The document outlines the methodology for brain tumor detection from MRI images. It involves four main stages: pre-processing, skull stripping, segmentation, and feature extraction. In pre-processing, MRI images are converted to grayscale and filters are applied to remove noise. Skull stripping removes non-brain tissues. Segmentation uses Otsu's thresholding and watershed methods to separate brain regions. Feature extraction uses morphological operators to extract the tumor region by subtracting it from the original grayscale image.

1. METHODOLOGY
• Four Stages:
A. Pre-Processing
C. Segmentation
D.Feature Extraction
B. Skull Stripping

2. Pre-Processing Stage
• Four Steps:
1. MRI Input Image
2. Gray Scale Image
3. High pass filter
4. Median pass filter

3. PROCESS:
• MRI images are magnetic resonance images which
can be acquired on computer when a patient is
scanned by MRI machine.
• This input image is converted to gray scale image
which is the black and white image.

4. • The converted gray scale image consist of some
noise.Hence we use High pass filter to remove
such noise and also to sharpen and brighten the
image.
• After passing through high pass filter, image is
passed through a median filter.

5. Gray level imageColor image

6. After applying HPF After applying median
Filter

7. Skull Stripping
• Skull stripping is an important process in biomedical
analysis and it is required for the effective examination
of brain tumour from the MR images.
• It is the process of eliminating all non-brain tissues in brain
images.

8. METHODS:
• Morphology-based
• Intensity-based

9. Step 1: Get the input image.
Step 2: Apply threshold and convert into binary image.
Step 3: Find the connected regions.
Step 4: Find the largest connected region.
Step 5: Fill the largest connected region with holes.
Step 6: Get the complement of the image.
Step 7: Apply morphological dilation.
Step 8: Get the complement of the image.
Step 9: Multiply pixel-wise with the output of step 1.
ALGORITHM

10. (a) Threshold image (b) Image showing connected
regions

11. (a) Largest connected
region filled with holes
(b) Complement of the image

12. (c) Dilated image
(d) Again complemented
to get final mask

13. Intensity-Based Methods:
• Region growing
• Edge based
• Histogram based

14. Before After

15. B.Segmentation Stage
>Otsu's Thresholding method
>Watershed Segmentation

16. Otsu's Thresholding
• It is a global thesholding method used to detect
regions in the gray scale image.
• Threshold value is choosen so as to minimize the
intraclass varience of a b/w image.
• Converts the grey scale image into binary image for
different regions in the image.

18. Watershed Segmentation
• The method we follow for
segmentation is Topological
Watershed Segmentation.
• The watershed transformation
treats the image it operates
upon like a topographic map,
with the brightness of each
point representing its height,
and finds the lines that run
along the tops of ridges.

20. C.Feature Extraction-
Morphological Operators
• Non-linear operations that relate to the shape of regions in the
image are carried out over stack of images.

21. From the obtained images the one responsible for
tumor is choosen.
That image is subtracted from the original gray
scale image to get the tumor detected image.

23. Thank you!!!