You are on page 1of 15

Brain Tumor

presented by

 Rim Chairet
 Haifa Ismail
 Ala Eddine Benhmida
 Ahmed Wajdi Abdallah
CONTENTS OF THIS
TEMPLATE

Introduction

APPROCHE
PROPOSÉE

CNN TRAINING PROCESS

04 Conclusion
Introduction

The brain tumors, are the most common and aggressive disease, leading to a very short life expectancy in
their highest grade. Thus, treatment planning is a key stage to improve the quality of life of patients.
APPROCHE PROPOSÉE

In a convolutional layer, the input image was divided into small regions and a filter or set of filters was
applied, resulting in an activation function. It is then transferred to the "ReLU" layer.
CNN TRAINING
PROCESS

Prepare the data Convolution layer Dropout layer Model Evaluation

Build the model Flattening (Flatten) Train the model


Prepare the data

In the domain of sciences and medical treatments, there are a number of available datasets that have
been validated and then published publicly

Like for example Kaggle websites from where we took our dataset
Prepare the data

In this python project, we used MRI brain images that were saved into Google drive then
treated in Google Colab environment
Build the
model

we will proceed with building our neural net- work using “Fastai” .

the trained model is called a “learner”. It is a general concept that can be taught to fit a
model.
Convolution layer

The first layer of a Convolutional Neural Network is always a convolutional layer. It


works in Image Pro-cessing. It is mostly used for multiplying two images together and yet
ensure that the resulting matrix has the same dimensions as required.
Flattening (Flatten)

It is the process of converting a matrix of the pooled feature map to a


vector
that is then passed to a fully connected layer .
Dropout layer

It is important in CNN’s training pro- cess, by preventing the overfitting problem on the
training set. Their absence, would cause the first batch of training samples to influence the
learning in a dis- proportionately way.
Train the model

Once a training procedure is done, we can observe that the


classification error rate on the validation dataset in
conjunction with a training and validation losses.
Model Evaluation

Various metrics for performance evaluation can


be used in image classification studies and these
perfor- mance evaluation metrics have become
standard in similar studies. These are precision,
specificity, sensi- tivity and accuracy.
Conclusion
● The proposed method can assist medical personnel such as surgeons and radiologists in
diagnosing brain cancer using MRI scans, particularly in the case of GBM, which
necessitates the detection of all possible malignant regions propagating

 The system has a detection rate of 81.25 percent on tumor photos and 100 percent on
non-tumor images, with a 94 percent overall accuracy .
THANKS

You might also like