Model for Brain Tumour Classification.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Brain tumors are a leading cause of death, with early detection crucial for treatability.
• Current classification relies on invasive biopsies, often post-surgery.
• Manual radiological diagnosis can be prone to errors and delays.

Purpose of the Study:

• To develop an automated, non-invasive image classification technique for brain tumors.
• To accelerate diagnosis and improve accuracy in brain tumor detection.
• To assist radiologists in identifying glioma, meningioma, and pituitary tumors using MRI.

Main Methods:

• A hybrid Convolutional Neural Network (CNN) architecture was developed.
• Two classification methods were explored: Google-Net with Support Vector Machine (SVM) and finely tuned Google-Net with SoftMax.
• The models were trained and validated on a dataset of 3460 MRI brain images (glioma, meningioma, pituitary tumors, and normal scans).

Main Results:

• The finely tuned Google-Net with SoftMax achieved 93.1% accuracy.
• The hybrid approach combining Google-Net feature extraction with SVM classification reached a superior accuracy of 98.1%.
• This demonstrates the effectiveness of the hybrid model in classifying different brain tumor types.

Conclusions:

• Hybrid deep learning models, particularly Google-Net with SVM, offer high accuracy for non-invasive brain tumor classification.
• This technology can significantly aid radiologists, reduce diagnostic errors, and expedite treatment.
• Automated MRI-based classification holds promise for improving patient outcomes in neuro-oncology.