Detection of brain tumor in three D MRI
This project is about the detection of brain tumors found in the human brain with a maximum accuracy of results with minimum cost(time, resource consumption). Using the latest advanced algorithms of machine learning and computer vision with sophisticated methods of digital image processing applied o
2025-06-28 16:32:01 - Adil Khan
Detection of brain tumor in three D MRI
Project Area of Specialization Mechanical EngineeringProject SummaryThis project is about the detection of brain tumors found in the human brain with a maximum accuracy of results with minimum cost(time, resource consumption). Using the latest advanced algorithms of machine learning and computer vision with sophisticated methods of digital image processing applied on a 3D MRI scan of patients using the fastest and one of the powerful software tools in the world MATLAB we expect to achieve an advance and state of the art software that will be able to identify tumors including its size, category(either it is T1, T2, T3 or T4), which will not only helps neurologist and doctors in early detection of brain tumor but also start immediate treatment as well. This project is majorly research-based as well as some development phase is added for the convenience of doctors.
The project includes series of image enhancers for the application of digital image processing followed by the state of the art algorithms of Computer Vision as well as machine learning, These algorithms will serve to create the most advanced and state of the art software for the detection of brain tumor.
Project ObjectivesBrain tumor detection and diagnosis of a tumor before and after surgery depend on the decision of neurosurgeons for image evaluation which takes months to complete along with the possibility of human error leading to the deduction of a wrong decision. The software will be developed for real-life applications. Our project aims to present an algorithm that automatically detects the presence of a brain tumor in 3D MRI and Introduction presents the output to the doctors for their comprehensive analyses on a brain tumor. In the Application an MRI to be checked will be inputted, after training of the models the inputted image will be tested and the suspected brain tumor patient’s 3D MRI is classified as healthy or brain tumor affected. Machine Learning is doing wonders in the medical industry these days. Our application will give a method which will be fast, robust, and reliable for the detection of brain tumor. “The project is based on building a real-life application that can classify if the subject has a tumor or not before and after the surgery based on 3D brain MRI images with an acceptable accuracy for medical image processing application.”
Project Implementation Method1. Acquiring Image: In the first step is to acquire a 3D MR image scan of the patient’s brain from the dataset of images using MATLAB. Then the image is loaded in the system using the built-in function of MATLAB to perform the operations mentioned below with which we receive information regarding the presence of tumors so that we could take proper measures later.
2. Selection of Random slice: A random slice from the 3D MR image which is in wrl format is then converted into jpg format. Then this 3D image is displayed on the 2D screen which is the concept of orthography in which different operations are performed on a 3D object which in this case is a 3D image but to a viewer, the object is shown as 2D.
3. Pre-Processing: In preprocessing the 3D MR image is sliced using the built-in function of MATLAB then converted into a 3D volume so that the performance of the below-mentioned operations becomes easier which in turn reduces the complexity of operations and in turn convert structure in volume shape. Once the 3D volume is acquired then it must be enhanced using filters like when the grey-scale is applied or contrast is adjusted on that part to enhance the 3D volume and remove any noise associated with it to get better clarity of that 3D volume to make it ready for segmentation.
4. Segmentation: In this operation, we solve the problem of differentiation between homogeneous(normal) and heterogeneous(cancerous) cells that the 3D MR image contains. To properly distinguish between normal brain cells and tumor cells we will apply segmentation. In segmentation, we propose the use of a watershed algorithm. The watershed algorithm is one of the algorithms used in this project to achieve higher accuracy. Its function is to treats the image it operates upon like a topographic map. A topographic map is required as it helps with the brightness of each point representing its height and finds the lines that run along the tops of ridges that would distinguish between normal and cancerous brain cells.
5. Self-Organized Mapping: SOM (self-organized mapping) is a machine learning algorithm that serves as the layer algorithm which would help in increasing the accuracy. Its function would be to identify the attributes of affected(cancerous) cells in the 3D image applied after the segmentation phase with the addition of the proposed method of Dot Product to find the Euclidean distance between points of tumors and providing the values depending upon their sizes and shapes.
6. Feature Extraction: The Euclidean distance received in the previous phase of the application is used to extract the useful part of the image so that it can be used as input for the classifier. In this operation’s different features of cancerous cells like their size, orientation, will be extracted.
7. Classification: The training model uses the data acquired from feature extraction and classifies possible attributes and its performance will be shown by confusion matrix in the training process.
“3D MRI images using a Brats 2020 dataset using SOM by introducing the concept of dot product for tumors to overcome the issue of detection of brain tumors. with this methodology, we expect to reduce training time and resource consumptions while with a little or no compromise on accuracy.”
The detection process of brain tumors in 3D MRI will be easy and accurate.
A brain tumor is an uncontrolled division of cells in your brain. The brain tumor is one of the foremost dangerous diseases that need correct detection methodology. The treatment of the tumors is incredibly necessary for the patient due to their rapid spread. Purposes:
• Facilitate doctors.
• Platform to get precise results.
Generally, the brain tumor has two types. A primary brain tumor that originates within the brain and infrequently unfold outside of it. The other is a Metastatic Brain Tumor which begins in some other part of the body and spreads to the brain. Magnetic resonance imaging is a medical imaging technique used in radiology to form pictures of the anatomy and the physiological processes of the body.
Machine learning has made so much progress in healthcare. Machine learning lends itself more to some applications than to others. Algorithms may provide an actual benefit to disciplines with reproducible or standard processes. Those with large image datasets are also viable targets, such as radiology, cardiology, and pathology. Machine learning can be trained to look at images, detect irregularities, and point to areas that require attention, thus increasing the accuracy of all these operations. In the Long-term, the family practitioner may benefit from machine learning. Machine learning provides an unbiased opinion to enhance performance, reliability, and precision.
Use of ML techniques in the health industry to analyze the data and funnel it back to doctors in real-time to assist with clinical decision-making. At the same time as a doctor examines a patient and inserts data into the application, there is machine learning at the back end that examines the data and guides the doctor about useful information to make a decision, schedule a test, or propose a preventive screen. We shall also use machine learning algorithms to accomplish this project
Final Deliverable of the Project Hardware SystemCore Industry ITOther Industries Medical Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 70000 | |||
| 3-D glasses | Equipment | 3 | 1000 | 3000 |
| Samsung 6 Series 3D LED EH6030 Features | Equipment | 1 | 56000 | 56000 |
| HP deskjet 2620 | Equipment | 1 | 11000 | 11000 |