A new super vised approach for breast cancer diagnosis based on ar tificial social bees

A new supervised approach for breast cancerdiagnosis based on artificial social bees

Hanane MENAD

GeCoDe LaboratoryDepartment of Computer Science,

Tahar MOULAY University of Saida, Algeria

Hanane MENAD A new supervised approach for breast cancer diagnosis based on artificial social bees 1 / 32

Contents

1 Introduction

2 Application of computer science in medicine

3 Approach Proposed

4 Results and Discussion

5 Conclusion and Perspective


Introduction

Introduction


Introduction

Introduction

Cancer is a chronic disease that can be fatal; it affects nowadaysincreasingly the human worldwide.The use of computers with automated tools, large volumes ofmedical data are being collected and made available to themedical research groups.


Introduction

Introduction

Cancer is a chronic disease that can be fatal; it affects nowadaysincreasingly the human worldwide.The use of computers with automated tools, large volumes ofmedical data are being collected and made available to themedical research groups.


Application of computer science in medicine

Application of computerscience in medicine




The electronic devices supplied with processing units became animportant component of our everyday life.Health care as a vital part of contemporary society model is alsoaffected by the same technical trends as the other branches ofbusiness.all computer methods that have proven to have technical and scientificpotential are quickly developed and utilized in medicine.




Example results produced by the DMD system.




Visualization of CTA volumetric dataset.




The hardware set - up of prototype of virtual three-dimensionaldesktop and picture taken during performance tests



Application of Computer Science in Medicine



Application of Computer Science in Medicine



Problematic

breast cancer presents a difficult issues for researchers, the mainissue is how can we diagnostic of cancer cellular ?



Proposed Solution

Our solution is based on suppervised classification using twoalgorithms:

Social Bees Algorithmthe Nearest neighbhour (1-NN)


Approach Proposed

Approach Proposed


Approach Proposed

Approach Proposed


Approach Proposed

Cell Nucleus


Approach Proposed

Representation of Data

For each cell nucleus, 10 real-valued features are computed:

1 Radius of individual nucleus.

2 The perimeter.3 Nuclear area.

4 The compactness of the cell nuclei (perimeter2

area− 1.0).

5 The smoothness of a nuclear contour (local variation in radiuslengths).

6 Concave points (number of concave portions of the contour).7 Symmetry8 The Concavity(severity of concave potions of the contour).9 The fractal dimension of a cell (”coastline approximation” - 1).

10 The texture (standard deviation of gray-scale values).


Approach Proposed


For each cell nucleus, 10 real-valued features are computed:1 Radius of individual nucleus.2 The perimeter.3 Nuclear area.


area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed




area− 1.0).





Approach Proposed

Dataset

the Wisconsin Diagnostic Breast Cancer represented by 30element vector containing 30 real values.This dataset contains 569 samples:

357 examples present Benign.212 examples present Malignant.


Approach Proposed

Social Worker Bees

Main tasks of natural worker bees:

Worker Bee Housekeeping (days 1 to 3).

Worker Bee Undertakers (days 3 to 16).Collecting Nectar for the Hive (days 12 to 18).


Approach Proposed

Social Worker Bees

Main tasks of natural worker bees:Worker Bee Housekeeping (days 1 to 3).Worker Bee Undertakers (days 3 to 16).Collecting Nectar for the Hive (days 12 to 18).


Approach Proposed

Social Worker Bees



Approach Proposed

Social Worker Bees



Approach Proposed

Artificial Worker Bees

Used distances:Manhattan (Man) :

D(x,y) = Σ|xi-yi|

Euclidean (Euc):

D(x,y) = Σ√

(xi − yi)2

Chebyshev (Cheb):

D(x, y)= Max(|xi-yi|)

Cosine (Cos) :

D(X,Y ) =Σxi × yi√

Σx2i ×√

Σy2i


Approach Proposed



D(x,y) = Σ|xi-yi|

Euclidean (Euc):

D(x,y) = Σ√

(xi − yi)2

Chebyshev (Cheb):


Cosine (Cos) :


Σx2i ×√

Σy2i


Approach Proposed



D(x,y) = Σ|xi-yi|

Euclidean (Euc):

D(x,y) = Σ√

(xi − yi)2

Chebyshev (Cheb):


Cosine (Cos) :


Σx2i ×√

Σy2i


Approach Proposed



D(x,y) = Σ|xi-yi|

Euclidean (Euc):

D(x,y) = Σ√

(xi − yi)2

Chebyshev (Cheb):


Cosine (Cos) :


Σx2i ×√

Σy2i


Approach Proposed

Normalisation of distances

In order to give the distances the same impact, we used to normaliseEuclidean, Manhattan, and Chebyshev distances as follow:

Di(x, y) =Di(x, y)−DMin(x, y)

DMax(x, y)−DMin(x, y)


Approach Proposed

Classification

We calculated the average distance of three calculated distances(tasks):

Daverage(x, y) =D1(X,Y ) + D2(X,Y ) + D3(X,Y )

3

then we classified according to the minimum distance average.


Approach Proposed

Evaluation

To evaluate our approach, we calculated the following metrics:

Accuracy (AC).Error (Er).Precision (Pr).Recall (Rc).F-measure.Entropy.Duration.Confusion Matrix.


Approach Proposed

Evaluation




Approach Proposed

Evaluation




Approach Proposed

Evaluation




Approach Proposed

Evaluation




Approach Proposed

Evaluation




Approach Proposed

Evaluation




Approach Proposed

Evaluation




Approach Proposed

Evaluation




Results and Discussion




Experiments conducted using Social Worker Bees

• Experiment1 (Ex1): Euclidean + Manhattan + Chebyshev.• Experiment2 (Ex2): Euclidean + Manhattan + Cosine.• Experiment3 (Ex3): Euclidean + Chebyshev + Cosine.• Experiment4 (Ex4): Manhattan + Cosine + Chebyshev.















Obtained Results by Social Worker Bees



Obtained Results by Nearest Neighbour (1-NN)


Conclusion and Perspective




Conclusion

Computers have become an essential part of every hospital. Thesemachines help in carrying out these tasks and medical proceduresmuch more efficiently and effectively.The increasing complexity of real-world problems motivates theresearchers to search for efficient methods. Divide and conquertechniques are the one way to solve large and complex problemswhich has been a practice in research since long time.



Conclusion

Computers have become an essential part of every hospital. Thesemachines help in carrying out these tasks and medical proceduresmuch more efficiently and effectively.The increasing complexity of real-world problems motivates theresearchers to search for efficient methods. Divide and conquertechniques are the one way to solve large and complex problemswhich has been a practice in research since long time.



Conclusion

We have introduced an approach inspired from Social Worker Beesthat proves that it is a good source to inspire from it, with itsmechanism and based on multi agent task.Our approach shows many advantages

Easy to understandEasy to implement because it is based on distance calculation.The main advantage of this approach is its efficiency



Conclusion





Conclusion





Conclusion





Future Works

Generate the approach to diagnostic of other cancer types.Combine this approach with data mining techniques that prove itseffiency such as K-NN we used in this study.Use other bio-inspired methods for this task.Use it for different medical applications.



Future Works




Future Works




Future Works



That’s all. Thanks for your attention! Any Questions?

Hanane MENADGeCoDe Laboratory, Department of Computer Science,

Tahar MOULAY University of Saida, AlgeriaEmail: [email protected]


A new super vised approach for breast cancer diagnosis based on ar tificial social bees

Data & Analytics

breast cancer diagnosis

medicine hanane menad

new supervised approach

computer methods

algeria hanane menad

perspective hanane menad

performance tests hanane

medicine example results