ProductsLogo
LogoStudy Documents
LogoAI Grader
LogoAI Answer
LogoAI Code Checker
LogoPlagiarism Checker
LogoAI Paraphraser
LogoAI Quiz
LogoAI Detector
PricingBlogAbout Us
logo

Deep Learning for Brain Tumor Segmentation: A Literature Review

Verified

Added on  2024/07/01

|41
|7683
|343
AI Summary

Contribute Materials

Your contribution can guide someone’s learning journey. Share your documents today.
Document Page
Literature Review (Secondary Research) Template
Student Name & CSU ID

Project Topic Title
Deep learning neural networks for medical image segmentation of brain tumor diagnosis
Version 1.0 _ Week 1 (5 Journal Papers from CSU Library)

1

Reference in APA format that will be in

'Reference List'

Zhao, X., Wu, Y., Song, G., Li, Z., Zhang, Y., & Fan, Y. (2018). A deep learning model integrating FCNNs and

CRFs for brain tumor segmentation. Medical image analysis, Vol. 43, pp. 98-111.

Citation that will be in the content
(Zhao, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S136184151730141X/1-s2.0-

S136184151730141X-main.pdf?

_tid=2da61e1d-7961-4740-9761-

0599e1e83c8b&acdnat=1532757075_b987

58be0e1f74b440498df61796621d

Q1
Conditional random fields, Brain tumour segmentation, Deep
learning, Fully convolutional neural networks

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:
Integrating
Conditional Random Fields (CRFs) and

Fully Conventional Neural Networks

(FCNNs)

Problem:
The efforts have been done for
developing automatic or semi-automatic brain

tumour segmentation techniques and the

preferred method is MRI. However, there are

few challenges of using MRI’s such as gliomas

ï‚·
Voting based fusion method
ï‚·
Image patches
ï‚·
Image Slices
ï‚·
Recurrent Neural Networks (RNNs)
ï‚·
TC1 and TC2 scans
1

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Tools: 2D image patches, Image slices,
voting based fusion method, BRATS

Applied Area:
Brain tumour diagnosis
treatment and planning

may appear similarly to gliosis and stroke in

MRI

Goal:
The aim is to develop a model based on
deep learning based segmentation model by

using 2D image patches.

The Process (Mechanism) of this Work; The process steps of the Technique/system

Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Imaging Data:
The first step is to obtain the imaging data from

BRATS 2015, BRATS 2016

It helps to collect basic information about the

MRI results that scans diagnosed patients with

gliomas

Only small volume of data is collected for

analysis

2
Brain tumour segmentation using FCNNs:
In this, the extracted image patches are

categorised by trained CNNs.

It reduces the issue of sample imbalance at the

testing stage

The relationship among the patches can be

lost while using patch-based segmentation

technique

3
Pre-processing of imaging data:
Using robust intensity normalisation technique

in order to compare the MRI scans and using

N41TK for correcting MRI data

By correcting the bias in the MRI data,

accurate results are obtained.

The results may not be accurate if the data is

incorrectly modified.

4
Comparing the results:
The results obtained are compared with the MRI

data for checking the accuracy

It helps to ascertain the viability of using the

proposed method

NA

Validation Criteria (Measurement Criteria)

2
Document Page
Dependent Variable Independent Variable
ï‚·
Brain tumour segmentation  Voxel’s label
ï‚·
Calculation of standard deviation ï‚· Collected MRI scans
Input and Output
Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
Three segmentation

models are trained by

using 2D image

patches and

segmenting the brain

tumours using voting

based fusion method.

The results indicate

that the proposed

methods eliminate the

bias of MRI scan and

increase the efficiency

of the diagnosis.

The proposed method of training FCNNs and CRFs

helps to differentiate the gliosis and gliomas which

helps to provide appropriate diagnosis for cancer

patients.

The major drawback of applying this technique is

that the segmentation performance of the trained

network gets affected because the numbers of

pixels varies according to their class in the image

slices.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

The current solution that is training FCNNs and

CRFs by using 2D image patches and segmenting

the images obtained from MRI scans.

The accuracy of the results obtained can be

evaluated from examining the scans which provides

appropriate analysis of the status of brain tumour.

There is a growing trend for using advanced

techniques for providing accurate diagnosis for

cancer. This research helps to understand the

techniques that can be used for developing future

technology in medical care.

Diagram/Flowchart

3
Document Page
Figure: Flowchart of the proposed deep learning model integrating FCNNs and CRFs for brain tumor segmentation
4

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Figure: The network structure of deep FCNNs
5
Document Page
2
Reference in APA format that will be in

'Reference List'

Soltaninejad, M., Yang, G., Lambrou, T., Allinson, N., Jones, T. L., Barrick, T. R., Howe, F.A., & Ye, X.

(2018). Supervised learning based multimodal MRI brain tumour segmentation using texture features from

supervoxels. Computer methods and programs in biomedicine, Vol. 157, pp. 69-84.

Citation that will be in the content
(Soltaninejad, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S016926071731355X/1-s2.0-

S016926071731355X-main.pdf?

_tid=4a4dd16b-bfb6-41e4-b262-

b20025595ecb&acdnat=1532759931_766b

835216f08196959681dfbbd7ba82

Q1
Diffusion tensor imaging, Supervoxel, Brain tumour
segmentation, Multimodal MRI

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:
3D
supervoxel based learning method

Tools:
Generating supervoxels, Gabor
Filters, random forests (RF) classifier

Applied Area:
Brain tumour segmentation
Problem:
Because of numerous types of
tumours types, is becomes difficult for

segmenting the brain tumour in MRI.

Goal:
The aim is to develop a 3D supervoxel
based learning method in order to accurately

identify the type of tumour after conducting

MRI scans.

ï‚·
Multimodal MRI images
ï‚·
Isotropic component
ï‚·
Anisotropic component
ï‚·
Diffusion Tensor Imaging (DTI)
ï‚·
Balanced Error Rate
6
Document Page
The Process (Mechanism) of this Work; The process steps of the Technique/system
Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Generation of Supervoxels:
The multi-model MRI dataset is used for

generating supervoxels

It helps to collect and store the MRI images.
The number of images used for analysing
results are limited

2
Calculating features of Supervoxels:
In this step, the features such as g histograms of

texton descriptor are calculated using Gabor

filters.

This process enables to calculate the features

of each supervoxel of different orientations.

Although MRI data is filtered, the accuracy

of using Gabor filters cannot be justified as

other techniques can be used.

3
Extracting statistical features:
After obtaining the images in the required

format, the statistical features are extracted.

This helps to analyse the MRI data accurately
NA
4
Entering the features in Random Forests
(RF) classifier:

The statistical features are entered into RF

classifier for classifying each supervoxel as

healthy brain tissue, tumour core, and oedema.

By classifying the data, the accuracy of the

results can be enhanced.

NA

Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
ï‚·
Training sample output ï‚· Structures of randomised tree in RF classifier
ï‚·
Generating supervoxels ï‚· Data obtained from MRI dataset.
7

Paraphrase This Document

Need a fresh take? Get an instant paraphrase of this document with our AI Paraphraser
Document Page
Input and Output Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
The MRI images are

collected for analysis

and supervoxels are

generated. Before

entering the data into

RF classifier, Gabor

Filters are used for

classifying the data.

The results obtained

by performing the

tests indicate that the

images provide

accurate results as to

the condition of the

tumour.

With the increasing number of cases related to brain

tumour, it is essential for developing a technology

which helps to accurately identify the tumours for

providing appropriate diagnosis.

The major drawback of using supervoxel is that

there is a minimum criterion that has to be fulfilled

with regards to its parameters.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

The proposed
3D supervoxel based learning
method helps to eliminate the issues of MRI

scans. This is essential for ascertaining the actual

nature of tumour.

The MRI images are accurately classified as
healthy
brain tissue, tumour core, and oedema which is

helpful while designing the type of diagnosis that is

to be provided to the patients.

3D supervoxel based learning method is a solution

for the issues currently being faced related to MRI

scans. This helps to develop an effective neural

networks and image processing technique.

Diagram/Flowchart

Figure: Flowchart of the proposed multimodal MRI segmentation method for segmentation of brain tumour.

8
Document Page
Figure: Flowchart of the multimodal normalization and histogram matching of the
MR dataset

9
Document Page
3
Reference in APA format that will be in

'Reference List'

Pinto, A., Pereira, S., Rasteiro, D., & Silva, C. A. (2018). Hierarchical Brain Tumour Segmentation using

Extremely Randomized Trees. Pattern Recognition.

Citation that will be in the content
(Pinto, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S0031320318301699/1-s2.0-

S0031320318301699-main.pdf?

_tid=6cec5e43-2f0c-46c9-941d-

276563fcb78d&acdnat=1532759868_53c8

b955cadbe0e1d6b5046a6a61f0c7

Q1
Magnetic resonance imaging, Extremely randomized trees,
Machine learning, Image segmentation

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:
Automatic
hierarchical brain tumour segmentation

using Extremely Randomized Trees (ERT)

Tools:
Binary classification, morphological
refinement, N4ITK method, histogram

matching method

Applied Area:
Brain Tumour
Segmentation

Problem:
MRI is the commonly used method
for identifying gliomas. But this requires

manual segmentation which is difficult as well

as time-consuming task.

Goal:
The aim is to develop an automatic
tumour segmentation technique using

Extremely Randomized Trees for accurate

analysis of the condition of tumour.

ï‚·
BRATS 2013
ï‚·
Bias field distortion
ï‚·
Intensity normalisation
ï‚·
Data augmentation
ï‚·
2-dimensional patches
ï‚·
Tumour detection
10

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
The Process (Mechanism) of this Work; The process steps of the Technique/system
Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Collecting the data:
The MRI images are collected for analysing the

accuracy of using Extremely Randomized Trees.

It helps to ascertain the accuracy that can be

achieved by using the proposed technique.

A large sample has to be analysed for

accurate analysis of the data.

2
Correcting the image bias:
The imperfections in the MRI images are

corrected using N4ITK method

This minimises the distortions in the images

and enhances the accuracy of the results.

Although the imperfections are reduced, it

does not help to completely eliminate wrong

data.

3
Intensity Normalisation:
The histogram matching method is used during

this stage for ensuring identical intensity

distribution of the same sequence

This enables reducing the differences between

the images

NA

4
Analysis of the data:
The results are compared with existing

techniques used such as T1-weighted, and d

Fluid Attenuation Inversion Recovery

This helps to understand the viability of using

this technique.

NA

Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
ï‚·
Tumour classifications ï‚· Complexity of the tumour
ï‚·
Calculation of intensities ï‚· intensity variability of the percentiles of the training images
Input and Output
Critical Thinking: Feature of this work, and Critical Thinking: Limitations of the research
11
Document Page
Why (Justify) current solution, and Why (Justify)
Input (Data)
Output (View)
For arriving at the

desired results, the

MRI scan images are

collected. The bias in

the images are

normalised before

entering the same into

Random Forest (RF

classifier).

The output is verified

by considering the

metrics, Positive

Predictive Value,

Dice Similarity

Coefficient and

Sensitivity.

The proposed technique helps to analyse the types

of tumour by eliminating biases. Hence, accuracy of

the diagnosis is ensured.

The intensity range of the necrosis tissues would

have been shifted affecting the training of

classifier.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

The use of the proposed technique reduces the

manual segmentation of tumour tissues thereby

enhancing the accuracy of the results.

The images are clearer and the affected brain area is

highlighted along with corresponding tissues.

The proposed research will help to develop a

novel technique which can be used for developing

advanced methods for detecting brain tumours.

Diagram/Flowchart

12
Document Page
Figure: Overview of the proposed method
Figure: T1c sequence of a HGG patient in the different stages of the pre-processing step. In the bias field heat map, warmer colours represent a

higher magnitude of the bias field

13

Paraphrase This Document

Need a fresh take? Get an instant paraphrase of this document with our AI Paraphraser
Document Page
4
Reference in APA format that will be in

'Reference List'

Gibson, E., Li, W., Sudre, C., Fidon, L., Shakir, D. I., Wang, G., Gray, R., Doel, T., Yipeng, H., & Whyntie, T.

(2018). NiftyNet: a deep-learning platform for medical imaging. Computer methods and programs in

biomedicine, Vol. 158, pp. 113-122.

Citation that will be in the content
(Gibson, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S0169260717311823/1-s2.0-

S0169260717311823-main.pdf?

_tid=e9244757-9c0d-4c85-8389-

29326062c28c&acdnat=1532759777_7f87

d9a18241216c2289274709be4746

Q1
Convolutional neural network, Image regression, Deep
learning, Generative adversarial network

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:
Using
TensorFlow framework for developing

NiftyNet platform.

Tools:
Image regression, segmentation of
multiple abdominal organs, TensorBoard

visualization

Problem:
Although deep learning platforms
are developed, they do not provide specific

functionality which requires huge

implementation efforts.

Goal:
The aim is to simplify the learning
platforms so that further research can be

conducted for improving the quality of the

medical services.

ï‚·
Image regression
ï‚·
Segmentation
ï‚·
Brain magnetic resonance images
ï‚·
Specified anatomical poses.
ï‚·
Representation learning applications
14
Document Page
Applied Area: Developing deep learning
platforms.

The Process (Mechanism) of this Work; The process steps of the Technique/system

Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Developing the infrastructure:
This stage involves designing various functions

such as image segmentation, generation and

regression.

This will help the researchers for developing

advanced functions.

This involves expert knowledge and

incurring huge expenses

2
Gathering the required data:
For developing an effective deep learning

platform, it is necessary to gather a large number

of images.

Collecting and storing information enables to

analyse different types of data.

The challenge is to store the information at

one location.

3
Formatting the images:
Storing the images in a suitable format

depending on the type of application being used.

Formatting ensures easy integration of images
Storing the images in unsupported format
may result in losing valuable data.

4
Analysing the stored data:
The stored images are reviewed and the time

taken for accessing these images is ascertained.

This helps to ascertain the effectiveness of the

proposed system

NA

Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
ï‚·
Data augmentation ï‚· Type of images saved
ï‚·
Sampling and output handling ï‚· Training performed
15
Document Page
Input and Output Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
For developing

NiftyNet platform, the

data has to be

collected and stored

for future analysis and

research.

The output is that the

data is easily

accessible and the

suitable information

can be obtained at

minimum duration of

time.

NiftyNet platform provides the solution faced for

developing a suitable deep learning platform for

improving medical care services.

The major concern is the storing large volumes of

data at one location because usually these require

specialised storage devices.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

NiftyNet helps to develop deep learning platform

by collecting the required information. Also, the

images helps to analyse the opportunities that can

be utilised for improving medical care.

The images stored are easily available and

accessible for further research.

NiftyNet helps to understand the importance of

developing deep learning platform which is

helpful for developing new methods that can be

used for brain tumour diagnosis.

Diagram/Flowchart

16

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Figure: Data flow implemented in typical deep learning projects
17
Document Page
Figure: A brief overview of NiftyNet components
5

Reference in APA format that will be in

'Reference List'

Charron, O., Lallement, A., Jarnet, D., Noblet, V., Clavier, J. B., & Meyer, P. (2018). Automatic detection and

segmentation of brain metastases on multimodal MR images with a deep convolutional neural network.

Computers in biology and medicine, 95, 43-54.

Citation that will be in the content
(Charron, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-
Q2 Convolutional neural network, Brain metastases, Magnetic
18
Document Page
com.ezproxy.csu.edu.au/
S0010482518300325/1-s2.0-

S0010482518300325-main.pdf?

_tid=8ab2cdc7-fd92-45a8-bf3d-

da4e799ed122&acdnat=1532759589_c1f7e

affa2beb33b36a92103aeac0f7a

resonance imaging,

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:
Automatic
segmentation of brain metastases

Tools:
3D convolutional neural network
(DeepMedic), 3DT1Gd MRI, Brain

Extraction Tool

Applied Area:
Brain tumour diagnosis
Problem:
There are various methods that have
been developed for segmentation of gliomas,

but there are no comprehensive studies for

metastases.

Goal:
The aim is to detect and segment brain
metastases in MRI by using existing 3D

convolutional neural network

ï‚·
Different MRI modalities
ï‚·
Detection and Segmentation
ï‚·
Necrotic parts
ï‚·
Spatial resolution
The Process (Mechanism) of this Work; The process steps of the Technique/system

Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Adapting network parameters:
This involves adapting network parameters for

brain metastases.

This is helpful while evaluating the results
There are no standard techniques and hence,
parameters have to be carefully selected.

2
Using MRI modalities: It helps to evaluate the network performance
related to segmentation and detection

Accuracy depends on the collected MRI

data.

19

Paraphrase This Document

Need a fresh take? Get an instant paraphrase of this document with our AI Paraphraser
Document Page
In this, using single or combined use of various
MRI modalities is explored.

3
Using additional database:
The additional information is used for separating

active parts of metastases from necrotic ones.

This reduces the risk of prescribing incorrect

diagnosis.

NA

4
Evaluating the effectiveness:
The results are evaluated for ascertaining the

benefits of using the proposed approach.

This enables for conducting further research.
NA
Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
ï‚·
Segmentation maps ï‚· Performance of the network
ï‚·
Adapting network parameters ï‚· MRI database
Input and Output
Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
For studying the

usefulness of

DeepMedic, it is

essential for using

MRI modalities for

detecting brain

metastases.

The results showcased

that the use of

multimodal MRI

enables to detect brain

metastases.

Brain metastases have to be detected since it affects

cancer patients and early diagnosis helps to cure

cancer. Hence, by using the proposed technique,

metastases can be detected.

The major concern with the research is the very

small size of the dataset. This affects the accuracy

of the results obtained.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
20
Document Page
for your project
3D convolutional neural network can be used for

detecting
metastases which is essential because of
increasing number of cancer patients.

The effectiveness of this technique helps to detect

and segment brain metastases on multimodal MRI.

The use of
3D convolutional neural network
(DeepMedic) while using multimodal MRI

enables to develop technologies for providing

medical care for patients suffering from brain

tumour.

Diagram/Flowchart

Figure: From left to right: cerebral metastasis visualized on 2DT1, 3DT1Gd and 2DFLAIR MRI

21
Document Page
Fig: Histogram of the brain metastasis diameters and volumes included in our database.
22

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Version 2.0 _ Week 2 (10 Papers = 5 NEW Journal Papers + Updated pervious 5 papers that you worked on them in Version 1.0)
6

Reference in APA format that will be in

'Reference List'

Hussain, S., Anwar, S. M., & Majid, M. (2018). Segmentation of glioma tumours in brain using deep

convolutional neural network.
Neurocomputing, 282, 248-261.
Citation that will be in the content
(Hussain, et.al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S0925231217318763/1-s2.0-

S0925231217318763-main.pdf?

_tid=7abf2a11-30ee-4ea4-9970-

2a9421028c2b&acdnat=1532766178_ce0b

22e219896889e97dc491477f9126

Q1
Segmentation, Convolutional Neural networks, Brain tumour,
Deep learning

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:

Automated brain tumor segmentation

algorithm

Tools:

BRATS 2013, BRATS 2015 datasets

Applied Area:

Medical industry

Problem:

The main problem in the detection of brain

tumour is the using the segmentation approach.

Goal:

The goal of the research is to develop a model

that will detect the tumour and diagnose it at

the early stage so that it can be cured.

ï‚·
Magnetic resonance imaging (MRI)
ï‚·
computed tomography (CT)
ï‚·
positron emission tomography (PET)
ï‚·
magnetic resonance spectroscopy (MRS)
ï‚·
cerebrospinal fluid (CSF)
ï‚·
BRATS
ï‚·
MICCAI
The Process (Mechanism) of this Work; The process steps of the Technique/system

23
Document Page
Process Steps Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Pre-processing-
A technique which is non-uniform and non-

parametric is used here to normalize the

intensity and removal of artefact in MRI.

Artefacts such as motion and in homogeneity

of the field is removed

NA

2
Convolutional Neural Networks-
The convolutional layers are overlapped to form

hierarchical fashion for the feature of maps.

It helps in development of image recognition.
Computation requirements
3
Post-processing-
Use of morphological operators is done here to

enhance segmentation results

Removal of small and false positives is done

around the corners of a image which is

segmented

NA

Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
Scanning the skull for imaging
MRI
Deals with multiple predictions
CRF
Input and Output
Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
Input in this particular

research is to develop

a model help the

Output from this

particular research is

that an automated

Tumor regions generally are placed near white

matter fiber and they have blurry boundaries which

makes it difficult to detect. The correct

segmentation of the brain tumor is very crucial in

the diagnosis and the treatment planning. Therefore,

The main limitation of this scheme is that its

capacity is limited. If the capacity of the system is

increased then the performance could be

enhanced. Further, this area of study can be

applied to other fields such as analysis of

24
Document Page
automation in the
segmentation of the

brain tumour.

brain tumour

segmentation scheme

is proposed that is

based on deep

convolutional neural

network.

automated brain tumor segmentation technique is

proposed to improve the segmentation of the tumor.

biomedical image analysis and the segmentation

of brain lesion.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

Introduction of automated segmentation of the

tumor in brain can help in the diagnosis of the

tumor and future treatment of this tumor. The

technique on which the segmentation is based on

is the Deep Convolutional Neural Networks.

In this system, the two networks are placed one

above another to make a new Tlinear nexus

architecture which can be used to get the best

results out of all the architectures present. The first

network contains the layers placed parallel and the

second network contains the layers linearly.

After a number of studies and evaluation done on

different architectures, the various settings have

been explored validation is given to the use of

parallel paths in an architecture. The results done

on two datasets i.e. BRATS2013 and BRATS2015

have shown that the performance is increased in

comparison with other state-of-art techniques.

Diagram/Flowchart

25

Paraphrase This Document

Need a fresh take? Get an instant paraphrase of this document with our AI Paraphraser
Document Page
Figure: Block diagram of the proposed methodology
7

Reference in APA format that will be in

'Reference List'

Rachmadi, M. F., Valdés-Hernández, M. D. C., Agan, M. L. F., Di Perri, C., Komura, T., & Alzheimer's

Disease Neuroimaging Initiative. (2018). Segmentation of white matter hyperintensities using convolutional

neural networks with global spatial information in routine clinical brain MRI with none or mild vascular

pathology.
Computerized Medical Imaging and Graphics, 66, 28-43.
Citation that will be in the content
(Rachmadi, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S089561111830082X/1-s2.0-

S089561111830082X-main.pdf?

_tid=f3f69cb0-469c-4b3a-9dfd-

5c67f9ac14e2&acdnat=1532767949_bff14

e82310fe9bdbe76018662fbbce3

Q2
Mild cognitive impairment, Convolutional neural network,
White matter hyper intensities, Alzheimer's disease, Deep

learning, Segmentation, Global spatial information

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:

Convolutional neural network (CNN)

Tools:

Linear nexus (LN), Two-path linear nexus

(TLinear)

Problem:

Small area or volume of white matter hyper

intensities are the main problem occurred in

the brain MRI and authors are focused on the

same.

Goal:

ï‚·
Glioblastoma multiforme (GBM)
ï‚·
Markov random field (MRF)
ï‚·
Local histograms
ï‚·
Raw pixel values
ï‚·
Fuzzy C-means (FCM)
ï‚·
Two-path Nexus (TPN)
ï‚·
Linear nexus (LN)
26
Document Page
Applied Area:
Medical industry

To enhance the brain MRI procedure by

introducing convolutional networks scheme.

The Process (Mechanism) of this Work; The process steps of the Technique/system

Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Training and testing processes-
In this process, all the gathered database got

tested and used for the training session and

further got evaluated by Fazekas score.

Accuracy of Machine learning algorithm
NA
2
Parameter setup-
All the important parameters for machine

learning got evaluate before the training process.

Accurate classification by Radial basis (RBF)

kernel

NA

3
Evaluation-
In the process, results of medical image

segmentation got evaluated.

Precision of the equation
NA
Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
T1 and T2
Nature of the tissues imaged
WMH partly
On their location in the brain
statistical analyses
grey-level histogram
Input and Output
Critical Thinking: Feature of this work, and Critical Thinking: Limitations of the research
27
Document Page
Why (Justify) current solution, and Why (Justify)
Input (Data)
Output (View)
Scheme of learning

which help to getting

information about

features of MIRs.

That could help to

enhance MIR

procedure in brain

tumour decease.

To develop and use

the deep learning

algorithms to for

betterment of

segmentation of

WMH.

Feature of the work is various deep learning

algorithms such as SVR and RF. These proposed

algorithms assist in brain related decease and

segmentation of WMH

The proposed techniques and algorithms used in

the proposal are CNN -GIS and they need to be

evaluated in the various fields and advancement

also required in the same.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

Research was focused on the brain tumor MIRs

and the algorithms which made the process of

WMH segmentation ease and advance, have been

discussed (Rachmadi, et. al., 2018).

Brain patients and doctors are the end users of the

proposed algorithms as it assists process of WMH

segmentation and MRI of brain tumour.

CNNs are the effective and useful algorithm in the

way of WMH segmentations and also assisting in

routine clinical brain MRIs.

Diagram/Flowchart

28

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Figure: Example image of WMH visualisation in two different types of MRI structural sequences: T2 based-fluid attenuated inversion recovery
8

Reference in APA format that will be in

'Reference List'

Izadyyazdanabadi, M., Belykh, E., Mooney, M., Martirosyan, N., Eschbacher, J., Nakaji, P., ... & Yang, Y.

(2018). Convolutional neural networks: Ensemble modeling, fine-tuning and unsupervised semantic

localization for neurosurgical CLE images.
Journal of Visual Communication and Image Representation, 54,
10-20.

Citation that will be in the content
(Izadyyazdanabadi, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

Q1
Ensemble modelling, Confocal laser endomicroscopy, Neural
network, Surgical vision, Unsupervised localization, Brain

29
Document Page
S1047320318300804/1-s2.0-
S1047320318300804-main.pdf?

_tid=a27da700-7e16-423e-b435-

1861656b6fe2&acdnat=1532770818_9e6d

6243cbedae01e13d217defad7166

tumor

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:

Deep learning image selection algorithm

Tools:

Confocal imaging instrument instrument,

intraoperative data collection process

Applied Area:

Medical industry

Problem:

Confocal Laser Endomicroscopy (CLE) is the

advanced method in the brain tumor imaging.

But sometimes a large number of CLE images

can be distorted.

Goal:

The goal is to automatically select a diagnostic

image for examination.

ï‚·
Fluorescein sodium (FNa)
ï‚·
Rectified linear unit (ReLU)
ï‚·
Normalization layer
ï‚·
Activation layer
ï‚·
Stochastic Gradient Descent (SGD)
ï‚·
Convolutional layers
The Process (Mechanism) of this Work; The process steps of the Technique/system

Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Dataset Preparation-
The 20,734 images from 74 tumor cases were

recorded

The images from different cases can be used to

understand the diagnosis easily

Professional trained staff required for

efficient work.

30
Document Page
2 Model Development-
After the splitting of the data, a patient related

five cross validation is developed for the model.

Different models are prepared to find the

optimal method

NA

3
Inter-observer Study-
The test is done on the different test datasets

developed.

The comparison is made between the human-

human and model-human inter-observer

agreements.

NA

4
Unsupervised histological feature
localisation-

The examination of neural activation is done at

two sites.

This detection feature could help the doctor in

treatment.

NA

Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
Determine the sensitivity
Reciever Operating characteristic (ROC) curve
Optimize the predictions
Stochastic Gradient Descent (SGD)
Input and Output
Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
The input in the

research is that a

model needs to be

prepared where the

image suitable for

diagnosis can be

He output is that the

deep learning model

is suggested so that

the diagnosable

images can

automatically be

CLE is used widely because of its ability to image

histopathological features of a tissue in real time.

The motion and the artifacts of blood that are there

in the images can be an obstruction in revealing

meaningful data from image. This necessity calls

for the selection of best image out of non-

diagnosable images to help the doctors.

The suggested model is at the beginning stage of

the evolution. Some techniques like combination

of different abstract features can be useful for

visual recognition. Thus these advanced features

can be added to the present system to increase the

efficiency and accuracy of the system.

31

Paraphrase This Document

Need a fresh take? Get an instant paraphrase of this document with our AI Paraphraser
Document Page
automatically
selected.

detected.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

The research suggests a deep learning algorithm

which will be able to automatically detect the

diagnostic CLE image for the brain tumor. The

datasets which are manually interpreted in-house

that are used here to test and train this approach.

The result of the research shows that both deep

fine-tuning and building a group of models can be

beneficial in enhancing the performance. Also their

combination could attain largest accuracy.

The proposed model in the research is also able to

confine some of the histological features of the

images to be diagnostic. The deep learning models

have helped achieving the accuracy in detection of

diagnostic images.

Diagram/Flowchart

32
Document Page
Figure: A schematic diagram of interobserver study.
9

Reference in APA format that will be in
Chen, H., Dou, Q., Yu, L., Qin, J., & Heng, P. A. (2018). VoxResNet: Deep voxelwise residual networks for
33
Document Page
'Reference List' brain segmentation from 3D MR images. NeuroImage.
Citation that will be in the content
(Chen, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S1053811917303348/1-s2.0-

S1053811917303348-main.pdf?

_tid=d044cee7-ffbf-4b9f-96c7-

2cbf09171f71&acdnat=1532774784_8a4c0

58e26ed5ffbd3008ee39f40f6fd

Q1
Brain segmentation, Multi-level contextual information,
Auto-context, Residual learning, Convolutional neural

network, Multi-modality, 3D deep learning

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:

Voxelwise Residual Network (VoxResNet)

Tools:

3D magnetic resonance (MR)

Applied Area:

Medical industry

Problem:

Segmentation of key brain tissues from 3d

images can be very beneficial for the medical

diagnosis. The manual segmentation is very

complicated as the anatomy of brain is hard to

understand.

Goal:

The goal is to propose a model which will help

in automated segmentation of the brain tissues.

ï‚·
CNN
ï‚·
C++
ï‚·
Matlab
ï‚·
Caffe library
ï‚·
3D magnetic resonance (MR)
ï‚·
Brain segmentation
ï‚·
3D deep learning
The Process (Mechanism) of this Work; The process steps of the Technique/system

34

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Process Steps Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Data Acquisition and Pre-Processing-
The datasets are acquired in this step

Evaluation of the algorithm is done
NA
2
Deep Residual Learning-
The multiple layers of neurons are stacked

hierarchically to form a low, middle and high

feature based representation.

It can improve the judgment capability of a

particular network.

NA

3
VoxResNet for volumetric image
segmentation-

2D deep residual network is extended to 3D

deep residual network to get more representative

data

Volumetric feature representation learning is

strengthened.

NA

4
Multi-Modality and Auto-context
information Fusion-

The volumetric based data is completely

acquired by using multiple imaging modalities.

Various tissues can be examined robustly.
Leaser accuracy
Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
C++ & Matlab
Caffe library
Input and Output
Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

35
Document Page
Input (Data) Output (View)
The input of this

research is to suggest

a model that will

automatically do the

segmentation of the

3D medical images.

The output that can be

obtained from this

research is that the

deep learning model

with a training

module is proposed

which can help in

auto-segmentation of

3D medical images.

The 3D magnetic resonance images can greatly

affect the diagnosis of the medical

neurodegenerative disease. The manual

segmentation of the 3D images can consume a lot

of time and it is also an extremely laborious work.

Moreover, manual segmentation is prone to errors.

Therefore, automated segmentation is highly

required in the medical industry.

The use of the current model is limited to some

parts of the diagnosis which makes it a limitation

of the model. The automated brain structure

segmentation can be offered to various fields like

neuro-imaging and neuroscience studies where it

is essential to have accurate segmentation.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

The research has proposed a model based on

VoxResNet which is a 3D residual network used

for automated segmentation of the brain structure

from 3D MR images. Various analyses have been

done on the capabilities in automated

segmentation. This model can also help the

advancement in brain structure segmentation

(
Chen, et. al., 2018).
The method proposed in the research has extended

the 2D residual network to the 3D residual network

to solve the issues that were faced during the

segmentation of volumetric data using a deeper

network than the other models. Multi-modal and

Multi-level contextual based information were

carefully integrated in the end-to-end based

network to increase performance.

The additional features of the proposed model are

the addition of auto-context version of VoxResNet

which can enhance its performance by integrating

low-level appearance based information, implicit

shape based information and the high-level

context. Various experiments and analysis proves

that the proposed model is one of the efficient

models proposed recently.

Diagram/Flowchart

36
Document Page
Figure: An overview of our proposed framework for integrating auto-context with multi-modality information
37

Paraphrase This Document

Need a fresh take? Get an instant paraphrase of this document with our AI Paraphraser
Document Page
10
Reference in APA format that will be in

'Reference List'

Raju, A. R., Suresh, P., & Rao, R. R. (2018). Bayesian HCS-based multi-SVNN: A classification approach for

brain tumor segmentation and classification using Bayesian fuzzy clustering.
Biocybernetics and Biomedical
Engineering
.
Citation that will be in the content
(Raju, et. al., 2018)
URL of the
Reference Level of Journal (Q1, Q2, …Qn) Keywords in this Reference
https://ac-els-cdn-

com.ezproxy.csu.edu.au/

S0208521617303923/1-s2.0-

S0208521617303923-main.pdf?

_tid=dff04457-be6b-4b96-9336-

6a2381155c83&acdnat=1532776500_f062

299f5f65b7d5eb23241808802155

Q3
Brain tumor classification, Bayesian fuzzy clustering, MRI
image, Support vector neural network, Brain tumor

segmentation

The Name of the Current Solution

(Technique/ Method/ Scheme/

Algorithm/ Model/ Tool/ Framework/ ...

etc )

The Goal (Objective) of this Solution &

What is the Problem that need to be solved

What are the components of it?

Technique/Algorithm name:

Bayesian HCS-based multi SVNN

classifier

Tools:

Harmony-Crow Search (HCS) algorithm,

BRATS database, machine learning

classification techniques

Applied Area:

Medical industry

Problem:

The segmentation of the brain structure is

advancement in technology in medical field

but the automatic segmentation is lacking to

identify which part of brain is tumorous and

which part is not.

Goal:

The goal is to automatically classify the

tumorous area and the non-tumorous area of

the brain using a model.

ï‚·
BRATS database
ï‚·
Fuzzy C-means clustering
ï‚·
Multi-SVNN classifier
ï‚·
Harmony Search Algorithm
ï‚·
Crow Search Algorithm (CSA)
ï‚·
Bayesian fuzzy clustering (BFC)
38
Document Page
The Process (Mechanism) of this Work; The process steps of the Technique/system
Process Steps
Advantage (Purpose of this step) Disadvantage (Limitation/Challenge)
1
Segmentation using the Bayesian Fuzzy
Clustering-

Identification of core and the edema regions.

Flexibility and the capability of the algorithm

are proved.

Require high accuracy

2
Feature Extraction-
Wavelet transform, scattering transform and

theoretical data is used to extract features.

Features are extracted for segmentation.
NA
3
Detection of tumour and tumour level-
The classification is done on the basis of

features extracted

The classification of the tumour is done
NA
Validation Criteria (Measurement Criteria)

Dependent Variable
Independent Variable
BFC model
Fuzzy data
Semi-automatic methods
User interaction
Optimization algorithm
Multi-SVNN classifier
Input and Output
Critical Thinking: Feature of this work, and
Why (Justify)

Critical Thinking:
Limitations of the research
current solution, and Why (Justify)

Input (Data)
Output (View)
The input of the
The output of this
In the advanced world, the MRI is not good enough

for detecting and segmenting the brain structure for

tumor. But using MRI images the segmentation can

The performance of the proposed framework is

very good based on the results and analysis done

on the model. But the limitation of this model can

39
Document Page
research is that a
model can be

proposed which will

automatically classify

the tumour region of

the brain and the level

of tumour.

research is that the

model is developed

which can

automatically look for

tumour region in brain

and also it can classify

the level of tumour.

be done. In the medical image processing,

segmentation plays an important role for diagnosis

and the analysis of images.

be that it is still in evolution field and more

accuracy and precision is needed in the model.

Moreover, additional datasets can be added to the

system so that the performance of the system can

be enhanced.

(
Describe the research/current solution) Evaluation Criteria How this research/current solution is valuable
for your project

The proposed model of Bayesian HCS-based

multi SVNN classifier is used to automatically

segment the brain structure and classify the level

of tumor based on the MRI images. The proposed

model is proved to be much accurate then its

counterparts from various experiments and

evaluation done on the system. The system is

much more efficient when compared with manual

process
(Raju, et. al., 2018).
The suggested model of the HCS algorithm is

integrated from CSA and the harmony search

optimization that engages itself into the calculation

of optimal weights to multi-SVNN. The proposed

system can easily and automatically classify the

tumor level of the brain using these algorithms.

The highly robust features of the system that are

obtained from the slices are employed to detect the

existence of non-tumor and edema tumors that

shows the clear-cut presence of abnormalities in

brain. The analysis done with the help of BRATS

database says that tumor classification obtains

accuracy of 0.93, sensitivity of 0.9690 and the

specificity of 0.9929.

Diagram/Flowchart

40

Secure Best Marks with AI Grader

Need help grading? Try our AI Grader for instant feedback on your assignments.
Document Page
Figure- Proposed HCS-based multi-SVNN for brain tumor segmentation and classification
41
1 out of 41
circle_padding
hide_on_mobile
zoom_out_icon
[object Object]

Your All-in-One AI-Powered Toolkit for Academic Success.

Available 24*7 on WhatsApp / Email

[object Object]