Brain Tumor Prediction
LITERATURE REVIEW:
[1]. MM-BiFPN: Multi-Modality Fusion Network with Bi-FPN for MRI Brain Tumor
Segmentation.
In this article, propose an end-to-end encoder-decoder network with multi-modality
encoders to exploit the relationship between different modalities in 3D MRI brain
tumor segmentation. The proposed MM-BiFPN model consists of an end-to-end fully
convolutional encoder-decoder network.
The performance of the proposed network was evaluated by calculating the Dice
coefficient, sensitivity, PPV and Jaccard similarity index of each tumor region:
complete tumor, core tumor, and enhancing tumor.
In this article, to imports MICCAI BraTS2018 and MICCAI BraTS2020 datasets are
used.
Research Gap:
This work, plan to focus more on reducing the time and computing limitations and
develop a better brain tumor segmentation model.
[2]. Time Distributed-CNN-LSTM: A Hybrid Approach Combining CNN and LSTM to
Classify Brain Tumor on 3D MRI Scans Performing Ablation Study.
In this article, propose a hybrid model named Time Distributed-CNN-LSTM (TD-
CNN-LSTM) combining 3D Convolutional Neural Network (CNN) and Long Short
Term Memory (LSTM) where each layer is wrapped with a Time Distributed
function.
To demonstrate the TD-CNN-LSTM network outperforms 3D CNN achieving the
highest test accuracy of 98.90%.
In this article, propose, two fully automatic deep learning approaches are carried out
for the classification of glioma brain tumors into HGG and LGG.
In this article, imports BRATS 2018, BRATS 2019 and BRATS 2020 datasets are
used.
Research Gap:
This work, lowest time complexity can be a useful approach in future research and
real world tumor diagnosis system.
[3].Pattern Descriptors Orientation and MAP Firefly Algorithm Based Brain Pathology
Classification Using Hybridized Machine Learning Algorithm.
In this article, the proposed methodology successfully achieves a reliable accuracy of
88.3% for classifying brain tumor cases and 99.2% for brain stroke classification.
In this article, HSVFC, SVM, FFNN, DC, ResNet-18 and KNN is used and
implementing GUI in MATLAB 2019b platform.
In this article, imports BRATS (Brain Tumor Segmentation) 2013 and ISLES
(Ischemic Stroke Lesion Segmentation) datasets are used.
Research Gap:
This work, deep neural network can be utilized for detecting multiple brain diseases
by collecting the input images from several databases without applying data
augmentation to achieve high accuracy and low error in predicting the diseases.
[4]. MVFusFra: A Multi-View Dynamic Fusion Framework for Multimodal Brain
Tumor Segmentation.
Findings from the evaluations suggest that fusion results from multi-views achieve
better performance than segmentation results from the single view, and also implying
effectiveness of the proposed multi-view fusion loss.
A comparative summary also shows that MVFusFra achieves better segmentation
performance, in terms of efficiency. The MVFusFra has been proposed to facilitate
the process of implementing the multi-view method to improve the performance of
brain tumor segmentation.
In this article, Multi-view Dynamic Fusion Framework and Deep Neural Network
algorithm is used.
In this article, imports BRATS 2015 and BRATS 2018 datasets are used.
Research Gap:
This work, how to improve the network parameters and the efficiency when utilizing
the complex network structure as the backbone network in the proposed framework.
[5]. HDC-Net: Hierarchical Decoupled Convolution Network for Brain Tumor
Segmentation.
Introduce a hierarchical decoupled convolution network (HDC-Net) which is a single
light-weight pseudo-3D network and intended to be a simple yet strong baseline for
brain tumor segmentation.
In existing methods for multi-class brain tumor segmentation design effective
methods using variants of FCN or U-Net with 2D or/and 3D convolutions.
In this article, HDC-Net, Light-weight network and FCN algorithms is used and
implemented in PyTorch.
This model imports BRATS 2018 and 2017 datasets are used.
Research Gap:
This work, exploits better fusion of the multi-modal data instead of simple
concatenation.
[6]. Machine Learning and Deep Learning Approaches for Brain Disease Diagnosis.
The application of artificial intelligence in medical science has increased the precision
and accuracy with which brain disease may be predicted and identified.
The author looked at the most recent advances in machine learning and deep learning
for diagnosing four main brain disorders: Alzheimer's (AD), brain tumor, epilepsy,
and Parkinson's disease.
Research Gap:
The grade detection can also be considered for the brain disease diagnosis.
The features considered can be reduced for reducing the training time levels.
[7]. Content-Based Brain Tumor Retrieval for MR Images Using Transfer Learning.
Automatic Content-Based Image Retrieval was used to find brain tumors in T1-
weighted, contrastenhanced MRI images.
Low-level MRI visual information and highlevel MRI visual information have
semantic gaps that must be filled. CBIR systems have significant challenges in this
area.
Typical feature extraction methods use a few generated features to fill in the gaps
between the low and high level features that are being extracted.
Research Gap:
The CBIR model used transfer learning that considers more classifier learning models
that increases the complexity levels of the model.
The linked classification techniques can be applied to get better accuracy rate.
[8]. Comparison of Feature Selection Methods and Machine Learning Classifiers for
Radiomics Analysis in Glioma Grading.
Using radiomics characteristics for feature selection and classification, the author
proposed assessing the predicted accuracy of different glioma grading systems using
their model. Patients with 210 GBMs and 75 LGGs who had MRIs were studied for
quantitative radiomics characteristics.
A ten-fold cross validation process and a percentage split test were used to compare
the diagnostic performance of sixteen feature selection and fifteen classification
algorithms.
Research Gap:
The comparison performed observed that every model has their limitations like
complexity levels, more training time, noisy data that need to be observed in
designing a model for better outcomes.
[9]. A Novel Approach to Improving Brain Image Classification Using Mutual
Information Accelerated Singular Value Decomposition.
A new technique proposed used Mutual Information-Accelerated Singular Value
Decomposition (MI-ASVD) to choose a relevant subset of features for the classifier's
input.
It's possible to apply an algorithm to determine whether or not an MRI brain scan
shows signs of cancer.
When it comes to tumor localization, a six-stage method is applied. This method
includes pre-processing, feature extraction, MI-ASVD analysis, and classification.
Research Gap:
The singular value decomposition technique performance is better but the image
classification model can be enhanced by applying optimization techniques that can be
used for enhanced performance levels.
[10]. Learning Methods of Convolutional Neural Network Combined With Image
Feature Extraction in Brain Tumor Detection.
Because CAD technology is underutilized in brain tumor detection terminals,
reducing interference that can distort diagnostic results is difficult.
The author employed convolutional neural networks in conjunction with MRI to
construct an algorithm for recognizing the features of brain tumors, such that
computer-aided techniques can assist brain tumor detection.
Research Gap:
The hidden layers can be increased for more training accuracy levels and the noise
elimination should be strictly considered for getting accurate results.
[11]. Combining Deep and Handcrafted Image Features for MRI Brain Scan
Classification.
Deep learning algorithms in medical imaging have recently resulted in considerable
advancements in the categorization and diagnosis of challenging disorders, such as
malignant gliomas of the brain.
The author presented a deep learning feature extraction strategy for extracting
relevant characteristics from MRI brain imaging.
To acquire handcrafted qualities, Modified Grey Level Co-Occurrence Matrix
(MGLCM) is employed in parallel.
Support vector machines (SVMs) are used to integrate extracted relevant
characteristics with produced features to improve the accuracy of MRI brain image
categorization.
Research Gap:
The feature vector can be optimized and the complexity levels of the model can be
reduced by applying dimensionality reduction and the Denoising application for
accurate classification.
The training samples can be still increased to get accurate precision levels.
[12]. Exploring Task Structure for Brain Tumor Segmentation from Multi-Modality
MR Images.
A task-structured brain tumor segmentation network (TSBTS net) with a task-
structured model is proposed in this research, which author termed the task-modality
structure and the task-task structure, respectively, as a result of these two qualities.
In order to examine the task-modality structure, the author devised a modality-aware
feature embedding approach to infer the significant weights of the paradigm data
during network learning.
Research Gap:
The weakly supervised technique can be integrated with the task structure
segmentation for better segmentation accuracy levels.
The task modality feature considers the features that are redundant that effects the
accuracy levels. The feature redundancy can be avoided to improve the accuracy
rates.
[13]. Combining Noise to Image and Image toImage GANs: Brain MR Image
Augmentation for Tumor Detection.
Using a Generative Adversarial Networks (GAN) based Data Augmentation (DA), the
author refined magnetic resonance images of the brain with and without tumors in two
steps:
Progressive Growing GAN (PGGANs) use a multistage noise-to-image GAN to
generate realistic/diverse 2562 X 256 pictures in order to generate highresolution MR
images.
Research Gap:
The image denoising process accuracy levels are not satisfactory.
The noise in images can be avoided by applying strong filtering techniques that need
to be considered.
[14]. A Hybrid Feature Extraction Method with Regularized Extreme Learning
Machine for Brain Tumor Classification.
For the construction of an accurate brain tumor classification algorithm, a regularized
extreme learning machine (RELM)-hybrid feature extraction technique is proposed.
As the first phase in this approach, a min-max normalization method is utilized to
boost contrast at the brain's edges and regions.
The features of the brain tumor are then extracted using a hybrid feature extraction
method.
Research Gap:
The feature extraction model is time consuming process that increases the load on the
system.
The regularized extreme learning model classification can be enhanced by considering
only relevant feature vector.
[15]. A Deep Learning Model Based on Concatenation Approach for the Diagnosis of
Brain Tumor.
The author proposed a strategy for detecting brain tumors early that extracts and
concatenates many levels of characteristics.
The use of two pre-trained deep learning models, Inception-v3 and DensNet201,
ensures the validity of this model.
These two models were used to evaluate two alternative scenarios for the
identification and categorization of a brain tumor.
The features from pre-trained Inceptionv3 models were obtained and concatenated
before concatenating these features for brain tumor classification.
Research Gap:
The concatenation approach performance is good however the values considered for
concatenation can be verified for removing the irrelevant set to get better accuracy.
The grade detection is not accurate because of considering the noisy irrelevant data.
REFERENCES:
[1] Syazwany, NurSuriza, Ju-Hyeon Nam, and Sang-Chul Lee. "MM-BiFPN: Multi-Modality
Fusion Network With Bi-FPN for MRI Brain Tumor Segmentation." IEEE Access 9 (2021):
160708-160720.
[2] Montaha, Sidratul, et al. "TD-CNN-LSTM: A Hybrid approach combining CNN and
LSTM to classify brain tumor on 3D MRI scans performing ablation study." IEEE
Access (2022).
[3] Deepa, B., et al. "Pattern Descriptors Orientation and MAP Firefly Algorithm Based
Brain Pathology Classification Using Hybridized Machine Learning Algorithm." IEEE
Access 10 (2021): 3848-3863.
[4] Ding, Yi, et al. "MVFusFra: A Multi-View Dynamic Fusion Framework for Multimodal
Brain Tumor Segmentation." IEEE Journal of Biomedical and Health Informatics 26.4
(2021): 1570-1581.
[5] Luo, Zhengrong, et al. "HDC-Net: Hierarchical decoupled convolution network for brain
tumor segmentation." IEEE Journal of Biomedical and Health Informatics 25.3 (2020): 737-
745.
[6] Khan, Protima, et al. "Machine learning and deep learning approaches for brain disease
diagnosis: principles and recent advances." IEEE Access 9 (2021): 37622-37655.
[7] Swati, ZarNawab Khan, et al. "Content-based brain tumor retrieval for MR images using
transfer learning." IEEE Access 7 (2019): 17809-17822.
[8] Sun, Pan, et al. "Comparison of feature selection methods and machine learning
classifiers for radiomics analysis in glioma grading." IEEE Access 7 (2019): 102010-102020.
[9] Al-Saffar, Zahraa A., and TülayYildirim. "A novel approach to improving brain image
classification using mutual information-accelerated singular value decomposition." IEEE
Access 8 (2020): 52575-52587.
[10] Wang, Weiguang, et al. "Learning methods of convolutional neural network combined
with image feature extraction in brain tumor detection." IEEE Access 8 (2020): 152659-
152668.
[11] Hasan, Ali M., et al. "Combining deep and handcrafted image features for MRI brain
scan classification." IEEE Access 7 (2019): 79959-79967.
[12] Zhang, Dingwen, et al. "Exploring task structure for brain tumor segmentation from
multi-modality MR images." IEEE Transactions on Image Processing 29 (2020): 9032-9043.
[13] Han, Changhee, et al. "Combining noise-to-image and image-to-image GANs: Brain MR
image augmentation for tumor detection." Ieee Access 7 (2019): 156966-156977.
[14] Gumaei, Abdu, et al. "A hybrid feature extraction method with regularized extreme
learning machine for brain tumor classification." IEEE Access 7 (2019): 36266-36273.
[15] Noreen, Neelum, et al. "A deep learning model based on concatenation approach for the
diagnosis of brain tumor." IEEE Access 8 (2020): 55135-55144.
[16] Devi, Polagani Rama, and SrikanthVemuru. "Brain Tumour Detection with Feature
Extraction and Tumour Cell Classification Model using Machine Learning–A Survey." 2022
International Conference on Electronics and Renewable Systems (ICEARS). IEEE, 2022.
[17] Bodapati, Nagaeswari, et al. "Brain Tumor Detection on MR Images Using Improved
Support Vector Machine." 2022 International Conference on Electronics and Renewable
Systems (ICEARS). IEEE, 2022.