Deep Learning based Multi-class Brain Tumor Classification

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 03 | Mar 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 778
Deep Learning based Multi-class Brain Tumor Classification
Yash A. Paddalwar1, Atharva M. Patil2, Saurav Sabu3, Harsh A. Singh4, Sheetal Gawande5
1Student, Department of Information Technology, Pillai College of Engineering, New Panvel, Maharashtra, India
5Professor, Department of Computer Engineering, Pillai College of Engineering, New Panvel, Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - A brain tumor is a kind of cancerthatcanimpact
people, sometimes fatally or with significant quality of life
impairment. Using Deep learning techniques, researchers can
identify tumors and treat them more efficiently. Brain MRI
pictures can be used in a variety of ways to find malignancies.
Deep learning techniques have significantly outperformedthe
rest of these techniques. Within the framework, comparison of
the models has been employed for tumor detection from brain
MRI scans. Among the Convolutional Neural Network (CNN)
architectures that can be employed are Custom CNN,
DenseNet169, MobileNet, VGG-16, and ResNet152models. The
same hyper-parameters can be used to train these models on
MR images that have undergone the same dataset and
preprocessing procedures. The goal is to develop an
architecture that will compare various models to classify the
Brain Tumor MRI. Machine learning and deep learning
algorithms can be used to directly scan and determine the
presence and type of tumor. Therefore, it is useful for
analyzing brain tumor detection performance using various
methods. The dataset used for Brain Tumor Detectionconsists
of approximately 5000 Brain MR Images.
Key Words: Classification, Neural Networks, Brain Tumor
Classification, Deep Learning, Artificial Intelligence
1.INTRODUCTION
A relatively large number of people are diagnosed with
secondary neuropathy. Although the exact number is
unknown but this type of brain tumor is on the rise. Withthe
use of extremely effective clinical imaging tools, early
detection can always speed up the processofcontrollingand
eliminating tumors in their early stages. A patient with the
tumor may become immobile because a tumor may place
pressure on the part of the brain that regulatesmovement of
the body [6] [14] [15]. The aim of this studyistoimprove the
detection accuracy of brain tumors on MRI picture using
image processing and machine learning algorithms, as well
as to develop a framework for rapidly diagnosing brain
tumors from MRI images [19]. Amin et al. proposed [3] a
three-step methodfordistinguishingbetweencancerous and
non-cancerous brain tissue MRI. Among the steps involved
are image processing, feature extraction, and image
classification. This framework is useful not only for the
medical staff but also for the other employees of the
company because there is a chance that teams will need to
divide the images into various categories. In such
circumstances, this can be used to distinguish between the
images and keep patient records secure. This could be a
crucial tool that is helpful for any hospital employee because
medical images are delicate and must be handled with
extreme care. The objective is to identify and classify
different types of tumors and most importantly to save time
of doctors/patients and provide a suitable remedy at an
early stage and to identify and supply good insights to
doctors [20]. Examining the Brain tumor MRI images in the
Healthcare industry implies the process of identifying
tumors in the early stage. The framework will help in
automatic detection of images containing tumors and
searching for correlations between neighboring slices along
with it can also do automatic detection of symmetrical axes
of the image [13]. Furthermore, this framework can be used
to create a full-fledged application to detect any type of
Cancerous Polyps.
2. BACKGROUND
2.1 Convolutional Neural Network
Convolutional Neural Networks (ConvNet/CNN) is a deep
learning technique that can accept input images and assign
gist to various elements and objects (learnable weights and
biases), and be able to distinguish between them. ConvNets,
by comparison, depends upon approximately less
preprocessingthanotherclassificationtechniques.ConvNets
can learn from filters and properties, but primitive
techniques create the filters manually. To classify multi-
grade brain tumors, a novel convolutional neural network
(CNN) is proposed [2]. Individual neurons perceive stimuli
only in this restricted area of the visual field.
Fig - 1: CNN Architecture

A layer that receives model inputs is called an input layer.
The total number of features in the data is the same as the
number of neurons (pixels in the case of images) in that
layer. Input Layer: The hidden layer receives the input from
the input layer [1] [7]. Depending on our model and the
volume of the data, there may be numerous hidden levels as
shown in Fig 1. The number of neurons in each hidden layer
varies, but usually exceeds the number of features. The
output of each layer is computed by multiplying the output
of the layer below by a learnable weight, adding a learnable
bias, and then computing an activation function that makes
the network nonlinear [21]. The output from the hidden
layer is then passed into the output layer, where it is
converted into the probability score for each class using a
logistic function like sigmoid or SoftMax. Applications for
CNN include decoding facial recognition, understanding
climate, and gathering historical and environmental data
[10].
2.2 Transfer Learning
Transfer learning is a term used in machine learning to
describe the use of a previously trained model for another
task. In transferlearning,machinesuseinformationgathered
from previous work to improve predictions about newtasks
[16]. For instance, if a user trains a classifier to determine if
an image contains food, the learned knowledge can be used
to distinguish between drinks. Transfer learning applies the
expertise of trained machine learning models to unrelated
but closely related problems. For example, in this case, if
user trained a simple classifier to predict whether an image
contains tumor, user can use modeling data to determine
whether the tumor in the image are present or not.
Fig - 2: Transfer Learning
Shows how transfer learning basically uses what the
machine or user has learned in one task to try to better
understand the concepts of another task. The weights are
automatically transferred from the network that performed
"task B" to the network that performed "task A". Because of
high CPU power requirement, transfer learning is usually
used in computer vision.
2.3 VGG-16
The key difference between the networks and the deeper
networks of 2014 is that some designs are flashier and
perform even better. The 16-layer VGG 16 architecture
consists of his two layers of convolutions, a layerforpooling,
and a fully connected layer. The concept of a much deeper
network with much smaller filters is known as a VGG
network. It currently has VGGNet models with 16 to 19
layers. One crucial aspect of these models is that they
consistently used 3 x 3 convolutional filters, which are the
smallest convolutional filter sizescapableofexaminingsome
of the nearby pixels. All the way acrossthenetwork,theyjust
maintained this basic structure of 3 x 3 convs with periodic
pooling. Because VGG had fewer parameters,itlayeredmore
of them rather than using larger filters. Insteadofusinghuge
filters, VGG uses smaller and deeper filters. With a 7 x 7
convolutional layer, it now has the same effective receptive
field as that layer. Convolutional layers,a poolinglayer,a few
other convolution layers, a few more pooling layers, and so
on are present in VGGNet. Fig 3 VGG architecturehas16fully
connected
Fig - 3: VGG-16
Convolution layers. In this case, VGG 16 contains16andVGG
19 contains 19, which means basicallythesamearchitecture,
but with some extra layers of convolution.
2.4 MobileNET
The MobileNet model, as its name suggests, is TensorFlows
first mobile computer vision model andisintendedforusein
mobile applications.
Fig - 4: MobileNET
Fig 4 shows Depth-separable folding is used in MobileNet.
When compared to conventional convolution nets withequal
depth folds, the number of parameters is significantly
reduced. As a result, a lightweight deep neural network has

been created. Two processes are used to create a depthwise
separable convolution.
1. Depthwise convolution.
2. Pointwise convolution.
The convolution described above is an excellent starting
point for training ridiculously small and extremely fast
classifiers. Google offers an open-source CNN class called
MobileNet.
2.5 ResNET-152
The residual block serves as ResNet’s primary base
component. The complexity of processing increasesasdelve
further into the network with several levels. Stack of these
layers on top of one another is done, with each layer
attempting to uncover some underlying mapping of the
desired function. However, instead of using these blocks,
user try to adjust the remaining mapping.
Because the input to these blocks is simply the incoming
input, classes are used directly to adjusttheremainderof the
function H(X) - X rather than the predicted function H(X).
Basically, the input is just taken and passed through as an
identity at the end of this block, where it takes the skip
connection on this right here. If there were no weight layers
between, the input would simply be the identity. If further
weight layers are not employed
Fig - 5: ResNET-152
to learn some delta from residual X, the result would be the
same as the output. In a word, while moving further into the
network, learning H(X) becomes increasingly difficultdueto
the high number of layers. As a result, F(x) direct input of x
as the outcome in this caseandemployedskipconnection.So
F(x) is referred to as a residual as shown in Fig 5. All of these
blocks are very closely stacked in ResNet. Another benefitof
this extremely deep architecture is that it allows for up to
150 levels of this, which gets periodicallystack.Additionally,
stride two is used to down sample spatially and double the
number of filters. Only layer 1000 was ultimately fully
connected to the output classes.
2.6 DenseNET-169
As demonstrated in the Fig 6, a forward pass in a
conventional convolutional neural network involvespassing
an input image through the network to obtain a predicted
label for the output.
Fig - 6: DenseNET-169
Except for the first convolutional layer, which uses the input
image, all subsequent convolutional layers build the output
feature map that is passed to the next convolution layers
using the output of the previous layer. The L layers have L
direct connections, one from each layer to the next. Each
layer in the DenseNet architecture is connected to every
other layer, thus the term densely connected convolutional
network. For L classes, there are L(L+1)/2 direct
connections. Each layer uses the feature map from all layers
before it as input, and its own feature map is used as input
for each layer after it. DenseNet layers take input as a
concatenation of features Map from previous level.
3 RELATED WORKS
Techniques based on deep learning have recently been used
to classify and detect brain tumours using MRI and other
imaging methods.
[4] A custom CNN algorithm was created,whichimproved
the model by training with additional MRI images to
distinguish between tumor and non-tumor images. This
paper only proposes to determine if an image contains a
tumor and introduces a mobile application as a medical tool.
A computer-based method for differentiating braintumor
regions in MRI images. The algorithm uses NN techniques to
complete the appropriate phases of image preprocessing,
image segmentation, image feature extraction, and image
sorting [5].
[12] suggests developing an intelligent mechanism to
detect brain tumours in MRI images using clustering
algorithms such as Fuzzy C Means and intelligent
optimization tools. A CAD system was used, and the results
showed that PSO improved classification accuracy aswell as
typical error rate accuracy to 92.8 percent.
An automated system for real-time brain tumor detection
was proposed by two distinct deep learning-based methods
for the detection and classification of brain tumors [11].

Sasikala and team [18] presented a genetic algorithm for
selecting wavelet features for feature dimension reduction.
The best feature vector that can be fed into a selected
classifier, such as an ANN, is the foundation of the method.
The findings demonstrated that the genetic algorithm was
able to achieve an accuracy of 98 percent by only selecting
four features out of a total of 29.
Sajjad et al. [17] suggested a CNN method for data
augmentation for the classification of brain tumors. The
method used to classify brain tumors based on MRI images
of segmented brain tumors. For classification, they utilizeda
pre-trained VGG-19 CNN architecture and achieved
accuracies of 87.39 percent and 90.66 percent for the data
before and after augmentation, respectively.
A sophisticated method for classifying and categorising
brain tumors from Magnetic Resonance Imaging (MR)
images has been proposed [9]. The operation of the two
image restoration filters, the active use of the adaptivemean
filter on the MR images, and the appropriate image
enhancement and clipping steps required for tumor
recognition are essential parameters needed to evaluate
image quality.Theestablishedtechniqueisprimarilyfocused
on tumor detection, specifically detecting abnormal mass
accumulation and significantly influencing the pixel-wise
intensity distribution of the image.
4. METHODOLOGY
4.1 About Dataset
The dataset contains a total of about 5000 MR images
divided into three different tumor types (Glioma,
Meningioma and Pituitary) and one class of healthy brain
depicting No tumor.
Sample MR Images of all classes are represented in Fig - 7:
Fig - 7: Sample Input Images
4.2 Flow of the Project
In contrast to the existing architecture, the proposedsystem
addresses all its disadvantages and limitations. Fig 8 is a
more detailed explanation of the proposed system: First,the
system will access the image, then prepare it. To ensure that
the image will stand up for the algorithm to predict
accurately, the image will be enlarged during preprocessing.
The image will then be scaled to meet a specific need. After
scaling, the image will then be reshaped into required
dimensions so that it can fit in the algorithms that it is going
to pass through. The image will then be transformed into an
array for additional modeling. After all preprocessing is
complete, the image is run throughfivedifferentbestmodels
which predicts effective output. Now since the model is
trained for predicting tumor from the existing four classes,
the type tumor that is present in the image will be predicted
based on the probability from the softmax activation
function. The tumor type that is associated with maximum
probability will be given as output. Because five strong and
robust models are used to test the same input, this design’s
unique feature is that it offers significant reliability of the
presence of tumor and its type. As a result, a detailed
comparison can be made between all five model outputs
based on the accuracy of the output as well as the predicted
output using a web-based interface.
Fig - 8: Proposed Methodology Architecture
Input Image
The first and foremost step is to collect the brain MRI
images.
This data can be collected from websites like Kaggle, UCI
repository etc. A lot of brain MRI images are required to
train and test our model it to check the accuracy and
precision of the proposed model.
Image Processing
After getting the data, data processing is done. It is a very
important step as if the image size is very large then the
model will take a very long time to train the model. As a
result, input images are resized to lower dimensions so that
the model does not take a lot of time to train. Sometimes the
number of images will be very less, which is very bad for
deep learning models as it is data hungry. Further, data
augmentation is performed to increase the number of
images. Brain MRI images are basically GrayScale images so
the value ranges from 0 to 255. So, normalization is done on
this data in the range of 0 to 1. Normalization is basically
done on top of images so that training the model becomes

much faster. The entire preprocessing of all the images can
be done at once by using the keras ImageDataGenerator
function in which custom parameters are provided and it
will automatically convert set of images into desired
processed image and also it will convert data into desired
batches so that it can be passed directly through the
convolution neural network ortransferlearningmodel.Once
the image processing is done then it will be passed through
next stage which is converting image to ND Array and based
on classification model predicts whether tumorispresent or
not.
Convert Image into N-D array
Images are basically in the form of pixels. It is basically a
GrayScale image so it has 1 single channel. The pixel values
range between 0 to 255. The images are represented in the
form of 2D matrix. Models like an ANN could not take 2D
matrix as input, as a result there is need to convert the
images into N-D array so that ANN could take this as an
input. This section is optional if ImageDataGenerator
function is used. Because the function ImageDataGenerator
will automatically convert the set of images into suitable
format for CNN and store it in a variable for further
processing.
Classification model
After converting images into N-D arrays, the next step is to
train the model. Before training the model, since this is a
classification problem and the data is in the form of images,
CNN and transfer learning methodologies can be used to
train the model. In the proposed system five different
ideologies are been used i.e. Custom CNN with 6-layered
architecture, MobileNET transfer learning model, VGG- 16
transfer learning model, ResNET-152 transfer learning
model and DenseNET-169 transfer learning model. These
were proven to be the best as far as robustness andaccuracy
are concerned. Once the model is trained and analyzed, the
same trained model can be used for any new unseen images
to predict the type of tumor that is present inside it. For
testing new unseen images, the same ideology of
preprocessing will be applied on it and then it will be passed
through these trained models and finally the prediction
whether the MRI contains the pituitarytumor,glioma tumor,
meningioma tumor or no tumor is present. Currently only
these four classes are available and in future more classes
can also be integrated with the existing data to extend the
scope of the project.
Detailed analysis and comparison of prediction
Since the work have used five robust models, aftertheimage
is passed through all the five models, the output can be seen
and finally a detailed comparison of each model’s output is
made with graphs and accuracies associated with each as
shown in Fig 9.
5. RESULT AND ANALYSIS
The obtained results are presented in the Fig 9 below.Inthis
figure, one can see the accuracy of 98.32 percentage withthe
custom CNN 6-layered model with some preprocessing on
the unseen images. One can also see that usage of the
transfer learningMobileNet architecturewithpreprocessing
has brought an accuracy of 98.63 percentage.UsingResNET-
152 and VGG-16 architecture one can see a great accuracyof
97.71 percentage and 98.62 percentage respectively on
testing data. With the usage of Transfer learning DenseNET
architecture, an accuracy of 96.56 percentage has been
obtained on the unseen images of brain tumor MRI.
Fig - 9: Comparison of Model Accuracies
5.1 Custom CNN
6-layered architecture of CNN has given pretty good
accuracy on unseen images [8].
Accuracy: 98.32 percentage as represented in Table 1.
Accuracy and Loss graph comparison can be seen in Fig 10
and 11 respectively.
TP TN FP FN Precision Recall f1-
score
Glioma
Tumor
288 1010 12 0 1.00 0.96 0.98
Meningioma
Tumor
300 991 6 11 0.96 0.98 0.97
No Tumor 404 902 1 4 0.99 1.00 0.99
Pituitary
Tumor
297 1008 3 3 0.99 0.99 0.99
Accuracy 0.98
Macro avg 0.98 0.98 0.98
Weighted
avg
0.98 0.98 0.98
Table - 1: CNN Model Performance

Fig - 10: Accuracy Comparison of CNN Model
Fig - 11: Loss Comparison of CNN Model
5.2 MobileNET
Pre-trained architecture of MobileNET has given great
accuracy on unseen images.
Accuracy and Loss comparison is shown in Fig 12 and 13
respectively.
score
Glioma 290 1008 10 0 0.98 0.97 0.97
Meningioma 303 995 3 10 0.94 0.99 0.97
No Tumor 403 906 2 0 1.00 1.00 1.00
Pituitary 290 999 10 1 1.00 0.97 0.98
Accuracy 0.98
Macro avg 0.98 0.98 0.98
Weighted
avg
0.98 0.98 0.98
Table - 2: MobileNET Model Performance
Fig - 12: Accuracy Comparison of MobileNET Model
Fig - 13: Loss Comparison of MobileNET Model
5.3 ResNET-152
Pre-trained architecture of MobileNET has given great
Accuracy and Loss graph comparison of ResNET-152 can be
seen in Fig 14 and 15 respectively.
score
Glioma
Tumor
284 1002 9 16 0.97 0.95 0.96
Meningioma
Tumor
295 998 17 11 0.95 0.96 0.95
No Tumor 405 905 1 0 1.00 1.00 1.00
Pituitary
Tumor
297 1008 3 3 0.99 0.99 0.99
Accuracy 0.98
Macro avg 0.98 0.98 0.98
Weighted
avg
0.98 0.98 0.98
Table - 3: ResNET-152 Model Performance

Fig - 14: Accuracy Comparison of ResNET-152 Model
Fig - 15: Loss Comparison of ResNET-152 Model
5.4 VGG-16
Pre-trained architecture of VGG-16 has given great accuracy
on unseen images.
score
Glioma
Tumor
286 1009 2 14 0.99 0.95 0.97
Meningioma
Tumor
302 991 14 4 0.96 0.99 0.97
No Tumor 405 905 1 0 1.00 1.00 1.00
Pituitary
Tumor
300 1010 1 0 1.00 1.00 1.00
Accuracy 0.99
Macro avg 0.99 0.99 0.99
Weighted
avg
0.99 0.99 0.99
Table - 4: VGG-16 Model Performance
Fig - 16: Accuracy Comparison of VGG-16 Model
Fig - 17: Loss Comparison of VGG-16 Model
5.5 DenseNET-169
Pre-trained architecture of DenseNET-169 has given a nice
score
Glioma
Tumor
273 1009 27 0 0.99 0.91 0.95
Meningioma
Tumor
290 978 16 25 0.91 0.95 0.93
No Tumor 404 900 1 6 0.99 1.00 0.99
Pituitary
Tumor
299 1001 1 10 0.97 1.00 0.98
Accuracy 0.97
Macro avg 0.97 0.96 0.96
Weighted
avg
0.97 0.97 0.97
Table – 5: DenseNET-169 Model Performance

Fig - 18: Accuracy Comparison of DenseNET-169 Model
Fig - 19: Loss Comparison of DenseNET-169 Model
6. CONCLUSION
In this report, a peculiar brain tumor detection architecture
has been developed that is beneficial forthecharacterization
of four MRI modalities. It implies that each modality has
distinctive qualities to effectively aid class distinction by the
network. CNN model (the most popular deep learning
architecture) can achieve performance close to that of
human observers by processing only the portionofthe brain
image that is close to the tumor tissue. It has also been
suggested to use an easy-to-use but effective cascade CNN
model to extract local and global characteristics in two
separate methods utilising extraction patches of various
sizes. The patches are chosen and fed into the network
whose centres are situated in the predicted region of the
tumour following the extraction of the tumour utilising our
method. Because a considerablenumberofuselesspixelsare
eliminated from the image during the preprocessing stage,
calculation time is decreased and the capacity for quick
predictions for clinical image classification is increased. The
comparative study of various techniquesmentionedaboveis
presented in this report.
REFERENCES
[1] Zeynettin Akkus, Alfiia Galimzianova,AssafHoogi,Daniel
L Rubin, and Bradley J Erickson. Deep learning for brain
mri segmentation: state of the art and future directions.
Journal of digital imaging, 30:449–459, 2017.
[2] Saad Albawi, Tareq Abed Mohammed, and SaadAl-Zawi.
Understanding of a convolutional neural network. In
2017 international conference on engineering and
technology (ICET), pages 1–6. Ieee, 2017.
[3] Javeria Amin, Muhammad Sharif, Mussarat Yasmin, and
Steven Lawrence Fernandes. A distinctive approach in
brain tumor detection and classification using mri.
Pattern Recognition Letters, 139:118–127, 2020.
[4] Sagaya Aurelia et al. A machine learning entrenched
brain tumor recognition framework. In 2022
International Conference on Electronics and Renewable
Systems (ICEARS), pages 1372–1376. IEEE, 2022.
[5] Ehab F Badran, Esraa Galal Mahmoud, and Nadder
Hamdy. An algorithm for detecting brain tumors in mri
images. In The 2010 International Conference on
Computer Engineering & Systems, pages 368–373. IEEE,
2010.
[6] Yoshua Bengio, Aaron Courville, and Pascal Vincent.
Representationlearning:Areviewandnewperspectives.
IEEE transactions on pattern analysis and machine
intelligence, 35(8):1798–1828, 2013.
[7] Jose Bernal, Kaisar Kushibar, Daniel S Asfaw, Sergi
Valverde, Arnau Oliver, Robert Martí, and Xavier Lladó.
Deep convolutional neural networks for brain image
analysis on magnetic resonance imaging: a review.
Artificial intelligence in medicine, 95:64–81, 2019.
[8] Daniel Berrar. Cross-validation., 2019.
[9] Rupsa Bhattacharjee and Monisha Chakraborty. Brain
tumor detection from mr images: Image processing,
slicing and pca based reconstruction. In 2012 Third
International Conference on Emerging Applications of
Information Technology, pages 97–101. IEEE, 2012.
[10] Yi Chang, Luxin Yan, Meiya Chen, Houzhang Fang, and
Sheng Zhong. Twostage convolutional neural network
for medical noise removal via image decomposition.
IEEE Transactions on Instrumentation andMeasurement,
69(6):2707–2721, 2019.
[11] Nadim Mahmud Dipu, Sifatul Alam Shohan, and KMA
Salam. Deep learning based brain tumor detection and
classification. In 2021 International Conference on
Intelligent Technologies (CONIT), pages 1–6. IEEE, 2021.

[12] N Nandha Gopal and M Karnan. Diagnose brain tumor
through mri using image processing clustering
algorithms such as fuzzy c means along with intelligent
optimization techniques. In 2010 IEEE international
conference on computational intelligence and computing
research, pages 1–4. IEEE, 2010.
[13] Yurong Guan, Muhammad Aamir, Ziaur Rahman,
Ammara Ali, Waheed Ahmed Abro, Zaheer Ahmed Dayo,
Muhammad Shoaib Bhutta, and Zhihua Hu.Aframework
for efficient brain tumor classification using mri images.
2021.
[14] Paul Kleihues and Leslie H Sobin. World health
organization classification of tumors. Cancer,
88(12):2887–2887, 2000.
[15] Sarah H Landis, Taylor Murray, Sherry Bolden, and
Phyllis A Wingo. Cancer statistics, 1999. CA: A cancer
Journal for Clinicians, 49(1):8–31, 1999.
[16] Priyanka Modiya and Safvan Vahora. Brain tumor
detection using transfer learning with dimensionality
reduction method. International Journal of Intelligent
Systems and Applications in Engineering,10(2):201–206,
2022.
[17] Muhammad Sajjad, Salman Khan, Khan Muhammad,
Wanqing Wu, Amin Ullah, and Sung Wook Baik. Multi-
grade brain tumor classification using deep cnn with
extensive data augmentation. Journal of computational
science, 30:174–182, 2019.
[18] M Sasikala and N Kumaravel. A wavelet-based optimal
texture feature set for classification of brain tumours.
Journal of medical engineering & technology, 32(3):198–
205, 2008.
[19] Javier E Villanueva-Meyer, Marc C Mabray,andSoonmee
Cha. Current clinical brain tumor imaging. Neurosurgery,
81(3):397–415, 2017.
[20] Sarmad Fouad Yaseen, Ahmed S Al-Araji, and Amjad J
Humaidi. Brain tumor segmentationandclassification:A
one-decade review. International Journal of Nonlinear
Analysis and Applications, 13(2):1879–1891, 2022.
[21] Qianru Zhang, Meng Zhang, Tinghuan Chen, Zhifei Sun,
Yuzhe Ma, and Bei Yu. Recent advances in convolutional
neural network acceleration. Neurocomputing, 323:37–
51, 2019.

Deep Learning based Multi-class Brain Tumor Classification

Recommended

More Related Content

Similar to Deep Learning based Multi-class Brain Tumor Classification (20)

More from IRJET Journal (20)

Recently uploaded (20)

Deep Learning based Multi-class Brain Tumor Classification