BRAIN TUMOUR DETECTION AND CLASSIFICATION

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 11 Issue: 02 | Feb 2024 www.irjet.net p-ISSN: 2395-0072
© 2024, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 77
BRAIN TUMOUR DETECTION AND
CLASSIFICATION
Aryan Galande1, Suraj Mohite2, Pavan Thorat3, Prof. Rajani Jadhav4
1Aryan Galande: Student, Dept. of Computer Engineering, PICT Pune.
2Suraj Mohite: Student, Dept. of Computer Engineering, PICT Pune.
3Pavan Thorat: Student, Dept. of Computer Engineering, PICT Pune.
4Prof. Rajani Jadhav: Professor, Dept. of Computer Engineering, PICT Pune.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Cancer is one of the largest health problems
the world faces today, and early detection is key to better
patient outcomes. Traditional tumour detection methods
sometimes involve intrusive procedures and have
disadvantages. The integration of Artificial Intelligence (AI)
and Machine Learning (ML) has led to breakthrough
developments in the field of medical diagnostics in recent
years, particularly in the identification of malignant
tumours. This research provides a thorough analysis of the
application of AI and ML techniques in the early diagnosis
and detection of cancer. Keywords—Cancer Detection,
Tumor Detection, Artificial Intelligence, Machine Learning,
Medical Imaging, Radiology, Healthcare, Early Diagnosis.
Key Words: Cancer Detection, Tumour Detection, Artificial
Intelligence, Machine Learning, Medical Imaging ,
Radiology, Healthcare, Early Diagnosis .
1.INTRODUCTION
A crucial area of medical research and clinical practice is
the detection of cancer in the brain. This field focuses on
the identification and diagnosis of tumours that originate
inside the central nervous system, which includes the
brain and spinal cord. Brain tumours, commonly referred
to as brain cancer, can be malignant (cancerous) or benign
(noncancerous). Early brain cancer detection is essential
for successful treatment and better patient outcomes. The
following are some important details about brain cancer
detection: Different Brain Tumour Types: Brain tumours
are categorized according to their origin, location, and
level of malignancy. Whereas secondary brain tumours
originate from cancer that has progressed to the brain
from other parts of the body, primary brain tumours
originate in the brain or spinal cord. Brain cancer
symptoms can vary greatly and include weakness,
eyesight issues, altered behaviour or cognitive function,
chronic headaches, and seizures. Due to the mild and non-
specific nature of these symptoms, early identification is
frequently difficult. Diagnostic Tools: A mix of diagnostic
tools and medical imaging techniques, such as magnetic
resonance imaging (MRI), computed tomography (CT),
and positron emission tomography (PET) scans, are
commonly used to detect brain cancer. The presence,
location, and size of brain tumours can be seen with the
aid of these imaging investigations. Biopsy: To obtain a
conclusive diagnosis, a biopsy often entails taking a tiny
sample of the tumour tissue to be examined under a
microscope. This aids in identifying the exact type of
tumour and whether it is malignant. Blood Tests: New
studies are investigating the possibilities of liquid
biopsies, which entail testing blood samples for the
presence of DNA mutations linked to brain cancer or
tumor-specific markers. Developments in Imaging: The
capacity to accurately map brain tumours and evaluate
their effects on surrounding brain structures has been
enhanced by developments in medical imaging technology,
including diffusion tensor imaging and functional
magnetic resonance imaging. Options for Treatment:
Surgery, radiation therapy, chemotherapy,
immunotherapy, targeted therapy, or a mix of these may
be used to treat brain cancer. Options for less invasive
treatment are frequently made possible by early detection.
Research and Innovation: Developing novel treatments for
brain tumours and enhancing the precision and
noninvasiveness of diagnostic methods are the two main
objectives of continuous efforts in the field of brain cancer
detection. Early Diagnosis and Prognosis: Early detection
of brain cancer is critical to improving patient quality of
life and raising the chance of a successful treatment plan.
Moreover, accurate prognosis is necessary to tailor
treatment plans to individual patients.
1.1 Literature Survey
In a 2021 study by Q.D. Buchlak et al., the applications of
machine learning in neuroimaging for glioma detection
and classification were explored. This research offers
valuable insights into the use of AI specifically for glioma
diagnosis, a common and aggressive type of brain tumor.
Understanding the specific methods and algorithms
employed for glioma diagnosis is essential for
comprehending the broader landscape of brain tumor
detection. M.K. Abd-Ellah et al.'s 2019 review provides a
comprehensive overview of brain tumor diagnosis from
MRI images, with a focus on the practical implications.
This study offers a holistic perspective on the challenges
and opportunities associated with using MRI for brain

tumor diagnosis, shedding light on the real-world
applications of this technology. The same work by M.K.
Abd-Ellah et al. emphasizes lessons learned from the
application of MRI in brain tumor diagnosis. The lessons
derived from this research are crucial for understanding
the evolving landscape of brain tumor diagnosis and serve
as valuable guidance for future studies. V.P. Grover et al.'s
2015 publication on magnetic resonance imaging
principles and techniques is instrumental in providing a
fundamental understanding of MRI, which forms the basis
for brain tumor imaging. Understanding the underlying
principles of MRI is essential for interpreting the results of
machine learning models applied to MRI data. In a 2000
study by H. Tang et al., MRI brain image segmentation
using multi-resolution edge detection and region selection
techniques was discussed. This study provides insights
into the image processing approaches used in the context
of brain tumor detection and their integration with
machine learning for more accurate tumor delineation. In
a 2010 study by K. Somasundaram et al., a fully automatic
brain extraction algorithm for axial T2-weighted MRI
images was presented. This addresses a crucial
preprocessing step for brain tumor diagnosis. The
automatic extraction of the brain region is an essential
component of the pipeline, and this study contributes to
the technical aspects of the process.
1.2 Methodology
A. Detecting brain tumors using a Support Vector
Machine (SVM) algorithm involves several steps, from data
collection and preprocessing to model training and
evaluation. Here's a step-by-step algorithm on how to
detect brain tumor using SVM:
2. ALGORITHM
1. Start
2. Data Collection
3. Data Preprocessing: Process the MRI images to prepare
them for training and testing.
4. Feature Extraction: Depending on your dataset and the
nature of the images, you may want to extract relevant
features from the images. Some common feature
extraction methods include Histogram of Oriented
Gradients (HOG) or Local Binary Patterns (LBP). 5. Feature
Vector Generation: Convert the processed or extracted
features into feature vectors that can be fed into the SVM.
Each image should be represented as a set of numerical
features.
6. Labeling: Assign labels to your feature vectors, such as 0
for non-tumor and 1 for tumor. Ensure that your labels
match the images in the training set.
7. SVM Model Selection: Choose the appropriate type of
SVM (linear, polynomial, radial basis function, etc.) based
on your dataset and problem. You can experiment with
different kernel functions to see which one works best for
your problem.
8. Model Training: Train the SVM model on the labeled
training data. The SVM algorithm will learn the decision
boundary that best separates the tumor and non-tumor
classes. You can use libraries like scikit-learn in Python for
SVM implementation.
9. Model Evaluation: Assess the performance of your SVM
model using appropriate evaluation metrics. Common
evaluation metrics for binary classification tasks include
accuracy, precision, recall, F1-score, and ROC curve
analysis.
10. Hyperparameter Tuning: Fine-tune the
hyperparameters of the SVM model, such as the
regularization parameter (C) or the kernel parameters, to
optimize the model's performance. You can use techniques
like cross-validation for this.
11. Testing: Use the trained SVM model to predict tumor
vs. non-tumor labels on your test dataset.
12. Post-processing: You can apply post-processing
techniques to improve the model's output, such as
thresholding to reduce false positives or false negatives.
13. Visualization: Visualize the SVM's decision boundary
or important features, which can help understand the
model's decision-making process.
14. Deployment
15. If the model performs well, we can deploy it for real
world brain tumor detection, possibly in a medical setting.
16. End
Using the above algorithm, we can detect samples
containing brain tumour.
2. EXPERIMENTS AND RESULTS
For this experiment, we utilized a diverse dataset
comprising MRI images of the brain, including both
cancerous and non-cancerous cases. The dataset was
carefully curated to encompass a wide range of tumor
types, sizes, and locations, ensuring the diversity necessary
for training and evaluating the SVM model effectively.
Fig. 2. Samples tested for no tumor.

The chosen dataset includes two types of data:-
1)Training Data
2)Testing Data
Each of these divisions of datasets includes MRI Scansof
Human Brain having no tumor and MRI Scans of Human
Brain having Pituitary Tumor.
The SVM Model is trained on the training data of chosen
dataset, Fig 2 shows results of testing of SVM model on the
testing data of chosen dataset for no tumor.
Fig. 3. Samples tested for pituitary tumor.
Fig 3 shows results of testing of SVM model on the
testingdata of chosen dataset for pituitary tumor.
Fig. 4. Testing Results.
After testing the model, we got the training score of
0.9887 andtesting score of 0.9592.
3. CONCLUSIONS
In conclusion, a promising direction in the field of medical
diagnostics is the application of Support Vector Machines
(SVM) for brain tumour detection. SVMs are a useful tool
for accurate classification, but there are issues that need to
be resolved, including issues with generalization, data
limitations, and ethics.
Investigating deep learning methods, multimodal data
fusion, and real-time detection systems, among other
possibilities, can significantly improve the efficiency and
[1] Q.D. Buchlak et al. Machine learning applications to
neuroimaging for glioma detection and classification: an
artificial intelligence augmented systematic review J. Clin.
Neurosci. (2021)
[2] M.K. Abd-Ellah et al. A review on brain tumor diagnosis
from MRI images: practical implications, key
achievements, and lessons learned Magn. Reson. Imag.
(2019)
[3] M.K. Abd-Ellah et al. A review on brain tumor diagnosis
from MRI images: practical implications, key
achievements, and lessons learned Magn. Reson. Imag.
(2019)
[4] V.P. Grover et al. Magnetic resonance imaging:
principles and techniques: lessons for clinicians Journal of
clinical and experimental hepatology (2015)
[5]. H. Tang et al. MRI brain image segmentation by
multiresolution edge detection and region selection
Comput. Med. Imag. Graph. (2000)
[6] K. Somasundaram et al. Fully automatic brain
extraction algorithm for axial T2-weighted magnetic
resonance images Comput. Biol. Med. (2010)
accuracy of tumour detection. Moreover, collaboration
between AI and medical experts is necessary for the
successful integration of SVM-based tumour detection in
clinical settings. These developments herald a bright
future for improving brain tumour identification, which
will eventually lead to better patient outcomes and care.
REFERENCES

BRAIN TUMOUR DETECTION AND CLASSIFICATION

Recommended

More Related Content

Similar to BRAIN TUMOUR DETECTION AND CLASSIFICATION (20)

More from IRJET Journal (20)

Recently uploaded (20)

BRAIN TUMOUR DETECTION AND CLASSIFICATION