Scribd
Upload
11 views

Batch 9_cse-A_segmentation of Brain Tumor Using Machine Learning (1)

Uploaded by

praminna395
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
11 views

Batch 9_cse-A_segmentation of Brain Tumor Using Machine Learning (1)

Uploaded by

praminna395
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 14

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

MINI PROJECT PRESENTATION

SECTION :CSE-A
BATCH : : 9

Name Roll No.

1. Allu Vinay Reddy 21BK1A0514


2. Begari Praminna 21BK1A0532
3. Erugadinla Sai Kumar 21BK1A0561 HOD CSE
PROJECT GUIDE
Mrs Raj Kumari
MINI PROJECT TITLE
SEGMENTATION OF BRAIN TUMOR
USING MACHINE LEARNING
PROBLEM STATEMENT
In recent years, brain tumors have become a significant health concern, leading to increased
mortality and morbidity rates. Early and accurate detection of brain tumors is crucial for effective
treatment planning and improving patient outcomes. Manual analysis of medical images, such as
MRI scans, is time-consuming, prone to human error, and requires expertise. There is a need for
an automated, reliable, and efficient system for brain tumor detection and segmentation.

This project aims to develop a machine learning-based system for the segmentation of brain
tumors from MRI images. The system should accurately differentiate tumor regions from normal
brain tissue, allowing for precise identification and classification. The challenge lies in the
complexity of MRI images, variations in tumor shapes and sizes, and differentiating between
healthy and diseased tissues.
PROPOSED SOLUTION
The proposed system aims to automatically segment brain tumors from MRI images
using machine learning techniques, particularly deep learning. The system will take MRI
scans as input and output segmented images where the tumor regions are distinctly
marked. This automated system will aid radiologists in identifying tumor boundaries
more efficiently, ensuring faster and more accurate diagnoses.

The objectives of the proposed system include:

● To detect brain tumors.


● To Draw a outline around tumors and Separate it from other parts of Brain.
● To provide integrity and accuracy.
● To collect a data set containing various brain images and do preprocessing on the
data set in order to train the model.
EXISTING SOLUTIONS VS PROPOSED SOLUTION
Feature Existing Solutions Proposed Solution

Model Architecture Traditional CNNs, basic U-Net Enhanced U-Net with skip connections,
residual blocks, and attention mechanisms
Preprocessing Basic normalization, skull stripping Advanced normalization, noise reduction,
and bias field correction
Data Augmentation Limited or no augmentation Extensive augmentation (rotation, scaling,
shifting)
Dataset Small, synthetic datasets, or Real-world datasets with multi-modal MRI
limited real-world datasets images
Performance on Small Prone to false negatives, misses Optimized for small tumor detection using
Tumors
small tumors advanced loss functions
Output Type 2D Segmentation Potential for both 2D and 3D segmentation
TECHNOLOGIES USED
SOFTWARE REQUIREMENTS:

Below are the software requirements for the application development:


● The required language is MATLAB
● Google Chrome, Firefox, Microsoft Edge

HARDWARE REQUIREMENTS:

● Operating System : windows


● Processor : intel i3(min)
● Ram : 4 GB(min)
● Hard Disk : 250GB(min)
UML DIAGRAMS
USE-CASE DIAGRAM
UML DIAGRAMS
CLASS DIAGRAM
UML DIAGRAMS
ACTIVITY DIAGRAM
UML DIAGRAMS
SEQUENCE DIAGRAM
PROJECT OUTPUT
BENEFITS AND LIMITATIONS
Advantages Disadvantages

Automated Tumor Detection High Computational Requirements

Reduces reliance on manual inspection, Requires significant computational power


speeding up diagnosis for large datasets

Removes noise and sharpens boundaries Poor quality images can lead to inaccurate
for better detection segmentation

Processes images faster than manual Clinicians still required to interpret and
segmentation verify results

Entirely non-invasive based on MRI or CT Small changes in parameters can


scans drastically affect outcomes
CONCLUSION
● The segmentation of brain tumors using machine learning with MATLAB has
demonstrated promising results in enhancing the accuracy and efficiency of medical
image analysis.
● The brain tumor segmentation method you have implemented primarily relies on
traditional image processing techniques such as anisotropic diffusion filtering,
thresholding, and morphological operations. These approaches can effectively
preprocess the image, reduce noise, and highlight tumor regions. The system can detect
tumors and provide useful visualizations such as bounding boxes and tumor outlines.
● Potential for early detection: Accurate segmentation can aid in early detection and
diagnosis of brain tumors, leading to improved patient outcomes.However, while
traditional techniques are valuable for basic image segmentation tasks, more advanced
methods like Convolutional Neural Networks (CNNs) have become the standard in
medical image analysis.
PROJECT GUIDE
THANK YOU

You might also like