Automated Brain Tumor Segmentation based on Convolutional Neural Network

Brain tumor is a fatal disease that causes abnormal cell growth in the brain. Brain tumor segmentation is a difficult and tedious task for clinicians to immediately detect, extract disease regions from healthy cells in magnetic resonance images (MRI). There are variety of techniques, including

Project Title

Automated Brain Tumor Segmentation based on Convolutional Neural Network

Project Area of Specialization

Artificial Intelligence

Project Summary

Brain tumor is a fatal disease that causes abnormal cell growth in the brain. Brain tumor segmentation is a difficult and tedious task for clinicians to immediately detect, extract disease regions from healthy cells in magnetic resonance images (MRI). There are variety of techniques, including MRI that is used to detect brain tumors, and the methods that doctors use for detecting and analyzing segmentation are traditional, time consuming and depends on Eye-sight, clinical expertise  while diagnosing each individual patient, which leads to human errors. These traditional methods have less tumor segmentation accuracy. The aim of proposed method are to Improve traditional methods of detection and Segmenting brain images in order to assist doctors and to Introduce a new novel pathological method of detecting and segmenting lesion region by utilizing deep learning. Previous state-of-the-art methods were mainly assessed on the same dataset of training and test, however, the currently proposed method will be evaluated in terms of generalization on two distinct datasets such as Brats 2019, Brats 2020. The proposed method would be helpful in clinical procedures and also offers a high accuracy without a physician efforts and also reduces the rate of human error

Project Objectives

1. To develop a novel pathological model for detecting and segmenting brain tumor regions which can be trained and tested on different data sets by using CNN.
2. To improve the previous tumor segmentation accuracy state-of-the-arts methods (SOTA).

Project Implementation Method

The proposed project is a trained model based on supervised learning of convolutional neural networks, which will help in generalization to different patients and their varying tumor regions. The model will be implemented in different hospitals that do treatment of cancer and use magnetic resonance imaging in order to detect, segment, and differentiate between tumor regions and healthy cells. Furthermore, it will provide a new automatic method for tumor segmentation rather than the usage of traditional methods that doctors currently use for prescribing because they have a lot of errors, so this model will also assist doctors and clinical experts in automatically doing their work for better accuracy and fewer errors, and it will also reduce their work burden.

Benefits of the Project

• accuracy in diseases diagnosis.
• early diagnosis of cancer can save a life.

Technical Details of Final Deliverable

The final delivery of the project will be trained software that will be used for detecting and segmenting brain tumors in order to accurately get results of different patients that have brain tumors and differentiate the tumor regions from healthy cells. Furthermore, it will provide high accuracy in segmenting the tumors for the purpose of assisting the doctors in a better way that has less clinical expertise, training, and eyesight to combat the false prescribing of tumor regions in patients.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Health

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com