See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/360224770

Brain Tumour Detection Using Machine Learning

Conference Paper in Journal of Electronics and Informatics · December 2021

DOI: 10.36548/jei.2021.4.005

CITATIONS READS

11 12,585

4 authors, including:

Manav Sharma
Sharda University
1 PUBLICATION 11 CITATIONS

SEE PROFILE

All content following this page was uploaded by Manav Sharma on 27 April 2022.

The user has requested enhancement of the downloaded file.

Journal of Electronics and Informatics (ISSN: 2582-3825)
www.irojournals.com/iroei/

Brain Tumour Detection Using Machine

Learning
Manav Sharma1, Pramanshu Sharma2, Ritik Mittal3, Kamakshi
Gupta4
1,2,3
Department of Computer Science & Engineering, SET, Sharda University, India.
4
Assistant professor, Department of Computer Science & Engineering, SET, Sharda University, India.
E-mail: [email protected], [email protected],
[email protected], [email protected]

Abstract

This paper presents a model which is based on machine learning algorithms to detect brain
tumours from magnetic resonance images with high accuracy. A Convolutional Neural
Network (CNN) has been used as the algorithm for feature extraction, and segmentation. The
dataset used has been acquired from an internet website. The results show that this technique
is promising and the accuracy of 97.79% has been achieved.

Keywords: Image segmentation, CNN, Augmentation, Image classification, MRI

Introduction

This research paper has discussed the different stages of brain tumours. Cerebrum
cancer division is a significant assignment in clinical picture handling. Early determination of
mind cancers assumes a significant part in further developing therapy prospects and expands
the endurance pace of the patients. Manual division of the mind growths for disease finding,
from the enormous measure of MRI pictures produced in clinical daily practice, is a
troublesome and tedious errand. Additionally, mind growth analysis requires an acute level of
precision, where a minor mistake in decision making may result in a calamity. Consequently,
cerebrum cancer division is difficult for clinical purposes.

Among the right now proposed mind division strategies, cerebrum cancer division
techniques dependent on conventional picture handling isn't sufficiently ideal. In customary
strategy, an MRI is produced by utilizing attractive field radiation through which a two-
dimensional picture (predominantly dependent on a particular dark scale) is created and
afterwards that picture is handled and inspected by a clinical expert. This makes a chance of
Journal of Electronics and Informatics, December 2021, Volume 3, Issue 4, Pages 298-308 298
DOI: https://doi.org/10.36548/jei.2021.4.005
Received: 15.03.2022, received in revised form: 01.04.2022, accepted: 17.04.2022, published: 27.04.2022
© 2022 Inventive Research Organization. This is an open access article under the Creative Commons Attribution-NonCommercial International (CC BY-NC 4.0) License
 Manav Sharma, Pramanshu Sharma, Ritik Mittal, Kamakshi Gupta

human mistake and increments the general danger element of a clinical case which can at
times prompt lamentable conceivable outcomes. That is the reason there is a requirement for
programmed mind cancer picture division.

Current models that are based on deep learning algorithms are facing a big issue and
that is their accuracy. And accuracy plays a crucial role in health care intelligent systems,
hence for solving this issue, this model which is highly accurate has been developed.

1.1 Convolutional Neural Network

It is a deep learning algorithm that is used for image processing. This algorithm uses
an image as an input and differentiates it on different bases or features.

1.2 Advantages

• Brain tumours are detected from MRI images.

• No human intervention and hence human errors are removed.

• Human life can be saved from earlier detection of the tumour.

• Artificial intelligent systems are more reliable.

1.3 Disadvantages

• System requirements for the proper functioning of the model are high.
• Time taken to train the dataset is high.
• Highly accurate but not completely accurate.

Related Work

In [1], the Fuzzy segmentation method (FCM) was applied to separate tumour and
non-tumour regions of the brain. Wavelet features were also extracted using a multilevel
discrete wavelet transform (DWT). Finally, deep neural networks (DNNs) were incorporated
to classify brain tumours with high accuracy. This technique was compared with the methods
of KNN classifier, Linear Discriminant Analysis (LDA) and Sequential Minimum
Optimization (SMO). The accuracy rate was 96.97 in the DNN-based brain tumour
classification analysis. But the complexity was very high and the performance was very poor.
In [2], a new biomechanical model of tumour growth was presented for step-by-step analysis
of patient tumour growth. It will be applied to gliomas and individual fringed solid tumours

Journal of Electronics and Informatics, December 2021, Volume 3, Issue 4 299

BRAIN TUMOUR DETECTION USING MACHINE LEARNING

to capture a significant tumour effect. Discrete and continuous methods were combined to
model tumour growth. The proposed scheme provides the possibility of implicit segmentation
of atlas-based registry-based tumour-bearing brain images. This technique was mainly used
to segment brain tissue. But the computation time was high.

Paper [3], exploited the new multi-feature feature (Multi FD) and improved the
AdaBoost classification scheme used for brain tumour detection and segmentation. Structures
of brain tumour tissue were extracted using the Multi FD feature extraction scheme.
Advanced AdaBoost classification was used to determine whether the donated brain tissue is
tumour tissue or non-tumour. Paper [4], explained a highly complex work that the Local
Independent Projection (LIPC) based classifier was used to classify brain voxels. Also, the
path function was extracted in this method.

In [5], a new method of segmenting granular tumours using the Cellular Automata
(CA) technique was presented, which is compared with the histogram-based segmentation
method. Seed selection and volume of interest (VOI) were calculated for efficient
segmentation of brain tumours. Segmentation of tumour sections was also incorporated into
this work. Thus, the complexity was less but the accuracy was also less. In [6], a brain
tumour segmentation method, also known as multimodal brain tumour segmentation diagram
was introduced. Also, it combined different segmentation algorithms to achieve high
performance compared to the existing method. But the complexity was high.

In [7], studies on brain tumour segmentation were presented. It discussed different

segmentation methods like Area Based Segmentation, Threshold Based Segmentation, C
means Fuzzy Segmentation, Map-Based Segmentation, Margo Random Field Segmentation
(MRF), Modelling formattable, geometry deformable model, Accuracy, robustness, validity,
analysed for different types of models. In [8], Hybrid feature selection by ensemble
classification was applied to the diagnostic process of brain tumours. It used the
GANNIGMAC, decision tree, and bagging C-based wrapper approaches to get the decision
rules. It also simplified decision rules with hybrid feature selection that includes a
combination of (GANNIGMAC + MRMR C + Bagging C + Decision Tree).

Paper [9], used a Convolutional neural network for their algorithm and the data set
used was BRATS 2015. The limitation was that the computation time was high. In [10], the
algorithm used was KNN (k- nearest neighbour) but the accuracy achieved while using the
proposed method was 62.07%. [11] used a Convolutional neural network for their algorithm

ISSN: 2582-3825 300

 Manav Sharma, Pramanshu Sharma, Ritik Mittal, Kamakshi Gupta

and the data set used was BRATS 2015. The limitation was that the computation time was
high. In [12], the data set was acquired from the internet website GitHub. 2 different
algorithms ANN (Artificial neural network) and CNN (Convolutional neural network) were
used. The final result achieved was CNN is more accurate than ANN.

In [13], the dataset was acquired from sources University of Pavia, Kennedy space
centre, Indian Pives. The algorithm used was CNN and the accuracy achieved was 88.75%. In
[14], the dataset was acquired from The Cancer Imaging Archive” (TCIA). The algorithm
used was KNN and classifiers are SVM, RF, LOG, MLP and PCA. The accuracy achieved by
the proposed method was 83%. In [15], the dataset was acquired from Fig share Cheng. The
algorithm used was CNN (Convolutional neural network) and the accuracy achieved was
84.19%.

Data Set

The dataset selected for our model is acquired from the internet. It contains 253 MRI
images. In this dataset, there were different folders with a different set of images one with a
healthy brain image and the other with a brain tumour image. Then the entire dataset is
trained. Took 50 images for validation and 25 for testing. The MRI images present in the
dataset are of different dimensions. This dataset is selected because acquiring data sets from
hospitals is not a very simple task. The data set used is available on the internet website
Kaggle and the link to reach there is provided below.

https://www.kaggle.com/datasets/navoneel/brain-mri-images-for-brain-tumor-detection.

Figure 1. MRI images of the brain without tumour and with tumour

Journal of Electronics and Informatics, December 2021, Volume 3, Issue 4 301

BRAIN TUMOUR DETECTION USING MACHINE LEARNING

Table 1. Set of images

Number of images Folder directory

253 Training

25 Testing

50 Validating

• Validation set – It is the set of images that will be used during training for adjusting
the parameters.
• Testing set – It is the set of images that will not be involved until checking the final
performance of the model.

Methodologies

In this section, the basic methodologies used to develop the model are discussed.

4.1 Data acquisition

The data collected had been separated into two categories as healthy and non-healthy
ones. Further, the images are of different dimensions so they are converted into the same
dimensions of 224*224.

4.2 Pre-processing

In this stage noise removal will be done from the MRI images to increase the
accuracy of the model. MRI images often consist of noise which will increase the redundancy
and hence decrease the accuracy of the model. There is a high chance of a tumour not getting
detected because of the noise present on the borders of an MRI. Hence affects the accuracy of
the model. Pre-processing was done by scaling, reducing and converting them into grayscale.
Image Pre-processing is done to enhance the quality, look and characteristics of the image.

4.3 Image smoothing

This is an act of simplifying images while preserving important information. The aim
is to reduce unnecessary noise or detail without creating too much distortion to simplify
subsequent analyses.

ISSN: 2582-3825 302

 Manav Sharma, Pramanshu Sharma, Ritik Mittal, Kamakshi Gupta

4.4 Feature extraction

Feature extraction is the process to extract useful information out of the image. Pixel-
based feature extraction is used to extract the information and get it classified into tumour or
non-tumour.

4.5 Classification

Classification of brain MRI images from tumour to non-tumour is done using the
Convolutional neural network. The classifier used for classification is done by CNN itself. It
is highly accurate while dealing with image related datasets. It is used for classifying tumour
or non-tumour MRIs.

Proposed Work

Figure 2. Flow chart of the proposed work

The technique used for making this model work is CNN (Convolutional Neural
Network). The different stages by which CNN is applied on the dataset are

1. At first, the required packages are imported.

2. Then the folder where the dataset is stored is imported.

3. Image reading is done after that it is labelled (such as 0 non-tumour and 1 for
tumour image) and then the images are stored in the data frame.

4. Then the size of the image is changed to 224*224 and the shape of the image.

5. Image normalization is completed.

Journal of Electronics and Informatics, December 2021, Volume 3, Issue 4 303

BRAIN TUMOUR DETECTION USING MACHINE LEARNING

6. The data was split into 3 parts –

• Training

• Validation

• Test

7. Sequential model creation is completed.

8. The model compilation is completed.

9. Then the model is applied to the training dataset and a validation dataset was
used for evaluating the model.

10. Then the accuracy of the model is evaluated using the test images.

11. The loss graph and the accuracy graph are plotted.

The entire implementation is done using python 3.8 and are executed on Google
Colaboratory and the model is stored in Google Drive.

Results

When the model is applied to the testing data set for 10 epochs, a validation accuracy
of 82.86% is obtained and the validation loss is also less.

Figure 3. Model loss

ISSN: 2582-3825 304

 Manav Sharma, Pramanshu Sharma, Ritik Mittal, Kamakshi Gupta

As seen in figure 3, when the model is applied to the validation, then a high loss is
obtained but once applied to the testing set, the loss gradually decreases with the increasing
number of epochs.

Figure 4. Model accuracy

The accuracy of the convolutional neural network model achieved after applying it to
the testing set was 97.79%. with a very minimal loss with increasing epochs. The difference
in model accuracy can be seen between the validation dataset and the training dataset in
Figure 4.

Figure 5. Experimental results

By the use of figure 5, it can be confirmed that the accuracy increases with the
increase in the number of epochs and there is a decrease in loss of the testing set.

Journal of Electronics and Informatics, December 2021, Volume 3, Issue 4 305

BRAIN TUMOUR DETECTION USING MACHINE LEARNING

Conclusion

The aim of this paper is to create a model with high accuracy to determine brain
tumours from the MRI images. The dataset used consists of 253 brain MRI images and was
sufficient to check the performance of the model. The model is based on the machine learning
algorithm CNN (Convolutional Neural Network). It helps to predict just by reducing and
resizing the image without losing any important information that will be used for predicting.
The created model achieves an accuracy of 97.79% when applied to the training set and an
accuracy of 82.86% when applied to the validation set. The loss gradually starts decreasing
with the increase in the number of epochs. The model loss is very less when applied to the
training set whereas it is high when applied to the validation set. In future, different datasets
would be applied to this model, to further increase the overall accuracy.

References

[1] H. Mohsen, E.-S. A. El-Dahshan, E.-S. M. El-Horbaty and A.-B. M. Salem,

“Classification using deep learning neural networks for brain tumors,” Future
Computing and Informatics Journal, pp. 68-71, 2018.
[2] S. Bauer, C. May, D. Dionysiou, G. Stamatakos, P. Buchler and M. Reyes, “Multiscale
modeling for Image Analysis of Brain Tumor Studies,” IEEE Transactions on
Biomedical Engineering, vol. 59, no. 1, pp. 25-29, 2012.
[3] Islam, S. M. Reza and K. M. Iftekharuddin, “Multifractal texture estimation for
detection and segmentation of brain tumors,” IEEE Transactions on Biomedical
Engineering, pp. 3204-3215, 2013.
[4] M. Huang, W. Yang, Y. Wu, J. Jiang, W. Chen and Q. Feng, “Brain tumor
segmentation based on local independent projection-based classification,” IEEE
Transactions on Biomedical Engineering, pp. 2633-2645, 2014.
[5] Hamamci, N. Kucuk, K. Karaman, K. Engin and G. Unal, “Tumor-cut: Segmentation of
brain tumors on contrast enhanced MR images for radiosurgery applications,” IEEE
Transactions on Medical Imaging, pp. 790-804, 2012.
[6] H. B. Menze, A. Jakab, S. Bauer, K. Farahani and K. Justin, “The Multimodal Brain
Tumor Image Segmentation Benchmark,” IEEE Transactions on Medical Imaging, pp.
1993-2024, 2014.

ISSN: 2582-3825 306

 Manav Sharma, Pramanshu Sharma, Ritik Mittal, Kamakshi Gupta

[7] Jin Liu, Min Li, Jianxin Wang, Fangxiang Wu, Tianming Liu and Yi Pan, “A survey of
MRI-based brain tumor segmentation methods,” Tsinghua Science and Technology, pp.
578-595, 2014.
[8] S. Huda, J. Yearwood, H. F. Jelinek, M. M. Hassan, G. Fortino and M. Buckland, “A
hybrid feature selection with ensemble classification for Imbalanced Healthcare Data: A
case study for brain tumor diagnosis,” IEEE Access, pp. 9145-9154, 2016.
[9] J. Seetha and S. S. Raja, “Brain tumor classification using Convolutional Neural
Networks,” Biomedical and Pharmacology Journal, pp. 1457-1461, 2018.
[10] P. Meena and V. Janani, “IMAGE SEGMENTATION FOR TUMOR DETECTION,”
A Monthly Journal of Computer Science and Information Technology, pp. 244-248,
2013.
[11] M. Aarthilakshmi, S. Meenakshi, A. P. Pushkala, N. B. Prakash and V. R.
Ramalakshmi, “Brain Tumor Detection Using Machine,” INTERNATIONAL
JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH, pp. 1976-1979, 2020.
[12] P. Gokila Brindha, M. Kavinraj, P. Manivasakam and P. Prasanth, “Brain tumor
detection from MRI images using Deep Learning Techniques,” IOP Conference Series:
Materials Science and Engineering, p. 012115, 2021.
[13] Y. Chen, H. Jiang, C. Li, X. Jia and P. Ghamisi, “Deep feature extraction and
classification of hyperspectral images based on Convolutional Neural Networks,” IEEE
Transactions on Geoscience and Remote Sensing, pp. 6232-6251, 2016.
[14] E. Alberts, G. Tetteh, S. Trebeschi, M. Bieth, A. Valentinitsch, B. Wiestler, C. Zimmer
and B. H. Menze, “Multi-modal image classification using low-dimensional texture
features for genomic brain tumour recognition,” Graphs in Biomedical Image Analysis,
Computational Anatomy and Imaging Genetics, pp. 201-209, 2017.
[15] N. Abiwinanda, M. Hanif, S. T. Hesaputra, A. Handayani and T. R. Mengko, “Brain
tumour classification using Convolutional Neural Network,” IFMBE Proceedings, pp.
183-189, 2018.

Author’s biography

Manav Sharma is going to receive a Bachelor’s degree in Computer engineering with a

specialization in Data Science in the year 2022 from Sharda University. His research area
includes machine learning, deep learning and data analytics.

Journal of Electronics and Informatics, December 2021, Volume 3, Issue 4 307

 BRAIN TUMOUR DETECTION USING MACHINE LEARNING

Pramanshu Sharma is going to receive a Bachelor’s degree in Computer engineering with a

specialization in Data Science in the year 2022 from Sharda University. His research area
includes machine learning, deep learning and data analytics.

Ritik Mittal is going to receive a Bachelor’s degree in Computer engineering with a

specialization in Data Science in the year 2022 from Sharda University. His research area
includes machine learning, deep learning and data analytics.

Kamakshi Gupta received a bachelor’s degree in Computer engineering from AKTU in

2012 and master’s degree from Guru Gobind Singh Indraprastha University in the year 2019.
She is currently working as an assistant professor at the department of computer science at
Sharda University. Her area of research includes Cryptography and Networking.

ISSN: 2582-3825 308

View publication stats