Breast Thermograms Features Analysis based on Grey Wolf Optimizer *Faculty of Computers and Information, Cairo University and SRGE member *Gehad Ismail Sayed and Aboul Ella Hassanien http://www.egyptscience.net Bio-inspiring and evolutionary computation: Trends, applications and open issues workshop, 7 Nov. 2015 Faculty of Computers and Information, Cairo University
17
Embed
Breast Thermograms Features Analysis based on Grey Wolf Optimizer
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Breast Thermograms Features Analysis based on Grey Wolf Optimizer
*Faculty of Computers and Information, Cairo University and SRGE member
*Gehad Ismail Sayed and Aboul Ella Hassanien
http://www.egyptscience.net
Bio-inspiring and evolutionary computation: Trends, applications and open issues workshop, 7 Nov. 2015 Faculty of Computers and Information, Cairo University
Overview Introduction
What is Thermography? How Thermal Imaging Works? Problem Definition Motivation
Related Work Proposed Approaches Results and Discussion Conclusion and Future Works
SRGE Workshop, Cairo University (7-November-2015)
2
Introduction
What is Thermography? Infrared Thermography is the science of
acquisition and analysis of thermal information from non-contact thermal imaging devices.
Thermography is non invasive functional imaging method, harmless, passive, fast, low cost and sensitive method.
SRGE Workshop, Cairo University Conference Hall (19-September-2015)
3
Introduction
How Thermal Imaging Works? All objects emit infrared energy (heat) as a function of
their temperature. The infrared emitted by an object is known as its heat
temperature, where the hotter an object is , the more radiation its emits
Thermal camera is essentially a heat sensor that is capable of detecting tiny differences in temperature
4
SRGE Workshop, Cairo University (7-November-2015)
Introduction
Problem Definition Breast cancer is the most common cancer among
women in the world. One out of 8 women will get breast cancer.
5
SRGE Workshop, Cairo University (7-November-2015)
Introduction
Motivation Mammogram is one of the most imaging technology for
diagnosing breast cancer. Although mammogram has recorded a high detection and
classification accuracy, it is difficult in imaging dense breast tissues, its performance is poor in younger women and harmful, it couldn’t detect breast tumor that less than 2 mm and it’s very difficult to detect cancer in early stage
IRT could be a good source of images to study and detect the cancer at the early stages.
6
SRGE Workshop, Cairo University (7-November-2015)
Related Work
Several approaches for classification Breast thermograms to normal or abnormal have been proposes which can be categorized to : Asymmetric classification based on comparison between
the extracted features from left and right breast Asymmetric classification based on extracted feature
from whole region of interest
7
SRGE Workshop, Cairo University (7-November-2015)
SRGE Workshop, Cairo University (7-November-2015)
Proposed Approach8
Preprocessing Phase
Breast Region of Interest Segmentation
Results and Discussion9
DatasetA benchmark database used to evaluate the proposed approach. This is a public database that are constructed by collecting the IR images from UFF University's Hospital and publicly published under the approval of the ethical committee where every patient should sign consent. 61 IR breast images with resolution (640 x 480 pixels) from this database were used in this paper (29 healthy and 32 malignant).
SRGE Workshop, Cairo University (7-November-2015)
Results and Discussion10
Grey Wolf Parameters SettingParameter Value(s)
Number of Features 127Number of Search Agents (Wolves) 200
Number of Iterations 5Range (Boundary of Search Space) [1 127]
Dimension 127Fitness Classification
AccuracySRGE Workshop, Cairo University (7-November-2015)
Comparison between proposed approach and other approaches
MI SDRSF
SSF
SSF
FS PCA GA
GWO
All Fea
tures
0.00%20.00%40.00%60.00%80.00%
100.00%
RBFLinearQua-dratic
SRGE Workshop, Cairo University (7-November-2015)
Conclusion and Future Works
Conclusion Several features extracted from breast region of interest have
been analyzed. Moreover, new features selector technique has been proposed and compared with 6 well known features selectors techniques and one of evolutionary techniques.
The obtained results shows the robustness of the It obtains over the all almost 97% accuracy
Future Works We plan to increase the number of breast thermograms images
dataset to evaluate the performance of the proposed approach and try new version of swarm.