BRAIN TUMOR Segmentation Using (AI) Artificial Intelligence

We propose a State-of-the-Art Artificial Intelligence (AI) based model that will introduce the Multi duality and multi-feature to the existing state of the art deep learning network U-Net to improve the accuracy of computer-aided brain tumor segmentation frameworks using MRIs.  The&

Project Title

Project Area of Specialization

Biomedical Engineering

Project Summary

We propose a State-of-the-Art Artificial Intelligence (AI) based model that will introduce the Multi duality and multi-feature to the existing state of the art deep learning network U-Net to improve the accuracy of computer-aided brain tumor segmentation frameworks using MRIs.

The brain tumor is a big problem in our country and our doctor and researcher have been working for a long time to get rid of this problem. Brain tumor segmentation is a challenging problem in medical image analysis.

It is a very time-consuming and error full task. In order to reduce the burden on physicians and improve the segmentation accuracy, the CAD (computer-aided detection) system needs to be developed.

Manual segmentation of Brain tumor imaging is still a tedious task and proper specialized examinations system and examiners are required.

Brain tissue segmentation aims to label each unit with unique brain tissue.

About 0.25 million people died due to brain tumors in 2020 according to WHO.

Project Objectives

1) To design and implement an AI and Deep learning-based Model that can predict with accuracy very closer to the human specialists on standard MRIs database.

2) The proposed model will be cost-efficient

3) The proposed model will save time and labor

Project Implementation Method

The proposed model will be implemented in Python.

First of all, a Standard Dataset of Brain MRIs will be chosen to train the model. The model will be designed, Trained, Validated, and tested. The model will make predictions of MRI segmentation using test MRIs from the standard Database and the accuracy will be calculated in terms of the ground truth. Results in terms of accuracy will be calculated and compared with existing models for improvement.

Benefits of the Project

The proposed project will help to segment the brain tumor with higher accuracy for the diagnosis of cancer. Brain tumor segmentation can predict the size, shape and age of the tumor that can help in the treatment of brain tumors.

The proposed model will be a computer-aided frame work that will save money and time.

Technical Details of Final Deliverable

The implementation and simulation of the proposed model will be carried out in Python using the Anaconda platform and Keras and PyTorch libraries. As the deep learning networks need high speed SSD drives and GPU so there is a requirement of a 512 GB SSD and Nvidia GPU.