Methods to chose for the Brain Tumor Detection

 The objective of the proposed system is to classify the brain tumor images using :-

 

• 1.       convolutional neural network (CNN)
• 2.       Support vector machine (SVM)
• 3.       Deep learning models
• 4.    Image Clustering 
• 5.    Using transfer learning

 

 Images from the dataset are downsized to reduce computation and some salt noise is added to make model robust and the dataset increases. 

The classification process undergoes following steps

 

 

1- Input

 

2-  Data Pre-processing

 

a. Importing libraries – importing libraries

b. Data augmentation – modified version of image

c. Import the augmented data

d. Convert the images to grayscale

e. Removal of noise using dilations and erosions and smoothening of images

f. Grab the largest contour.

g. Find the extreme points of the contoured image

h. Resize the image

i. Crop the images using the extreme points

j. Splitting of dataset.

 

3- Algorithms used:

 

The algorithms used in the proposed work are:

1. Convolutional Neural Network [CNN].

2. Support Vector Machine [SVM]

3. Deep Learning [ML]

 

 4- Output

The system is trained to detect the tumour in the MRI of the patient and thus predict whether the patient is suffering from tumour or no.

• CNN does the extraction using convolution layers and as the depth increases level of feature goes higher.

• SVM features are extracted depends on type of texture or pattern in the image and classes which have similar features, can be classified easily.

• Deep Learning (DL) algorithms enabled computational models consist of multiple processing layers that represent data with multiple levels of abstraction