VGG-16 Convolutional Neural Networks For Brain Tumour
Detection
ABSTRACT
Brain tumors are among the most frequent and severe types of cancer, with a life expectancy
of only a few months in the most advanced stages. Various image modalities, including CT
Scans, MRI and ultrasound images, are commonly used to assess tumors in the various
organs of the body. However, such huge amounts of data presented by these image scanning
techniques poses a difficulty in analyzing them manually and requires a lot of human effort.
Convolutional neural networks (CNNs) have been shown to outperform standard methods
when it comes to classifying brain tumor In this model, we provide a completely automated
computerized system for brain tumor categorization that employs optimal deep features
from a variety of well-known CNNs and abstraction levels. We use Transfer Learning from
ImageNet for optimized feature training. VGG-16 is architecture is used with modification to
its number of filters. The architecture is further Improved by adding a dropout layers are also
used to avoid overfitting. When tested on photos from the Kaggle Dataset, the suggested
technique gets excellent classification results.
KEY WORDS
VGG-16, Brain Tumor, Convolutional Neural Network (CNN).
INTRODUCTION
Background
A brain tumour which is a life-threatening malignancy is caused by uncontrolled and
abnormal cell partitioning. Deep learning improvements in the medical imaging area have
assisted in the identification of a number of diseases. For visual learning and image
identification, the CNN architecture is the most popular and extensively used machine
learning approach. In this report, we use a convolutional neural network (CNN) technique [1]
combined with Data Augmentation and Image Processing to analyses MRI images to
determine which images have and which do not have brain tumors.
Tumor detection using MRI data is a critical, yet time-consuming, and challenging task that
is often done by hand by medical experts. Medical image segmentation takes a long time for
1
, radiologists and other medical experts. However, precisely identifying a brain tumor takes a
long time, and there is a lot of difference among doctors.
To overcome these constraints, computer-assisted technology is critical, as the medical
field requires quick and reliable procedures to identify life-threatening diseases such as
cancer, which is the top cause of death for patients worldwide. Hence, utilizing Brain MRI
Images, we present a method for classifying MRI images into those with and without the
presence of a brain tumors utilizing a data. augmentation strategy and a convolutional neural
network model in our study.
Motivation
The effects of a brain tumor are long-term, and not only physical but psychological also. A
brain tumor is caused by a tissue anomaly that develops inside the brain or central spine,
interfering with normal brain function. A primary brain tumor affects around 700,000 people
in the most developed countries today, with another 85.000 expected to be diagnosed in
2021. Tumors of the brain can be life-threatening. Brain tumors can be fatal, have a severe
impact on quality of life, and completely transform a patient's and their family's lives. They
afflict men, women, and children of all races and nationalities without discrimination.
Additionally, manual analysis of MRI images and the risk of inaccurate result make the
situation even graver. Therefore, all these above factors motivate us to design an algorithm
to effectively and accurately predict presence of brain tumor and contribute towards society
Contribution
In this work we have used Convolutional Neural Network to segment MRI images into two
categories, those that have tumor and those that do not. The dataset consists of 253 images,
with 155 with tumor and 98 without tumor. Once the model is prepared, individual MRI
images can be segmented by taking as input. The major contributions are listed below:
➤ Convolutional neural networks work by accessing information extracted from images to
perform tasks like tumor segmentation.
➤ The network is first trained using a manually segmented dataset before being used to
segment patient images.
➤Segmented pictures can be used to predict clinical outcomes such as survival and
treatment. response
2
Detection
ABSTRACT
Brain tumors are among the most frequent and severe types of cancer, with a life expectancy
of only a few months in the most advanced stages. Various image modalities, including CT
Scans, MRI and ultrasound images, are commonly used to assess tumors in the various
organs of the body. However, such huge amounts of data presented by these image scanning
techniques poses a difficulty in analyzing them manually and requires a lot of human effort.
Convolutional neural networks (CNNs) have been shown to outperform standard methods
when it comes to classifying brain tumor In this model, we provide a completely automated
computerized system for brain tumor categorization that employs optimal deep features
from a variety of well-known CNNs and abstraction levels. We use Transfer Learning from
ImageNet for optimized feature training. VGG-16 is architecture is used with modification to
its number of filters. The architecture is further Improved by adding a dropout layers are also
used to avoid overfitting. When tested on photos from the Kaggle Dataset, the suggested
technique gets excellent classification results.
KEY WORDS
VGG-16, Brain Tumor, Convolutional Neural Network (CNN).
INTRODUCTION
Background
A brain tumour which is a life-threatening malignancy is caused by uncontrolled and
abnormal cell partitioning. Deep learning improvements in the medical imaging area have
assisted in the identification of a number of diseases. For visual learning and image
identification, the CNN architecture is the most popular and extensively used machine
learning approach. In this report, we use a convolutional neural network (CNN) technique [1]
combined with Data Augmentation and Image Processing to analyses MRI images to
determine which images have and which do not have brain tumors.
Tumor detection using MRI data is a critical, yet time-consuming, and challenging task that
is often done by hand by medical experts. Medical image segmentation takes a long time for
1
, radiologists and other medical experts. However, precisely identifying a brain tumor takes a
long time, and there is a lot of difference among doctors.
To overcome these constraints, computer-assisted technology is critical, as the medical
field requires quick and reliable procedures to identify life-threatening diseases such as
cancer, which is the top cause of death for patients worldwide. Hence, utilizing Brain MRI
Images, we present a method for classifying MRI images into those with and without the
presence of a brain tumors utilizing a data. augmentation strategy and a convolutional neural
network model in our study.
Motivation
The effects of a brain tumor are long-term, and not only physical but psychological also. A
brain tumor is caused by a tissue anomaly that develops inside the brain or central spine,
interfering with normal brain function. A primary brain tumor affects around 700,000 people
in the most developed countries today, with another 85.000 expected to be diagnosed in
2021. Tumors of the brain can be life-threatening. Brain tumors can be fatal, have a severe
impact on quality of life, and completely transform a patient's and their family's lives. They
afflict men, women, and children of all races and nationalities without discrimination.
Additionally, manual analysis of MRI images and the risk of inaccurate result make the
situation even graver. Therefore, all these above factors motivate us to design an algorithm
to effectively and accurately predict presence of brain tumor and contribute towards society
Contribution
In this work we have used Convolutional Neural Network to segment MRI images into two
categories, those that have tumor and those that do not. The dataset consists of 253 images,
with 155 with tumor and 98 without tumor. Once the model is prepared, individual MRI
images can be segmented by taking as input. The major contributions are listed below:
➤ Convolutional neural networks work by accessing information extracted from images to
perform tasks like tumor segmentation.
➤ The network is first trained using a manually segmented dataset before being used to
segment patient images.
➤Segmented pictures can be used to predict clinical outcomes such as survival and
treatment. response
2
