EARLY DETECTION OF CANCER USING

BACKPROPAGATION NEURAL NETWORKS AND

CONVOLUTION NEURAL NETWORKS

GUIDED BY PRESENTED BY

PROF.P.BRUNDHA P.Sathya
(CSE) M.E student (CSE)
2054005
OUTLINE:

 DOMAIN
 ABSTRACT
 INTRODUCTION
 OBJECTIVE
 LITERATURE SURVEY
 MOTIVATION
 PROPOSED SYSTEM
 SYSTEM REQUIREMENTS
 SYSTEM ARCHITECTURE
 MODULES
 RESULTS
 CONCLUSION
 FUTURE WORK
 REFERENCE
 PROJECT SCHEDULE
DOMAIN:
 MEDICAL IMAGE PROCESSING:
Medical image processing use 3D image datasets of human body from
CT (or)
MRI
Scan for research purpose or for surgical planning.
USES:
LUNG
BRAIN
NODULE
Treat Medical Conditions TUMOUR
DETECTION CLASSIFICAT
ION

Visualization of internal body

organs DL
Identify disease FETAL
THYROID LOCALIZ-
DIAGNOSIS ATION
Testing
BENEFITS OF
MEDICAL IMAGE PROCESSING
 ABSTRACT:
 Cancer is the most often Diagnosed Disease in children and adults.
 Cancer ……
Changes in group of normal cells  form Lumps  Tumour.
 When cancer care is delayed there is a lower chance of survival.
 The main aim is to detect cancer with high accuracy and provide cancer
treatment methods.

Detect various kinds of

cancer

Stages of cancer
(ABSTRACT) Cntd.,
 Accurate diagnosis….
Clinical Professionals relay on CAD technologies.
Deep Learning Techniques….
CAD systems have embraced deep learning techniques such as the
convolution neural network (CNN)
Training…..Back Propagation Methods.
Optimization……Genetic Algorithm.
 The model trained using GA outperforms the Adam optimizer with a high
classification accuracy.
INTRODUCTION:
 Cancer is the most often Diagnosed Disease in children and adults.
 Early detection of this condition through microarray technology significantly
increases the likelihood of survival.
 Micro array is a Laboratory tool used to detect the expression of thousands of
genes at the same time.
 The accuracy of this diagnosis, on the other hand, will be compromised
because of the type of image used.
 Throughout this project, I’ll learn how to train a convolution neural network
and improve prediction accuracy.
OBJECTIVE:
 To detect CANCER from MEDICAL micro array images.
 To extract relevant features from medical images using ANACONDA and
further classification using deep learning algorithms.
 Classification….
1. CNN
2.BPNN Algorithm.
 LITERATURE SURVEY:
Title Publication Authors Methods and Inference
Algorithm

Automatic IEEE Sunyi Zheng, Jiapan 3D CNN Need to improve

Pulmonary Transactions Guo, Xiaonan Cui, detection
Nodule Detection on Medical Raymond performance on
in CT scans using Imaging , N.J.Veldhuis, small nodules
CNN based on Volume:39, Matthijs Oudkerk, ground nodules
Maximum Issue:3, and Peter M.A van since they can
Intensity March 2020 Ooijen. easily overlooked
Projection during the
screening.

Graph IEEE LIMING XU 1,2, Graph To capture

Regularized Transactions XIAOPENG YAO1,3, Regularized semantic
Hierarchical on Medical LISHA ZHONG1 , Hierarchical relationship of
Diffusion Process Imaging , JIANBO LEI1,4, Diffusion Process medical images,
With Relevance Volume:40, AND ZHIWEI (GRHDP) which has been
Feedback for Issue:1, HUANG 1,3 method proved that it can
Medical Image March 2021 preserve the
Retrieval underlying
manifold
structure of
images
(LITERATURE SURVEY) cntd.,
Title Publication Authors Methods and Inference
Year Algorithm

AI in Medical IEEE Journal of Andreas S. semi-supervised Accuracy result

Imaging Biomedical and Panayides and prediction Time
Informatics: Health unsupervised is Extension
Current Informatics,Vol. methods
Challenges and 24,No.7,July
Future 2020.
Directions
A New Approach IEEE Sensors Guotai Wang , R-CNN Automatic
to Polyp Journal , Vol Wenqi Li , Maria Extraction ,
Detection by 21,No.10,May A. Zuluaga , Need to combine
pre-processing 15,2021. Rosalind Pratt, R-CNN and
of images and Premal A. Patel, Socially inspired
Enhanced Faster Michael Aertsen classify algm,
R-CNN. Clean up
disturbance due
to specular
reflection.
(LITERATURE SURVEY) cntd.,

Title Publication Authors Methods and Drawbacks

Year Algorithm

Deep Fuzzy IEEE Benjamin Hou , Fuzzy Neural Further need

Neural Transaction Bishesh Khanal , Network feature
Networks for s on Fuzzy Amir Alansary , selection using
Biomarker Systems, Steven evolutionary
Selection for Volume 28, McDonagh, Alice algorithms,
Accurate Issue 12, Davidson, different
Cancer 2020 statistical
Detection methods.
MOTIVATION:

 Computer supported analysis technique is a tool to support

radiologists
physicians.
 Computer supported analysis can be used to identify cancerous tumours at

primitive stage
 The process of differentiating at primitive stages is complex as the affected
person can observe the impacts only at later stages.
 Earlier Artificial neural network classification system is used. They provide
POOR POOR
DETECTION ACCURACY
(MOTIVATION) Cntd.,

 The ANN algorithm undergoes…

Hardware dependence
Unrecognized behavior of the network
Assurance of proper network structure
 ANN does not give proper explanation for output.
 ANN does not give proper explanation for structure.
Obtained by Trial and Error….
PROPOSED WORK:
 Back propagation is a supervised learning algorithm based on the gradient
descent of the artificial neural network (ANN)
 It computes the error function gradient and reference to the mass of the
neural network under consideration.
 The technique was used in the study of "backward progression of errors“
 Detect Cancer in early stages.
 The main aim is to

Detect various kinds of

cancer

Stages of cancer
(PROPOSED WORK)Cntd.,
Image preprocessing –Locating gene,
”MEDIAN FILTER”
Filter –1.Remove the noise
2.Improve image quality

Image segmentation-”Canny edge detection” Algorithm

1.Extract Useful Structural Information
2.Reduce data to be processed

Feature Extraction-”Gray Level Co-Occurrence Matrix”

Data Mining –Process of extracting and discovering
Pattern.

Image Classification – CNN,BPNN

1.Input Image
Assign Importance
Differentiate one from other
2.Accurate way to solve classification than others
MODEL ARCHITECTURE:

3D Image Data set

DATASET
Image Set Partition
PARTITION
Train Image Test Image
PRE
Image Pre-processing (Reshaping, stripping,
PROCESSING
Noise Removal)
PHASE

FEATURE
3D CNN Training GLCM Feature Extraction
EXTRACTION
PHASE
CNN Feature Extraction

BPNN Classifier
PREDICTION
PHASE
Malignan
Benign t
Normal
SYSTEM REQUIREMENTS:

Software Tool Anaconda IDE

Front End Tool Jupyter Notebook

Language Python

Datasets Real time Cancer Datas

OS Windows 10

HDD 500 GB

RAM 4 GB
MODULES:

 Input: Micro Array Image

 Output: Classified result along with accuracy
 Algorithm: CNN,BPNN
 System is divided into four modules
Module 1:Image Preprocessing.
Module 2:Image Segmentation.
Module 3:Feature extraction.
Module 4:Image Classification.
INPUT 3D IMAGE:
Module1:Image Preprocessing

Medical Image

Gaussian Filter
NOISE IMAGE

Processed Image

NOISE REMOVAL IMAGE

IMAGE PREPROCESSING:

 Crucial step in data analysis:

Data pre-processing or data cleansing

 Image processing performs refining image, locating genes, and extracting the raw
data.

 Aim:
Improvement of the image data that suppresses undesired distortions
Enhances some image features relevant for further processing and analysis task.

 Image pre-processing
Term for operations on images at the lowest level of abstraction.
GAUSSIAN FILTER

 GAUSSIAN BLUR
Blurring an image –Reduce noise
Reduce detail.
 It is a technique for converting a signal into elementary frequency
component.
 Usage
Image compression.
 The high-frequency component extracted by the Gaussian filtering from
the input image.
MODULE 2-IMAGE SEGMENTATION:

Processed image

Canny Edge Detection

SEGMENTED IMAGE

Segmented image
IMAGE SEGMENTATION:

Image segmentation :
Partitioning a digital image into multiple segments.
 Image segmentation Usage:
locate objects and boundaries (lines, curves, etc.) in images.
 Goal:
Simplify and/or change the representation of an image into
something that is more meaningful.
CANNY EDGE DETECTION:

 The Canny edge detector

Edge detection operator
Uses a multi-stage algorithm
To detect a wide range of edges in images.

 The Canny method finds edges by looking for local maxima of the
gradient of the image.

 The number of edges detected by Canny is much more than edges detected
by Sobel means the Canny edge detector works better than Sobel edge
detector.
FEATURE EXTRACTION:

 Popular Statistical Method:

Gray Level Co-Occurrence Matrix (GLCM)

 Data Mining:
Process of extracting and discovering patterns.
Data mining phase is used to select more efficient genes in detecting cancer.
The enhanced neural network technique is used to optimize the result.

 Following are the features which are considered while detection of Cancer:
Statistical: Statistical parameters like mean, variance, standard deviation.
Textural: Textural features correlation factor, entropy, contrast and energy.
Geometrical: It includes geometrical features like area of cell, perimeter,
radius, eccentricity, symmetry and concavity.
MODULE 4- IMAGE CLASSIFICATION:

Extracted Features

CNN,BPNN Classifier

CANCER Detection
CLASSIFICATION:
IMAGE CLASSIFICATION:
 Key Technique of CAD
Image Classification
 Image Classification:
Classify &
Labelling group of pixels
 Supervised Image Classification
Procedure for identifying spectrally similar areas on an image by
identifying 'training' sites of known targets and then extrapolating those
spectral signatures to other areas of unknown targets.
CNN:
 A Convolutional neural network (CNN)
Neural network
One or more convolutional layers
Used for image processing, classification, segmentation and also for other auto
correlated data.

 CNNs are used for image classification and recognition because of its high accuracy.

 The CNN
Follows a hierarchical model
Works on building a network, like a funnel
Finally gives out a fully-connected layer and the output is processed .

 CNN or the convolutional neural network (CNN) is a class of deep learning neural networks
Take in an input image
Assign importance (learnable weights and biases) to various aspects/objects in
the image
Be able to differentiate one from the other.
BPNN:
 Back propagation in neural network is a short form for "backward
propagation of errors.“

 Train Convolutional neural networks.

 Calculate the gradient of a loss function with respect to all the weights in
the network.

 The steps involved in Back propagation:

Step — 1: Forward Propagation
Step — 2: Backward Propagation
Step — 3: Putting all the values together and calculating the updated
weight value
RESULT:
Train and Test Score
Percentage based Confusion Matrix
PREDICTED PREDICTED
NO YES

Actual
NO

ACTUAL
YES
CONCLUSION:

 Research contains
Detection of various kinds of cancer and
Classification of the types of cancer from CT/MRI scanned image samples using
image processing.

 If misclassification is being investigated, the Confusion Matrix may be a critical

measure.
Shows…. which classes are correctly classified during a predicted class.
which groups are assisting because they know what they are being
misclassified as. In this case, it's advantageous to have an accurate model.

 The system should have high reliability, accuracy and efficacy, less processing time,
smaller error, less cost and must be robust.

 Early identification of Cancer yields in providing the appropriate treatment to the patient.
FUTURE WORK:

 Computer-Aided Detection (CAD) is one of the constantly growing

technologies that help in detecting cancer by feeding in certain inputs
containing patient-related information such as scans like CT-Scan, X-Ray,
MRI Scan, unusual symptoms in patients or biomarkers, etc.

 SVM, CNN, ANN, Watershed Segmentation, Image enhancement,

Image processing are a few methods used to improve the accuracy and aid
the process.

 For training, the most popular datasets used are Super Bowl Dataset
2020, and LIDC-IDRI2020.
CONFERENCE:
Presented a paper in the International Conference entitled “Early Detection of
Cancer using Convolutional Neural Network and Back Propagation Neural
Network”, 4th International E-Conference on Incipient Research in information
technology. Commerce, management and linguistics-2021 in “The Quaide Milleth
College for Men”, Chennai.

Presented a paper in the International Conference entitled “Computer-Aided

Detection and Diagnosis of Breast Cancer with Convolutional Neural Network
Analysing Deep Features” in the ICETMCCT-2021 held on 9th and 10th December
2021 Organized by Annai Vailankanni College of Engineering, AVK nagar,
Pothaiyadi salai, Azhagappapapuram.P.O, K.K.District, Tamilnadu, India.
JOURNAL:
 Published a paper entitled “Computer-Aided Detection and Diagnosis of Breast
Cancer with Convolutional Neural Network Analysing Deep Features” in the
International Journal of Scientific research in science, engineering and technology,
volume 9, issue 8/7676
REFERENCES:
[1] Andreas S. Panayides, Ashish Sharma, Sotirios A. Tsaftaris, Nhan Do , Spyretta
Golemati,Konstantina S. Nikita, Michalis Zervakis, Constantinos S. Pattichis AI in
Medical Imaging Informatics: Current Challenges and Future Directions ,IEEE
JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 24, NO. 7,
JULY 2020

[2] Guotai Wang , Wenqi Li , Maria A. Zuluaga , Rosalind Pratt, Premal A. Patel,
Michael Aertsen, Tom Doel, Anna L. David, Jan Deprest, Sébastien Ourselin, and
Tom Vercauteren Interactive Medical Image Segmentation Using Deep Learning With
Image-Specific Fine Tuning IEEE TRANSACTIONS ON MEDICAL IMAGING,
VOL. 37, NO. 7, JULY 2019

[3] Yuchuan Qiao, Baldur van Lew, Boudewijn P. F. Lelieveldt, and Marius Staring
Fast Automatic Step Size Estimation for Gradient Descent Optimization of Image
Registration IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 35, NO. 2,
FEBRUARY 2018
 REFERENCES:
[4] A.F. AgarapOn all types cancer detection: an application of machine learning
algorithms on the Wisconsin diagnostic dataset, CoRR abs/1711.07831

[5] Benjamin Hou , Bishesh Khanal , Amir Alansary , Steven McDonagh, Alice
Davidson, Mary Rutherford , Jo V. Hajnal, Daniel Rueckert , Ben Glocker , and
Bernhard Kainz 3-D Reconstruction in Canonical Co-Ordinate Space From
Arbitrarily Oriented 2-D Images. IEEE TRANSACTIONS ON MEDICAL
IMAGING, VOL. 37, NO. 8, AUGUST 2019

[6] Z. Han, B. Wei, Y. Zheng, Y. Yin, K. Li, S. Li All types cancer multi-classification
from histopathological images with structured deep learning model
 REFERENCES:
[7] Shu, Xin; Zhang, Lei; Wang, Zizhou; Lv, Qing; Yi, Zhang (2020). “Deep Neural
Networks with Region-based Pooling Structures for Mammographic Image
Classification” in IEEE Transactions on Medical Imaging, Volume: 39, Issue: 6, 1–
1. doi:10.1109/TMI.2020.2968397

[8] J. C. Souza et al., “An automatic method for brain segmentation and
reconstruction in chest X-Ray using deep neural networks,” Comput. Meth. Prog.
Bio., vol.177, pp. 285-296. Aug. 2019.

[9] M. Khened, V. A. Kollerathu, G. Krishnamurthi, “Fully convolutional multi-scale

residual DenseNets for cardiac segmentation and automated cardiac diagnosis
using ensemble of classifiers,” Med. Image. Anal., vol. 1, no. 51, pp. 21-45, Jan.
2019
 REFERENCES:
[10] Thosini K. Bamunu Mudiyanselage, Xueli Xiao, Yanqing Zhang, Yi Pan, "Deep
Fuzzy Neural Networks for Biomarker Selection for Accurate Cancer Detection" in
IEEE Transactions on Fuzzy Systems ,volume 28,issue 12,2020

[11] Shaban, Muhammad; Awan, Ruqayya; Fraz, Muhammad Moazam; Azam,

Ayesha; Tsang, Yee-Wah; Snead, David; Rajpoot, Nasir M. (2020). “Context-
Aware Convolutional Neural Network for Grading of Colorectal Cancer Histology
Images in IEEE Transactions on Medical Imaging”, volume 39,issue 7(), 1–
1. doi:10.1109/TMI.2020.2971006

[12] Yang, Heechan; Kim, Ji-Ye; Kim, Hyongsuk; Adhikari, Shyam P.

(2019). “Guided Soft Attention Network for Classification of Breast Cancer
Histopathology Images. IEEE Transactions on Medical Imaging”, Volume:
39, Issue: 5, 1–1. doi:10.1109/TMI.2019.2948026
 REFERENCES:
[13] Byra, Michal; Dobruch-Sobczak, Katarzyna; Klimonda, Ziemowit; Piotrzkowska-
Wroblewska, Hanna; Litniewski, Jerzy (2020). “Early prediction of response to
neoadjuvant chemotherapy in breast cancer sonography using Siamese convolutional
neural networks” in IEEE Journal of Biomedical and Health Informatics, Volume:
25, Issue: 3, 1–1. doi:10.1109/JBHI.2020.3008040

[14] Wu, Nan; Phang, Jason; Park, Jungkyu; Shen, Yiqiu; Huang, Zhe; Zorin, Masha;
Jastrzebski, Stanislaw; Fevry, Thibault; Katsnelson, Joe; Kim, Eric; Wolfson, Stacey;
Parikh (2019). “Deep Neural Networks Improve Radiologists’ Performance in Breast
Cancer Screening” IEEE Transactions on Medical Imaging, Volume: 39, Issue: 4, 1–
1. doi:10.1109/TMI.2019.2945514

[15] Alawad, Mohammed; Yoon, Hong-Jun; Gao, Shang; Mumphrey, Brent; Wu, Xiao-
Cheng; Durbin, Eric B.; Jeong, Jong Cheol; Hands, Isaac; Rust, David; Coyle, Linda;
Penberthy, Lynne; Tourassi, Georgia (2020). “Privacy-Preserving Deep Learning NLP
Models for Cancer Registries” in IEEE Transactions on Emerging Topics in Computing,
Volume: 9, Issue: 3, 1–1. doi:10.1109/TETC.2020.2983404
 REFERENCES:
[16] Zheng, Sunyi; Guo, Jiapan; Cui, Xiaonan; Veldhuis, Raymond N. J.; Oudkerk, Matthijs;
van Ooijen, Peter M.A. (2019). “Automatic Pulmonary Nodule Detection in CT Scans
Using Convolutional Neural Networks Based on Maximum Intensity Projection”. IEEE
Transactions on Medical Imaging, Volume: 39, Issue: 3,
1–1. doi:10.1109/TMI.2019.2935553

[17] Wang, Yi; Wang, Na; Xu, Min; Yu, Junxiong; Qin, Chenchen; Luo, Xiao; Yang, Xin;
Wang, Tianfu; Li, Anhua; Ni, Dong (2019). “Deeply-Supervised Networks with Threshold
Loss for Cancer Detection in Automated Breast Ultrasound. IEEE Transactions on
Medical Imaging”, Volume: 39, Issue: 4, 1–1. doi:10.1109/TMI.2019.2936500

[18] Wang, Chunyu; Guo, Junling; Zhao, Ning; Liu, Yang; Liu, Xiaoyan; Liu, Guojun; Guo,
Maozu (2019). “A Cancer Survival Prediction Method Based on Graph Convolutional
Network. IEEE Transactions on NanoBioscience”, Volume: 19, Issue: 1, 1–
1. doi:10.1109/TNB.2019.2936398
 REFERENCES:
[19] V. G. Kanas, E. I. Zacharaki, C. Davatzikos, K. N. Sgarbas, and V.
Megalooikonomou, “A low cost approach for medical image tumor segmentation
based on intensity modeling and 3D Random Walker,” Biomedical Signal
Processing and Control, vol. 22, pp. 19–30, 2015.

[20] Vijayalakshmi M M, "Melanoma Skin Cancer Detection using Image Processing

and Machine Learning", International Journal of trend in scientific Research and
Development, vol 3, 2019.
Project schedule
 September 2021 – Conference paper work and First Module complete
 October 2021 – Full code work complete
 November 2021– journal preparation and report preparation
 January 2022– phase 2 work implementation started
 February 2022– phase 2 report work preparation
THANK YOU