CVG-Net: novel transfer learning based

deep features for diagnosis of brain

tumors using MRI scans
Shaha Al-Otaibi1 , Amjad Rehman2 , Ali Raza3 , Jaber Alyami4 and Tanzila Saba2
1
Department of Information Systems, Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia
2
Artificial Intelligence & Data Analytics Lab CCIS, Prince Sultan University, Riyadh, Saudi Arabia
3
Institute of Computer Science, Khwaja Fareed University of Engineering and Information Technology, Rahim
Yar Khan, Pakistan
4
Department of Diagnostic Radiology, Faculty of Applied Medical Sciences, King Abdulaziz University,
Jeddah, Saudi Arabia

ABSTRACT
Brain tumors present a significant medical challenge, demanding accurate and timely
diagnosis for effective treatment planning. These tumors disrupt normal brain functions
in various ways, giving rise to a broad spectrum of physical, cognitive, and emotional
challenges. The daily increase in mortality rates attributed to brain tumors underscores
the urgency of this issue. In recent years, advanced medical imaging techniques,
particularly magnetic resonance imaging (MRI), have emerged as indispensable tools
for diagnosing brain tumors. Brain MRI scans provide high-resolution, non-invasive
visualization of brain structures, facilitating the precise detection of abnormalities
such as tumors. This study aims to propose an effective neural network approach
for the timely diagnosis of brain tumors. Our experiments utilized a multi-class
MRI image dataset comprising 21,672 images related to glioma tumors, meningioma
tumors, and pituitary tumors. We introduced a novel neural network-based feature
engineering approach, combining 2D convolutional neural network (2DCNN) and
VGG16. The resulting 2DCNN-VGG16 network (CVG-Net) extracted spatial features
from MRI images using 2DCNN and VGG16 without human intervention. The newly
created hybrid feature set is then input into machine learning models to diagnose
Submitted 7 November 2023 brain tumors. We have balanced the multi-class MRI image features data using the
Accepted 1 April 2024 Synthetic Minority Over-sampling Technique (SMOTE) approach. Extensive research
Published 17 May 2024 experiments demonstrate that utilizing the proposed CVG-Net, the k-neighbors
Corresponding author classifier outperformed state-of-the-art studies with a k-fold accuracy performance
Amjad Rehman, [email protected] score of 0.96. We also applied hyperparameter tuning to enhance performance for
Academic editor multi-class brain tumor diagnosis. Our novel proposed approach has the potential
Shibiao Wan to revolutionize early brain tumor diagnosis, providing medical professionals with a
Additional Information and cost-effective and timely diagnostic mechanism.
Declarations can be found on
page 20
DOI 10.7717/peerj-cs.2008 Subjects Algorithms and Analysis of Algorithms, Artificial Intelligence, Computer Vision, Data
Mining and Machine Learning
Copyright Keywords Machine learning, Deep learning, Brain tunmor, Image processing, Transfer learning,
2024 Al-Otaibi et al.
Magnetic resonance imaging
Distributed under
Creative Commons CC-BY 4.0

OPEN ACCESS

How to cite this article Al-Otaibi S, Rehman A, Raza A, Alyami J, Saba T. 2024. CVG-Net: novel transfer learning based deep features for
diagnosis of brain tumors using MRI scans. PeerJ Comput. Sci. 10:e2008 http://doi.org/10.7717/peerj-cs.2008
 INTRODUCTION
Brain tumors are abnormal growths of cells within the brain that can disrupt normal brain
function and pose serious health risks (Miao et al., 2023). Detecting and diagnosing brain
tumors is paramount for timely intervention and treatment. MRI brain scans have become
a crucial tool (Zhu et al., 2023). MRI scans utilize powerful magnets and radio waves to
generate detailed images of the brain structure, allowing healthcare professionals to visualize
abnormalities (Liu et al., 2023b). The MRI scans are especially effective in identifying brain
tumors because they provide high-resolution images that can differentiate between healthy
brain tissue and tumor masses.
Brain tumors can be classified into various types, each with its distinct characteristics
and potential challenges in diagnosis and treatment (Mehnatkesh et al., 2023; Liu et al.,
2023a). Glioma tumors originate in the glial cells, which provide support and protection
to neurons (Oztek et al., 2023). They are often classified according to their aggressiveness,
with high-grade gliomas being particularly concerning due to their rapid growth and
infiltrative nature. On the other hand, meningioma tumors develop from the meninges,
the protective membranes surrounding the brain and spinal cord. While often benign, they
can cause symptoms by compressing nearby brain structures (Bhatele & Bhadauria, 2023).
Pituitary tumors, which arise in the pituitary gland located at the base of the brain, can
affect hormone production and regulation, leading to a wide range of health issues.
The mortality rates caused by brain tumors are increasing daily. Between 2014 and
2018, the average annual age-adjusted incidence rate for primary malignant brain tumors
was 7.06 per 100,000 individuals. Projections indicated that in 2022, an estimated 25,050
new diagnoses and 18,280 fatalities would be attributed to primary malignant brain
tumors (Thierheimer et al., 2023). Brain tumors were 24.25 per 100,000 individuals,
comprising 7.06 per 100,000 for malignant cases and 17.18 per 100,000 for non-malignant
cases (Li et al., 2023a). Similarly, based on the 2018 data from the Global Cancer Registry,
a total of 18,078,957 cancer cases were reported across all genders, with 29,681 cases
attributed to brain cancer (Khazaei et al., 2020).
Traditional diagnostic methods for brain tumors have primarily relied on radiological
imaging techniques such as computed tomography (CT) (Zhang et al., 2023) and MRI.
Classical treatment modalities for brain tumors encompass surgical resection, radiation
therapy, and chemotherapy, with the specific approach chosen depending on tumor type,
location, and patient characteristics. Deep learning based methods use MRI images analyzed
using convolutional neural networks (CNN) for brain tumor recognition (Rajinikanth,
Kadry & Nam, 2021; Rajinikanth et al., 2022; Maqsood, Damasevicius & Shah, 2021;
Maqsood, Damaševičius & Maskeliūnas, 2022; Badjie & Deniz Ülker, 2022; Kurdi et al.,
2023; Khan et al., 2023).
In this study, we propose an advanced image-processing mechanism that effectively
diagnoses brain tumors without human intervention. We introduce a novel neural network,
CVG-Net, for transfer learning-based spatial feature extraction from MRI images. These
newly created transfer features are then employed to diagnose brain tumors using machine
learning models.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 2/24

 Our contributions regarding brain tumor diagnosis are as follows:
• We have introduced a novel neural network method, CVG-Net, for feature engineering.
This method combines 2DCNN and VGG16 to extract spatial features from MRI images.
The resulting hybrid spatial feature set is input into machine learning models to diagnose
brain tumors.
• We employed the SMOTE technique to balance the multi-class MRI image feature
data, resulting in more accurate and reliable diagnoses of brain tumors. This approach
addresses class imbalance issues within medical image analysis.
• We utilized four advanced machine learning models and two deep learning models
based on neural networks for comparative performance evaluations. Each method’s
performance is optimized through hyperparameter tuning and validated using k-fold
cross-validation. Additionally, we calculated the computational complexity of each
method for diagnosing brain tumors.
The rest of the study is organized as follows: ‘Literature analysis’ analyzes the limitations
in previous literature. ‘Proposed Methodology’ demonstrates our proposed research
methodology for diagnosing brain tumors. ‘Results and discussions’ comparatively
evaluates the results of the applied neural network approaches. Our findings are concluded
in ‘Conclusions and future work’.

LITERATURE ANALYSIS
This literature analysis section comprehensively examines existing studies and
methodologies related to both neural network applications in medical imaging and
transfer learning techniques. It delves into the foundational work in medical image
analysis, emphasizing the pivotal role of neural networks in automating the detection and
diagnosis of various pathologies, including brain tumors. Additionally, it synthesizes the
key findings and methodologies from relevant studies, providing a robust foundation for
the subsequent implementation of transfer learning techniques in the proposed diagnosis
framework for brain tumors using MRI scans

Machine learning models analysis

The study Stadlbauer et al. (2022) investigates the potential of employing multiclass
machine learning (ML) algorithms on an extensive array of radiomic features derived
from advanced MRI (advMRI) and physiological MRI (phyMRI) data. This innovative
methodology, referred to as radiophysiomics, aims to achieve precise classification of
contrast-enhancing brain tumors. The study leveraged a substantial MRI database,
encompassing more than 1700 MR examinations of brain tumor patients, for the purpose
of model development. The findings revealed that the application of adaptive boosting and
random forest algorithms in conjunction with both advMRI and phyMRI data surpassed
human reading in various performance metrics, including accuracy (0.875 vs. 0.850),
precision (0.862 vs. 0.798), and F-score (0.774 vs. 0.740). Nevertheless, it was observed
that radiologists demonstrated higher sensitivity (0.767 vs. 0.750) and specificity (0.925 vs.
0.902) compared to the machine learning models.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 3/24

 The study Latif et al. (2022) introduces an innovative method for categorizing glioma
tumors, utilizing features based on deep learning in conjunction with a support vector
machine (SVM) classifier. Through the application of a deep convolutional neural network
for feature extraction from MR images sourced from the BraTS dataset, the study achieved
noteworthy levels of accuracy in classification. Specifically, an accuracy of 96.19% was
observed for the HGG Glioma type using the FLAIR modality, and a commendable
accuracy of 95.46% was obtained for the LGG glioma tumor type when employing
the T2 modality. Notably, this method encompassed the classification of four distinct
glioma classes, including edema, necrosis, enhancing, and non-enhancing. This pioneering
approach shows promising potential in the domain of glioma tumor classification.

Deep neural networks models analysis

This study Archana, Karthigha & Suresh Lavanya (2023) introduces a comparative analysis
of several optimizers employed in CNNs for the purpose of brain tumor detection.
Within deep learning (DL), artificial neural networks (ANNs) play a crucial role in
discerning brain tumors. The medical image dataset is employed for applied methods
training and evaluation. Among the various approaches investigated, the application of the
AlexNet architecture in conjunction with the stochastic gradient descent (SGD) optimizer
demonstrated superior performance. This proposed approach achieved a diagnostic
accuracy of 0.80.
This article Sharif et al. (2020) introduces a novel approach to actively employ deep
learning for the segmentation and recognition of brain tumors in MRI images. The authors
utilize the pre-trained Inception V3 CNN model for robust deep feature extraction during
the classification phase. These features are integrated with the dominant rotated local binary
pattern (DRLBP) to enhance texture analysis. Subsequently, a particle swarm optimization
(PSO) technique is employed to fine-tune the concatenated feature vector for classification
using a softmax classifier. The study is conducted in two phases. In the initial phase, the
segmentation approach (SbDL) is validated on BRATS2017 and BRATS2018 datasets,
demonstrating promising Dice scores. For instance, the BRATS2017 dataset achieves Dice
scores of 83.73% for the core tumor, 93.7% for the whole tumor, and 79.94% for the
enhanced tumor. Moreover, the BRATS2018 dataset attains a Dice score of 88.34%. In
the subsequent phase, the classification strategy is applied to BRATS2013, 2014, 2017, and
2018 datasets, yielding an average accuracy of over 92%.
The study Kibriya et al. (2022) is focused on developing an automated method for precise
brain cancer identification through tumor diagnosis. Previous systems have struggled with
issues related to subpar accuracy and high false-positive rates. To address this, the study
introduces a pioneering 13-layer CNN architecture specifically designed for classifying
brain tumors from MRI scans. The effectiveness of the proposed model was assessed using
a standardized dataset comprising 3064 MRI images encompassing three distinct categories
of brain cancer: glioma, pituitary, and meningioma. The results demonstrated an average
accuracy of 97%.
The study Fabelo et al. (2019) introduces a novel framework centered on deep learning
techniques, tailored for the analysis of hyperspectral images depicting live human brain

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 4/24

 tissue. Through rigorous evaluation using a diverse database of 26 in vivo hyperspectral
cubes from 16 individual patients, encompassing a total of 258,810 annotated pixels,
the framework proves adept at generating a thematic representation that delineates the
brain’s parenchymal area and accurately identifies the tumor’s position. This information
serves as a crucial aid to surgeons, significantly enhancing the precision of tumor resection
procedures. Notably, the deep learning pipeline 2D convolutional neural network exhibits
an impressive overall accuracy of 80% for multiclass classification, which is low compared
to baseline.

Hybrid models analysis

The author of this Sudharson et al. (2022) study proposed a hybrid deep learning-based
neural system for the detection of brain tumors. The study introduced fundamental
concepts of image separation to address the challenges associated with partitioning brain
MRI images. Various MRI pre-processing techniques were then elaborated, encompassing
tasks such as image registration, bias field correction, and non-brain tissue removal.
Additionally, the author presented a hybrid CNN-based partitioning approach, which
proved advantageous for delineating brainstem tumors in MRI scans when augmented
with prior knowledge. The proposed approach demonstrated an accuracy score of 0.93.
The study Yazdan et al. (2022) introduces an innovative approach to diagnosing brain
tumors. It employs a dual-pronged strategy, utilizing a multi-scale CNN (MSCNN)
architecture to develop a robust classification model, while also addressing the impact
of Rician noise on MSCNN performance. This model enables the multi-classification
of MRIs, distinguishing between glioma, meningioma, pituitary, and non-tumor cases.
Additionally, a denoising process is implemented using a fuzzy similarity-based non-local
means (FSNLM) filter to enhance classification results. The study utilizes an openly
accessible MRI dataset comprising 3,264 MRIs. Experimental results demonstrate that
the proposed MSCNN model outperforms both AlexNet and ResNet, exhibiting superior
accuracy and efficiency while incurring lower computational expenses. Specifically, the
MCNN2 iteration achieves an impressive accuracy and F1-score of 91.2% and 91%,
respectively.
The analyzed works with their research limitations are summarized in Table 1.

PROPOSED METHODOLOGY
This section comprehensively describes the materials and methods used to evaluate our
research experiments. We examine the MRI scans dataset and the advanced neural network
approaches applied. Figure 1 illustrates our novel proposed research methodology for
diagnosing brain tumors. The following steps outline the process carried out in our
proposed methodology for conducting experiments.
• Step 1: At the outset of our research experiments, we acquired a brain tumor-related
dataset of MRI scan images for evaluating results. This dataset encompasses multiple
target classes, allowing for a comprehensive assessment of our findings.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 5/24

 Table 1 The literature summary and limitations analysis.

Ref Dataset Proposed Performance Research limitations

technique accuracy
Archana, Karthigha Brain tumours medical AlexNet with SGD 0.80 Low diagnose performance
& Suresh Lavanya (2023) images scores
Sudharson et al. (2022) Brain MRI image scans Hybrid CNN 0.93 Low performance with a state of
art approach
Sharif et al. (2020) BRATS2017 and Inception V3 pre- 0.92 Old datasets with low perfor-
BRATS2018 datasets trained CNN model mance accuracy for diagnosis
Kibriya et al. (2022) 3,064 MRI images 13-layer CNN architec- 0.97 Classical neural networks were
ture used for classification with low
images data record
Yazdan et al. (2022) 3,264 MRIs Scans Multi-Scale CNN 0.91 Poor accuracy performance for
(MSCNN) architecture multi class data with low images
data record
Stadlbauer et al. (2022) 1,700 MRI examina- Adaptive Boosting and 0.87 Low diagnose performance
tions from advMRI and Random Forest scores with low images data
phyMRI data record
Fabelo et al. (2019) 26 in vivo hyperspectral 2D convolutional neural 0.80 Poor accuracy performance for
cubes from 16 different network multiclass classification
patients, among which
258,810 pixels
Latif et al. (2022) BraTS dataset Support vector machine 0.96 Classical machine learning ap-
(SVM) proaches were used with out-
dated segmented images data
record

2D-CNN

Features
Splitting

VGG-16
Multi Class Brain Tumors Novel Transfer Learning SMOTE Based Features
Feature Engineering Data Balancing Training
MRI Image Scans
Testing

Normal

Glioma_tumor

Meningioma_tumor

Pituitary_tumor
Diagnosis of Brain Tumors Model Evaluations Hyper-Parameterized Artificial
Using MRI Scans & Predictions Intelligence Approach

Figure 1 Workflow diagram of our proposed methodology for brain tumor diagnosis.
Full-size DOI: 10.7717/peerjcs.2008/fig-1

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 6/24

 Figure 2 The target class labels distributions analysis.
Full-size DOI: 10.7717/peerjcs.2008/fig-2

• Step 2: Next, for the extraction of rich-level features from MRI images, we have
proposed a novel neural network approach that leverages transfer learning for spatial
feature extraction, ultimately creating a new feature set.
• Step 3: We then analyzed the dataset and identified an imbalance. To address this issue,
we applied the SMOTE approach to balance the features of the dataset for multi-class
classification.
• Step 4: The image feature data is then divided into two portions: training and testing.
The split ratio employed is 80% for training and 20% for testing.
• Step 5: We have employed advanced machine learning and deep learning approaches
in this step. Using 80% of the training data, we have implemented the applied methods.
• Step 6: Then, the 20% portion of the data is utilized for the evaluations and predictions
of applied methods through various performance metrics in comparison.
• Step 7: Ultimately, the superior neural network method during experiments is employed
for brain tumor diagnosis.

Multi class brain tumors MRI image data

This research used multiclass MRI scan images (Hashemi, 2023) related to brain tumors to
conduct experiments. The dataset comprises 21,672 MRI scan images. Figure 2 illustrates
the distribution of multi-class target labels. This analysis reveals that the ‘Normal’ target
class (0) contains 3,066 samples, the ‘Glioma_tumor’ target class (1) contains 6,307 samples
‘Meningioma_tumor’ target class (2) contains 6,391 samples, and the ‘Pituitary_tumor’

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 7/24

 Figure 3 The sample dataset image analysis with the target class.
Full-size DOI: 10.7717/peerjcs.2008/fig-3

target class (3) contains 5,908 samples. Additionally, we conducted a sample image analysis,
as shown in Fig. 3.
We utilized the state-of-the-art dataset: Multi-class Brain Tumour MRI Image Data,
comprising 21,672 MRI scan images. We chose this dataset because it aligns with the
problem addressed in our research study, specifically, brain tumor detection. Moreover,
this dataset encompasses a substantial number of images, enhancing the generalizability of
our research findings.

Novel transfer learning approach

This section comprehensively describes our novel proposed CVG-Net approach for transfer
feature engineering. The step-by-step architectural workflow of the proposed approach
is illustrated in Fig. 4. The brain tumor MRI image data is initially fed into the VGG16
approach, which extracts rich-level spatial features based on transfer learning. Similarly,
the brain tumor MRI image data is input into the 2DCNN approach, which also extracts
rich-level spatial features using transfer learning. The spatial features obtained from both
approaches are then combined to create a new feature set. This newly generated feature set
is subsequently employed in building the machine learning models applied in this study
and for further results evaluations.
The Algorithm 1 shows the step-by-step workflow of the proposed transfer learning
feature engineering.

SMOTE based data balancing

SMOTE has emerged as a pivotal tool in addressing class imbalance issues within medical
image analysis, particularly in diagnosing brain tumors using multi-class MRI image
datasets (Li et al., 2023b). Class imbalance, where certain classes are underrepresented

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 8/24

 VGG16 Features Extraction Output:
(21672, 200)

Hybrid Spatial
Input:(128, 128, 3) SoftMax Features
Conv2D Flatten (21672, 400)
MaxPooling2D
Layer
Repeated 5 Times

2D-CNN Features Extraction Output:

(21672, 200)

Input:(128, 128, 3) SoftMax

Conv2D MaxPooling2D Flatten
Layer

Figure 4 The architectural workflow analysis of our proposed CVG-Net feature engineering.
Full-size DOI: 10.7717/peerjcs.2008/fig-4

Algorithm 1 CVG-Net Algorithm

Input: Multi-class MRI scans related to brain tumors.
Output: New Transfer learning-based deep features.
initiate;
1- Fcnn ←− CNNprediction (MiS) // MiS  MRI images set , here MiS is original image data
and Fcnn is the deep spatial feature set extracted.
1- Fvgg 16 ←− VGG16prediction (MiS) // MiS  MRI images set , here MiS is original image
data and Fvgg 16 is the deep spatial feature set extracted.
3- FSpat ←− Fcnn + Fvgg 16 // here FSpat is final hybrid deep spatial features set used for
osteoarthritis detection.
end;

relative to others, can significantly hinder the performance of machine learning models,
as they tend to be biased towards the majority class. In medical imaging, this can have
critical implications for patient care. SMOTE operates by generating synthetic samples in
the feature space of the minority class, thereby alleviating the imbalance. When applied
to multi-class MRI datasets for brain tumor diagnosis, SMOTE ensures that the model
is exposed to a more representative data distribution, effectively discriminating between
various tumor types. This methodology has demonstrated promising results, contributing
to more accurate and reliable diagnoses, paramount in clinical decision-making processes.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 9/24

 Applied neural network methods
The application of artificial intelligence (AI) based machine learning, deep learning, and
transfer learning methods has emerged as a transformative approach in medical imaging
(Rehman et al., 2023; Qadri et al., 2023; Raza et al., 2022b), particularly in diagnosing brain
tumors using brain MRI scans. These advanced techniques have revolutionized the way
medical professionals analyze and interpret complex neuroimaging data. Machine learning
algorithms can efficiently extract intricate patterns and features from large datasets of MRI
scans, enabling them to differentiate between normal brain tissue and pathological regions
associated with brain tumors.
Deep learning, in particular, has been instrumental in automating the tumor detection,
segmentation, and classification process, as it can automatically learn hierarchical
representations from raw MRI data, making it highly effective in identifying subtle
abnormalities (Abunadi & Kumar, 2021; Inbarani H & Azar, 2020; Azar et al., 2023).
Transfer learning techniques leverage pre-trained models on large datasets and have shown
promise in enhancing diagnostic accuracy by transferring knowledge learned from other
domains or medical imaging tasks to improve the performance of brain tumor diagnosis.
Integrating AI-driven methods in brain tumor diagnosis accelerates the detection process
and aids healthcare providers in making more accurate and timely decisions.
In this research, we employed several advanced machine learning and deep learning-
based approaches for performance evaluations. We utilized two neural network approaches,
2D-CNN and VGG16, for both classification and transferring feature mechanisms from
MRI images. The machine learning models we employed include the k-neighbors classifier,
logistic regression, and random forest. We have provided a detailed description of the
working mechanisms of each applied neural network in Table 2 and layer architecture in
Table 3.

Hyperparameter setting
Hyperparameter tuning is a crucial aspect of optimizing neural network models in
machine learning. These hyperparameters are configurations external to the model and can
significantly impact its performance. In our research, we focus on systematically exploring
and optimizing these hyperparameters to enhance the overall efficiency and effectiveness of
neural networks. We employ a recursive process of training and evaluations to determine
the best-fit hyperparameters. The selected hyperparameters for the neural network applied
to diagnose brain tumors are presented in Table 4.

RESULTS AND DISCUSSIONS

The results and discussions of our study methods proposed for the diagnosis of brain tumors
using brain MRI scans are analyzed in this section. We have provided a comprehensive
analysis of the results obtained through the application of neural network methods. In
this section, we dive into the key findings and their implications for the field of medical
imaging and brain tumor diagnosis.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 10/24

 Table 2 The working mechanism description analysis of applied neural network approaches.

Method Working description

2D-CNN A 2D-CNN (Archana, Karthigha & Suresh Lavanya, 2023; Zhou et al.,
2023) works by systematically scanning and analyzing small regions
(kernels) of a 2D input image, such as a Brain MRI scan. These ker-
nels extract features like edges, textures, and patterns through con-
volution operations. Multiple layers of convolutions and pooling are
used to hierarchically learn and combine these features, enabling the
CNN to detect complex patterns indicative of brain tumors, making
it an effective method for automated diagnosis using MRI scans.
VGG-16 The VGG-16 neural network method (Rohith et al., 2023; Shourie,
Anand & Gupta, 2023) for diagnosing brain tumors using brain MRI
scans operates by utilizing a deep CNN architecture. It employs a se-
ries of convolutional layers to extract intricate features from the MRI
images, gradually learning hierarchical representations of the input
data. These features are then passed through fully connected layers
for classification, distinguishing between tumor and non-tumor re-
gions based on learned patterns, ultimately aiding in accurate brain
tumor diagnosis.
KNC K-neighbors classifier (KNC) method for diagnosing brain tumors
using brain MRI scans operates by analyzing the proximity of each
MRI scan to its neighboring data points in a high-dimensional fea-
ture space. KNC assigns a class label to a given MRI scan based on
the majority class among its k nearest neighbors, where k is a user-
defined parameter. This proximity-based approach allows KNC to ef-
fectively identify brain tumors by leveraging the collective character-
istics of nearby MRI scans in the dataset, making it a valuable tool in
medical image analysis (Raza et al., 2023; Raza et al., 2022a).
LR Logistic regression (LR) (van Ravesteijn et al., 2010) for diagnosing
brain tumors using MRI scans begins by collecting a dataset of brain
MRI images along with corresponding binary labels indicating the
presence or absence of tumors. The LR model calculates the proba-
bility of a brain tumor being present in each MRI scan by applying a
logistic function to a linear combination of image features extracted
from the scans. It learns the optimal weights for these features during
training. Once trained, the LR model can classify new MRI scans as
tumor-positive or tumor-negative based on the probability threshold
(typically 0.5), providing a valuable tool for non-invasive brain tu-
mor diagnosis.
RF Random forest (RF) (Dutta et al., 2023) for diagnosing brain tumors
using MRI scans work by first collecting a diverse dataset of brain
MRI images with corresponding tumor labels. The RF algorithm
then creates multiple decision trees using random subsets of the data
and features, allowing each tree to classify tumor presence or absence
independently. RF combines the results from these trees through vot-
ing or averaging, resulting in a robust and accurate diagnosis of brain
tumors from MRI scans.

Experimental setup
Our proposed experimental study setup includes the use of the Google Colab platform to
evaluate research results. The platform offers GPU backend support, providing 90 GB of
disk space and 13 GB of RAM. We have employed the Python 3.0 programming language
to build the neural network used in our research. We utilized the Sklearn, Tensorflow, and

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 11/24

 Table 3 The layered architecture analysis of applied neural network approaches.

Layer (type) Output Shape Param #

2D-CNN
conv2d (Conv2D) (None, 126, 126, 64) 1,792
max_pooling2d (MaxPooling2D) (None, 63, 63, 64) 0
dropout (Dropout) (None, 63, 63, 64) 0
flatten (Flatten) (None, 254,016) 0
dense (Dense) (None, 4) 1,016,068
VGG-16
vgg16 (Functional) (None, 4, 4, 512) 14,714,688
dropout_2 (Dropout) (None, 4, 4, 512) 0
flatten_2 (Flatten) (None, 8,192) 0
dense_2 (Dense) (None, 4) 32,772

Table 4 The hyper-parameters tuning analysis of applied neural network approaches.

Method Hyper-parameters description

2D-CNN activation= ‘softmax’, Dropout=0.02,
optimizer=‘adam’, loss=‘ategorical_crossentropy’,
metrics=[‘accuracy’,‘Precision’,‘Recall’]
VGG-16 weights = ‘imagenet’, include_top = False,
activation= ‘softmax’, Dropout=0.02,
optimizer=‘adam’, loss=‘categorical_crossentropy’,
metrics=[‘accuracy’,‘Precision’,‘Recall’]
KNC n_neighbors=3, weights=‘uniform’, leaf_size=30, p = 2
LR penalty=‘l2’, tol=1e−4, C = 1.0, solver=‘lbfgs’
RF max_depth=50, criterion=‘‘gini’’, random_state=0,
n_estimators=50, max_features=‘‘sqrt’’

Keras modules for evaluating the results. During performance evaluations, true positives
(TP), true negatives (TN), false positives (FP), and false negatives (FN) are used as basic
parameters. The performance metrics used for performance evaluation are examined as
follows:
• Accuracy: in diagnosing brain tumors using brain MRI scans refers to measuring how
effectively the diagnostic system or method correctly identifies and classifies the presence
or absence of brain tumors in the MRI images.
TP + TN
Accuracy = . (1)
TP + TN + FP + FN
• Precision: in diagnosing brain tumors using brain MRI scans refers to measuring how
accurately the diagnostic test identifies TP cases among all the positive cases it detects.
TP
Precision = . (2)
TP + FP
• Recall: in diagnosing brain tumors using brain MRI scans, is a fundamental performance
metric that measures the ability of a diagnostic model or process to correctly identify

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 12/24

 and classify TP cases of brain tumors among all the actual cases of brain tumors present
in the MRI scans.
TP
Recall = . (3)
TP + FN
• F1: in diagnosing brain tumors using brain MRI scans, is a measure that combines
precision and recall to assess the accuracy of the classification model.
2 · Precision · Recall
F 1 Score = . (4)
Precision + Recall
• Runtime computations: is a neural network method’s brain tumor diagnosis time by
inputting MRI images.
• Standard deviations (SD): refers to a statistical measure of the variability of the
evaluation metrics obtained from multiple folds of the dataset during the cross-validation
process.
v
u
u1X N
σ =t (xi − µ)2 . (5)
N
i=1

Performance analysis with classical approaches

The performance analysis of the applied classical deep learning-based approaches, 2DCNN
and VGG16, is conducted in this section. MRI brain scan images are fed into the 2DCNN
and VGG16 methods, and the results are then evaluated.

Results of CNN model

The performance results of the applied CNN model during training are illustrated in Fig. 5.
We conducted a time series analysis over 20 epochs of the CNN model. The evaluation is
based on loss scores, accuracy, precision, and recall. The analysis reveals that the training
loss is very high in the first two epochs. Subsequently, there is a notable reduction in loss,
while the validation loss gradually increases. Similarly, the training accuracy starts off low
in the initial two epochs, but then exhibits a steady increase. The validation accuracy scores
fluctuate around 0.80. The precision and recall metrics exhibit a similar pattern to the
accuracy performance scores throughout training. In conclusion, this analysis indicates
that the CNN model achieved acceptable scores for brain tumor diagnosis during training.
Nevertheless, there is still a need for performance enhancement.
The performance analysis of the applied CNN model for unseen testing data is described
in Table 5. The analysis shows that the CNN model achieved an acceptable score of 0.76
during brain tumor diagnosis. However, the loss scores are high for the testing data. The
class-wise performance scores also demonstrate moderate results. This analysis concludes
that classical approaches have not proven fruitful in this case; there is a need for advanced
mechanisms for performance enhancement.

Results of VGG16 model

The performance results of the applied pre-trained VGG16 model during training are
illustrated in Fig. 6. We conducted a time series analysis over 20 epochs of the VGG16
model. The analysis reveals that the training loss is very high in the first five epochs.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 13/24

 Computer Science Manuscript to be reviewed

(a) Loss (b) Loss

(c) Accuracy (d) Accuracy

(e) Precision (f) Precision

(g) Recall (h) Recall

Figure 5. The time series-based performance results comparison of CNN model during training. Each
Figure 5 The time series-based performance results comparison of CNN model during training. Each
performance
performancemeasure
measureisisevaluated
evaluated during
during the
the 20 epochsof
20 epochs oftraining.
training.
Full-size DOI: 10.7717/peerjcs.2008/fig-5
283 4.2 Performance analysis with classical approaches
284 The performance analysis of the applied classical deep learning-based approaches, 2DCNN and VGG16,
285 is Table 5 The
conducted inresults analysis
this section. of applied
MRI 2D-CNN
brain scan neural
images network.
are fed into the 2DCNN and VGG16 methods, and
286 the results are then evaluated.
Accuracy Loss Target class Precision Recall F1
287 4.2.1 Results of CNN Model Normal 0.72 0.76 0.74
288 The performance results of the applied CNN model during training
Glioma_tumor 0.70 are illustrated 0.68in Figure 5.0.69
We
289 conducted
0.76 a time series analysis over 20 epochs of the CNN model. The evaluation is based on loss scores,
3.64 Meningioma_tumor 0.78 0.75 0.77
290 accuracy, precision, and recall. The analysis reveals that the training loss is very high in the first two
291
Pituitary_tumor 0.87 0.87
epochs. Subsequently, there is a notable reduction in loss, while the validation loss gradually increases. 0.87
292 Similarly, the training accuracy starts off low in the initial two epochs, but then exhibits a steady increase.
Average 0.77 0.76 0.77
293 The validation accuracy scores fluctuate around 0.80. The precision and recall metrics exhibit a similar
294 pattern to the accuracy
Subsequently, thereperformance
is a notablescores throughout
reduction in training. In conclusion,
loss, while this analysis
the validation indicates that
loss gradually
295 the CNN model achieved acceptable scores for brain tumor diagnosis during training. Nevertheless, there
296
increases.
is still a needSimilarly, the training
for performance accuracy starts off low in the initial two epochs, but
enhancement.
297 then The performance
exhibits analysis
a steady of the The
increase. applied CNN model
validation for unseen
accuracy testing
scores data is around
fluctuate described0.92.
in Table
The
298 5. The analysis shows that the CNN model achieved an acceptable score of 0.76
precision and recall metrics exhibit a similar pattern to the accuracy performance scores during brain tumor
299 diagnosis. However, the loss scores are high for the testing data. The class-wise performance scores also

11/19
PeerJ Comput. Sci. reviewing PDF | (CS-2023:10:92458:1:1:NEW 18 Jan 2024)
Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 14/24
 Normal 0.72 0.76 0.74
Glioma tumor 0.70 0.68 0.69
0.76 3.64 Meningioma tumor 0.78 0.75 0.77
Pituitary tumor 0.87 0.87 0.87
Average 0.77 0.76 0.77

(a) Loss (b) Loss

(c) Accuracy (d) Accuracy

(e) Precision (f) Precision

(g) Recall (h) Recall

Figure
Figure6.6 The
Thetime
timeseries-based performanceresults
series-based performance resultscomparison
comparisonofof VGG16
VGG16 model
model during
during training. Each
training.
performance measure
Each performance is evaluated
measure during
is evaluated the 20
during the epochs ofoftraining.
20 epochs training.
Full-size DOI: 10.7717/peerjcs.2008/fig-6

302 4.2.2 Results of VGG16 Model

303
throughout
The performancetraining. In the
results of conclusion, this analysis
applied pre-trained VGG16indicates
model that thetraining
during VGG16 aremodel achieved
illustrated in Figure
304 6. We conducted a time series analysis over 20 epochs of the VGG16 model. The analysisareveals
good scores for brain tumor diagnosis during training. Nevertheless, there is still need that
305 the
fortraining loss is very
performance high in the first five epochs. Subsequently, there is a notable reduction in loss,
enhancement.
306 whileThethe performance
validation loss analysis
graduallyofincreases.
the appliedSimilarly,
VGG-16 the training
model accuracy
on unseen starts off low
testing in the
data is initial
outlined in Table 6. The analysis reveals that the VGG-16 model achieved a commendable
score of PDF
0.90| (CS-2023:10:92458:1:1:NEW
in brain tumor diagnosis.18Nevertheless, 12/19
the loss scores were notably high for
PeerJ Comput. Sci. reviewing Jan 2024)
the testing data. The class-wise performance scores also demonstrate positive results, with
0.91 precision and F1-scores. This analysis leads to the conclusion that classical approaches
yielded a moderately satisfactory performance. However, there is a pressing need for
advanced mechanisms to enhance performance.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 15/24

 Table 6 The results analysis of applied VGG-16 neural network.

Accuracy Loss Target class Precision Recall F1

Normal 0.99 0.89 0.94
Glioma_tumor 0.91 0.85 0.88
0.90 2.66 Meningioma_tumor 0.83 0.93 0.87
Pituitary_tumor 0.96 0.95 0.95
Average 0.91 0.90 0.91

Table 7 The results analysis of applied neural networks with novel unbalanced features.

Method Accuracy Target class Precision Recall F1

Normal 0.87 0.81 0.84
Glioma_tumor 0.79 0.82 0.81
LR 0.81 Meningioma_tumor 0.76 0.73 0.75
Pituitary_tumor 0.87 0.91 0.89
Average 0.82 0.82 0.82
Normal 0.95 0.85 0.90
Glioma_tumor 0.82 0.83 0.83
RF 0.85 Meningioma_tumor 0.81 0.78 0.80
Pituitary_tumor 0.87 0.95 0.91
Average 0.87 0.85 0.86
Normal 0.94 0.93 0.94
Glioma_tumor 0.91 0.96 0.93
KNC 0.94 Meningioma_tumor 0.96 0.87 0.91
Pituitary_tumor 0.95 0.98 0.96
Average 0.94 0.94 0.94

Performance analysis with Novel approach

The performance analysis of the applied advanced machine learning methods with the
novel proposed CVG-Net feature engineering is presented in this section. The proposed
approach extracts novel transfer features by inputting MRI brain scan images into the
2DCNN and VGG16 methods, creating a hybrid feature set. The newly created features are
then input into machine learning models, and the results are subsequently evaluated.

Results with unbalanced features

In this section, the novel extracted features that are unbalanced are inputted into applied
machine learning methods for performance evaluations. The performance results of
machine learning methods with unbalanced features are described in Table 7. The analysis
demonstrates that applied machine learning models achieved high-performance scores
using the novel extracted transfer features. The LR and RF achieved accuracy performance
scores of 0.81 and 0.85, respectively. The applied KCN approach achieved a good score
of 0.94. This analysis concludes that with unbalanced novel features, good performance
is achieved. However, the performance can be further improved by applying advanced
data-balancing approaches.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 16/24

 Table 8 The results analysis of applied neural networks with novel balanced features.

Method Accuracy Target class Precision Recall F1

Normal 0.91 0.93 0.92
Glioma_tumor 0.82 0.79 0.80
LR 0.84 Meningioma_tumor 0.74 0.72 0.73
Pituitary_tumor 0.87 0.90 0.89
Average 0.84 0.84 0.84
Normal 0.96 0.97 0.96
Glioma_tumor 0.83 0.83 0.83
RF 0.88 Meningioma_tumor 0.82 0.76 0.79
Pituitary_tumor 0.90 0.95 0.92
Average 0.88 0.88 0.88
Normal 0.96 0.99 0.97
Glioma_tumor 0.93 0.95 0.94
KNC 0.95 Meningioma_tumor 0.95 0.88 0.91
Pituitary_tumor 0.96 0.99 0.98
Average 0.95 0.95 0.95

Results with balanced features

After evaluating the performance scores with unbalanced feature data, we applied the
famous data balancing Synthetic Minority Oversampling Technique (SMOTE) to the
extracted feature data. The evaluation results of the applied machine learning approach with
a balanced dataset are demonstrated in Table 8. The analysis reveals that balanced features
help improve performance scores for the diagnosis of brain tumors. The proposed KNC
method outperformed other applied methods with a high accuracy of 0.95 in comparison.
This analysis concludes that SMOTE-based dataset balancing helps us improve performance
scores.
The confusion matrix allows visualization of the performance of applied machine
learning methods for the diagnosis of brain tumors. The overall performance validation,
based on the confusion matrix, is illustrated in Fig. 7. The analysis shows that the applied
LR and RF have high wrong predictions for target classes 2 and 3. The minimum error rate,
indicating the fewest wrong predictions, is achieved by the proposed K-nearest neighbors
classification (KNC) approach. This analysis validates the high-performance scores of the
proposed KNC method for the diagnosis of brain tumors using MRI scans.

Validating performance with k-fold mechanism

We have validated the performance of applied machine-learning approaches in this section
using both unbalanced and balanced feature data. The results of k-fold cross-validations
are demonstrated in Table 9. To validate accuracy performance results, we used a fold split
of 10 and determined the outcomes. The analysis shows that with unbalanced features, LR
and RF achieved acceptable validation accuracy scores in the range of 0.81 to 0.84. The
proposed approach achieved a commendable score of 0.94. When we evaluated results
with the balanced dataset, performance scores increased for each applied approach. The

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 17/24

 Computer Science Manuscript to be reviewed

KNC LR
1200
1324 2 7 3 1200 1240 28 51 17

0
1000
1000
23 1196 40 4 32 997 207 27 800
800

1
Actuals

Actuals
600 600
35 82 1099 39 56 169 907 123

2
400 400

2 7 5 1245 200 32 23 65 1139 200

3

3
0 1 2 3 0 1 2 3
Predictions Predictions
RF
1200
1290 14 17 15
0

1000

25 1046 161 31 800

1
Actuals

600
24 178 959 94
2

400

5 25 27 1202 200
3

0 1 2 3
Predictions

Figure 7.
Figure 7 Confusion Confusion
matrix analysismatrix analysis
of applied neuralofnetwork
appliedapproaches.
neural network approaches.
Full-size DOI: 10.7717/peerjcs.2008/fig-7

355 4.7 Computational Complexity analysis

Table 9 K-fold cross-validations results analysis of applied neural networks.
356 In this section, we present a computational complexity analysis of applied machine learning methods. The
357 runtime
Methodcomputations, measured in seconds,
Unbalanced of these applied machine learning
features methods,
Balanced featuresare determined
358 during each model-building process using a newly created features dataset. The computation results are
10-fold accuracy SD (+/-) 10-fold accuracy SD (+/-)
359 reported in Table 10. Our analysis demonstrates that the proposed KNC method achieved the minimum
360 runtime
LR computations0.81 compared to other applied methods. The proposed
0.005 0.83 KNC method takes only 0.0281
0.009
361 seconds
RF to diagnose 0.84
brain tumors. 0.007 0.88 0.005
KNC 0.94 0.004 0.96 0.005
Table 10. Computational complexity analysis of applied neural networks with novel balanced features.
analysis concludes that the proposed KNC approach achieved the highest k-fold accuracy
Method Runtime (seconds)
scores of 0.96 for the diagnosis of brain MRI scans.
LR 3.7201
RF 24.215
Computational complexity analysis
KNC 0.0281
In this section, we present a computational complexity analysis of applied machine
learning methods. The runtime computations, measured in seconds, of these applied
machine learning methods, are determined during each model-building process using a
362 4.8 Comparison with state-of-the-art studies
newly created features dataset. The computation results are reported in Table 10. Our
363 For a fair performance comparison, we have included state-of-the-art studies in brain tumor diagnosis
364 inanalysis demonstrates
this analysis. that the comparison
The performance proposed KNC
of ourmethod achieved
proposed approachthe minimuminruntime
is illustrated Table 11. The
365 studies considered for performance comparison are from the years 2020, 2022, and 2023. The analysis
366 reveals that our proposed approach outperformed state-of-the-art studies with high-performance scores of
367 0.96 for the diagnosis of brain tumors using MRI brain scans.
Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 18/24
15/19
PeerJ Comput. Sci. reviewing PDF | (CS-2023:10:92458:1:1:NEW 18 Jan 2024)
 Table 10 Computational complexity analysis of applied neural networks with novel balanced features.

Method Runtime (seconds)

LR 3.7201
RF 24.215
KNC 0.0281

Table 11 The performance comparison with state-of-the-art studies for brain tumor diagnosis.

Ref. Year Proposed technique Accuracy

Archana, Karthigha & Suresh Lavanya (2023) 2023 AlexNet with SGD 0.80
Sudharson et al. (2022) 2022 Hybrid CNN 0.93
Sharif et al. (2020) 2020 Inception V3 with CNN 0.92
Yazdan et al. (2022) 2022 Multi-Scale CNN 0.91
Our 2024 CVG-Net-KNC 0.96

computations compared to other applied methods. The proposed KNC method takes only
0.0281 s to diagnose brain tumors.

Comparison with state-of-the-art studies

For a fair performance comparison, we have included state-of-the-art studies in brain
tumor diagnosis in this analysis. The performance comparison of our proposed approach
is illustrated in Table 11. The studies considered for performance comparison are from the
years 2020, 2022, and 2023. The analysis reveals that our proposed approach outperformed
state-of-the-art studies with high-performance scores of 0.96 for the diagnosis of brain
tumors using MRI brain scans.

Discussions
In this research, we have introduced a novel neural network method, CVG-Net, for feature
engineering and the timely diagnosis of brain tumors. This proposed method combines
2DCNN and VGG16 to extract spatial features from MRI images. Our experiments
utilized a multi-class MRI image dataset comprising 21,672 images related to glioma
tumors, meningioma tumors, and pituitary tumors. Our commitment to privacy and
confidentiality underscores the implementation of robust data anonymization techniques,
thereby mitigating the risk of any inadvertent disclosure of sensitive personal details. This
explicit ethical discourse is pivotal not only in upholding the integrity of our research but
also in fostering trust within the scientific community and, more importantly, among the
individuals whose health data form the foundation of our study.

Study limitations
In this research study, we have introduced a novel CVG-Net for the prompt diagnosis
of brain tumors. Nevertheless, our proposed research does have some limitations. The
accuracy score of our approach, currently at 96%, could be further improved by minimizing
the loss error rates. To address potential overfitting issues or biases in the machine learning
models, we have employed the k-fold cross-validation mechanism, which validates the

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 19/24

 performance accuracy scores. Moreover, the deep learning-based neural network models
we utilized are computationally expensive and can be optimized by reducing their layered
architectures.

CONCLUSIONS AND FUTURE WORK

This research proposes an effective neural network approach for the timely diagnosis
of brain tumors. Our experiments utilized a multi-class MRI image dataset comprising
21,672 images related to glioma tumors, meningioma tumors, and pituitary tumors. We
introduced a novel neural network-based feature engineering approach called CVG-Net,
which combines 2DCNN and VGG16. The proposed CVG-Net method extracts spatial
features from MRI images using 2DCNN and VGG16 without human intervention.
The newly created hybrid feature set is then input into machine learning models to
diagnose brain tumors. We utilized four advanced machine learning models and two
deep learning models based on neural networks for comparative performance evaluations.
We also balanced the multi-class MRI image features data using the SMOTE approach.
Extensive research experiments demonstrate that utilizing the proposed CVG-Net, the
k-nearest neighbors classifier outperformed state-of-the-art studies with a k-fold accuracy
performance score of 0.96. We also applied hyperparameter tuning to enhance performance
for multi-class brain tumor diagnosis.

Future work
In the future, we intend to design a graphical user interface app for medical specialists to
facilitate the effective diagnosis of brain tumors using MRI scans. Also we plan to enhance
the performance by implementing transfer learning-based advanced neural networks. The
designed app will utilize the proposed approach in the backend for diagnosis.

ADDITIONAL INFORMATION AND DECLARATIONS

Funding
This research is supported by Princess Nourah bint Abdulrahman University Researchers
Supporting Project number (PNURSP2024R136), Princess Nourah bint Abdulrahman
University, Riyadh, Saudi Arabia. Prince Sultan University, Riyadh Saudi Arabia
supported the Article Processing Charges (APC) of this publication. Princess Nourah
bint Abdulrahman University had a role in the study design, conduct, data analysis and
interpretation (determined the scope of the work, directed the analysis, gave feedback and
dictated the desired output/results verification), manuscript writing, and dissemination of
results.

Grant Disclosures
The following grant information was disclosed by the authors:
Princess Nourah bint Abdulrahman University Researchers: PNURSP2024R136.
Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.
Prince Sultan University, Riyadh Saudi Arabia.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 20/24

 Competing Interests
The authors declare there are no competing interests.

Author Contributions
• Shaha Al-Otaibi conceived and designed the experiments, analyzed the data, authored
or reviewed drafts of the article, and approved the final draft.
• Amjad Rehman conceived and designed the experiments, prepared figures and/or tables,
and approved the final draft.
• Ali Raza performed the experiments, performed the computation work, prepared figures
and/or tables, authored or reviewed drafts of the article, and approved the final draft.
• Jaber Alyami analyzed the data, performed the computation work, prepared figures
and/or tables, and approved the final draft.
• Tanzila Saba analyzed the data, authored or reviewed drafts of the article, and approved
the final draft.

Data Availability
The following information was supplied regarding data availability:
The dataset is available at Kaggle: https://www.kaggle.com/datasets/mohammadhossein77/
brain-tumors-dataset.
The code is available in the Supplementary Files.

Supplemental Information
Supplemental information for this article can be found online at http://dx.doi.org/10.7717/
peerj-cs.2008#supplemental-information.

REFERENCES
Abunadi I, Kumar RL. 2021. Blockchain and business process management in health
care, especially for COVID-19 cases. Security and Communication Networks
2021:2245808 DOI 10.1155/2021/2245808.
Archana B, Karthigha M, Suresh Lavanya D. 2023. Comparative analysis of op-
timisers used in CNN for brain tumor detection. In: 2023 5th international
conference on inventive research in computing applications (ICIRCA). 687–692
DOI 10.1109/ICIRCA57980.2023.10220625.
Azar AT, Khan ZI, Amin SU, Fouad KM. 2023. Hybrid global optimization algo-
rithm for feature selection. Computers, Materials and Continua 74:2021–2037
DOI 10.32604/cmc.2023.032183.
Badjie B, Deniz Ülker E. 2022. A deep transfer learning based architecture for brain
tumor classification using MR images. Information Technology and Control
51(2):332–344 DOI 10.5755/j01.itc.51.2.30835.
Bhatele KR, Bhadauria SS. 2023. Multiclass classification of central nervous system
brain tumor types based on proposed hybrid texture feature extraction methods
and ensemble learning. Multimedia Tools and Applications 82(3):3831–3858
DOI 10.1007/s11042-022-13439-1.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 21/24

 Dutta P, Paul S, Jana GG, Sadhu A. 2023. Hybrid genetic algorithm random forest
algorithm (HGARF) for improving the missing value imputation in hepatitis medical
dataset. In: 2023 international symposium on devices, circuits and systems (ISDCS), vol.
1. 01–05 DOI 10.1109/ISDCS58735.2023.10153526.
Fabelo H, Halicek M, Ortega S, Shahedi M, Szolna A, Piñeiro JF, Sosa C, O’Shanahan
AJ, Bisshopp S, Espino C. 2019. Deep learning-based framework for in vivo
identification of glioblastoma tumor using hyperspectral images of human brain.
Sensors 19(4):920 DOI 10.3390/s19040920.
Hashemi SMH. 2023. Crystal clean: brain tumors MRI dataset. Kaggle. Available at
https://www.kaggle.com/ds/3505991 (accessed on 09 May 2023).
Inbarani HH, Azar AT. 2020. Leukemia image segmentation using a hybrid histogram-
based soft covering rough k-means clustering algorithm. Electronics 9(1):188
DOI 10.3390/electronics9010188.
Khan MA, Khan A, Alhaisoni M, Alqahtani A, Alsubai S, Alharbi M, Malik NA,
Damaševičius R. 2023. Multimodal brain tumor detection and classification using
deep saliency map and improved dragonfly optimization algorithm. International
Journal of Imaging Systems and Technology 33(2):572–587 DOI 10.1002/ima.22831.
Khazaei Z, Goodarzi E, Borhaninejad V, Iranmanesh F, Mirshekarpour H, Mirzaei B,
Naemi H, Bechashk SM, Darvishi I, Ershad Sarabi R. 2020. The association between
incidence and mortality of brain cancer and human development index (HDI): an
ecological study. BMC Public Health 20(1):1696 DOI 10.1186/s12889-020-09838-4.
Kibriya H, Masood M, Nawaz M, Nazir T. 2022. Multiclass classification of brain
tumors using a novel CNN architecture. Multimedia Tools and Applications
81(21):29847–29863 DOI 10.1007/s11042-022-12977-y.
Kurdi SZ, Ali MH, Jaber MM, Saba T, Rehman A, Damaševičius R. 2023. Brain tumor
classification using meta-heuristic optimized convolutional neural networks. Journal
of Personalized Medicine 13(2):181 DOI 10.3390/jpm13020181.
Latif G, Ben Brahim G, Iskandar DA, Bashar A, Alghazo J. 2022. Glioma Tumors’
classification using deep-neural-network-based features with SVM classifier.
Diagnostics 12(4):1018 DOI 10.3390/diagnostics12041018.
Li S, Wang C, Chen J, Lan Y, Zhang W, Kang Z, Zheng Y, Zhang R, Yu J, Li W. 2023a.
Signaling pathways in brain tumors and therapeutic interventions. Signal Transduc-
tion and Targeted Therapy 8:8 DOI 10.1038/s41392-022-01260-z.
Li Z, Wang Y, Jiang Z, Luo Z, Wu J, Toe TT. 2023b. Ensemble of CNN models for
identifying stages of Alzheimer’s disease: an approach using MRI scans and SMOTE
algorithm. In: 2023 3rd international symposium on computer technology and
information science (ISCTIS). 496–500 DOI 10.1109/ISCTIS58954.2023.10213182.
Liu H, Huo G, Li Q, Guan X, Tseng M-L. 2023a. Multiscale lightweight 3D segmentation
algorithm with attention mechanism: brain tumor image segmentation. Expert
Systems with Applications 214:119166 DOI 10.1016/j.eswa.2022.119166.
Liu Z, Tong L, Chen L, Jiang Z, Zhou F, Zhang Q, Zhang X, Jin Y, Zhou H. 2023b. Deep
learning based brain tumor segmentation: a survey. Complex & Intelligent Systems
9(1):1001–1026 DOI 10.1007/s40747-022-00815-5.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 22/24

 Maqsood S, Damasevicius R, Shah FM. 2021. An efficient approach for the detection
of brain tumor using fuzzy logic and U-NET CNN classification. Lecture Notes in
Computer Science 12953 LNCS:105–118 DOI 10.1007/978-3-030-86976-2_8.
Maqsood S, Damaševičius R, Maskeliūnas R. 2022. Multi-modal brain tumor de-
tection using deep neural network and multiclass SVM. Medicina 58(8):1090
DOI 10.3390/medicina58081090.
Mehnatkesh H, Jalali SMJ, Khosravi A, Nahavandi S. 2023. An intelligent driven deep
residual learning framework for brain tumor classification using MRI images. Expert
Systems with Applications 213(Part C):119087 DOI 10.1016/j.eswa.2022.119087.
Miao Y-B, Zhao W, Renchi G, Gong Y, Shi Y. 2023. Customizing delivery nano-
vehicles for precise brain tumor therapy. Journal of Nanobiotechnology 21(1):32
DOI 10.1186/s12951-023-01775-9.
Oztek MA, Noda SM, Romberg EK, Cole BL, Wright JN, Ishak GE, Perez FA. 2023.
Changes to pediatric brain tumors in 2021 World Health Organization classification
of tumors of the central nervous system. Pediatric Radiology 53(3):523–543
DOI 10.1007/s00247-022-05546-w.
Qadri AM, Raza A, Munir K, Almutairi MS. 2023. Effective feature engineering
technique for heart disease prediction with machine learning. IEEE Access
11:56214–56224 DOI 10.1109/ACCESS.2023.3281484.
Rajinikanth V, Kadry S, Damasevicius R, Sujitha RA, Balaji G, Mohammed MA. 2022.
Glioma/Glioblastoma detection in brain MRI using pre-trained deep-learning
scheme. In: Proceedings of the 2022 3rd international conference on intelligent com-
puting, instrumentation and control technologies: computational intelligence for smart
systems, ICICICT 2022. 987–990 DOI 10.1109/ICICICT54557.2022.9917904.
Rajinikanth V, Kadry S, Nam Y. 2021. Convolutional-neural-network assisted seg-
mentation and svm classification of brain tumor in clinical MRI slices. Information
Technology and Control 50(2):342–356 DOI 10.5755/j01.itc.50.2.28087.
Raza A, Rustam F, Siddiqui HUR, Diez IdlT, Ashraf I. 2023. Predicting microbe
organisms using data of living micro forms of life and hybrid microbes classifier.
PLOS ONE 18(4):284–522 DOI 10.1371/journal.pone.0284522.
Raza A, Rustam F, Siddiqui HUR, Diez IdlT, Garcia-Zapirain B, Lee E, Ashraf I. 2022a.
Predicting genetic disorder and types of disorder using chain classifier approach.
Gene 14(1):e0284522 DOI 10.3390/genes14010071.
Raza A, Siddiqui HUR, Munir K, Almutairi M, Rustam F, Ashraf I. 2022b. Ensemble
learning-based feature engineering to analyze maternal health during pregnancy and
health risk prediction. PLOS ONE 17(11):e0276525 DOI 10.1371/journal.pone.0276525.
Rehman A, Raza A, Alamri FS, Alghofaily B, Saba T. 2023. Transfer learning-based
smart features engineering for osteoarthritis diagnosis from knee X-ray images. IEEE
Access 11:71326–71338 DOI 10.1109/ACCESS.2023.3294542.
Rohith S, Prakash MS, Anitha R, Kumar KS, Yogeswara Sai K. 2023. Detection of brain
tumor using VGG16. In: 2023 8th international conference on communication and
electronics systems (ICCES). 1400–1405 DOI 10.1109/ICCES57224.2023.10192639.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 23/24

 Sharif MI, Li JP, Khan MA, Saleem MA. 2020. Active deep neural network features
selection for segmentation and recognition of brain tumors using MRI images.
Pattern Recognition Letters 129:181–189 DOI 10.1016/j.patrec.2019.11.019.
Shourie P, Anand V, Gupta S. 2023. An intelligent VGG-16 framework for brain tumor
detection using MRI-scans. In: 2023 international conference on smart systems for ap-
plications in electrical sciences (ICSSES). 1–6 DOI 10.1109/ICSSES58299.2023.10200833.
Stadlbauer A, Marhold F, Oberndorfer S, Heinz G, Buchfelder M, Kinfe TM, Meyer-
Bäse A. 2022. Radiophysiomics: brain tumors classification by machine learning and
physiological MRI data. Cancer 14(10):2363 DOI 10.3390/cancers14102363.
Sudharson K, Sermakani A, Parthipan V, Dhinakaran D, Eswari Petchiammal G,
Usha N. 2022. Hybrid deep learning neural system for brain tumor detection.
In: 2022 2nd international conference on intelligent technologies (CONIT). 1–6
DOI 10.1109/CONIT55038.2022.9847708.
Thierheimer M, Cioffi G, Waite KA, Kruchko C, Ostrom QT, Barnholtz-Sloan JS. 2023.
Mortality trends in primary malignant brain and central nervous system tumors vary
by histopathology, age, race, and sex. Journal of Neuro-Oncology 162(1):167–177
DOI 10.1007/s11060-023-04279-6.
van Ravesteijn VF, van Wijk C, Vos FM, Truyen R, Peters JF, Stoker J, van Vliet
LJ. 2010. Computer-aided detection of polyps in CT colonography using
logistic regression. IEEE Transactions on Medical Imaging 29(1):120–131
DOI 10.1109/TMI.2009.2028576.
Yazdan SA, Ahmad R, Iqbal N, Rizwan A, Khan AN, Kim D-H. 2022. An efficient multi-
scale convolutional neural network based multi-class brain MRI classification for
SaMD. Tomography 8(4):1905–1927 DOI 10.3390/tomography8040161.
Zhang H, He H, Gao Y, Mady A, Filipović V, Dyck M, Lv J, Liu Y. 2023. Applications
of computed tomography (CT) in environmental soil and plant sciences. Soil and
Tillage Research 226:105574 DOI 10.1016/j.still.2022.105574.
Zhou Q, Huang Z, Ding M, Zhang X. 2023. Medical image classification us-
ing light-weight CNN with spiking cortical model based attention mod-
ule. IEEE Journal of Biomedical and Health Informatics 27(4):1991–2002
DOI 10.1109/JBHI.2023.3241439.
Zhu Z, He X, Qi G, Li Y, Cong B, Liu Y. 2023. Brain tumor segmentation based on the
fusion of deep semantics and edge information in multimodal MRI. Information
Fusion 91:376–387 DOI 10.1016/j.inffus.2022.10.022.

Al-Otaibi et al. (2024), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.2008 24/24