ºÝºÝߣ

ºÝºÝߣShare a Scribd company logo

Cervical cancer classification using gabor filters 1026

•Download as PPTX, PDF•
•
E
es712

This document proposes using Gabor filters and K-means clustering to classify cervical biopsy images as normal, CIN1, CIN2, CIN3 or malignant. Images are preprocessed using Gabor filters to extract texture features, then segmented and classified using K-means clustering based on ratios of normal and abnormal cells. Evaluation shows this approach achieved sensitivities between 82-89% and specificity of 85% for cervical cancer grading.

1 of 17
Downloaded 18 times
CERVICAL CANCER CLASSIFICATION USING GABOR FILTERS 2011 First IEEE International Conference on Healthcare Informatics, Imaging and Systems Biology Advisor : Yin-Fu Huang Student : Chen-Ju Lai
OUTLINE ï‚¢ INTRODUCTION ï‚¢ DATA COLLECTION ï‚¢ METHODOLOGY AND RESULT ï‚¢ CONCLUSION
INTRODUCTION ï‚¢ Cervical cancer ï‚¢ Biopsy test ï‚¢ Cervical intraepithelial neoplasia (CIN) ï‚¢ Input : histology images Feature extraction : texture , using Gabor filter Classification method : K-Means Clustering Output : Normal/CIN1/CIN2/CIN3/Malignant Pre-cancer
DATA COLLECTION ï‚¢ Pathology anatomy laboratory of Saiful Anwar hospital ï‚¢ Biopsy images : resolution 4080 x 3072 pixels (categorized by an expert pathologist) ï‚¢ 475 labelled images are used in this study Normal CIN1 CIN2 CIN3 Malignant 60 70 50 50 245
DATA COLLECTION
CANCER GRADING METHODOLOGY AND RESULT
GABOR FILTER Spatial domain Gabor elementary function 2D Gaussian function x'=x cos θ +y sin θ and y'=-x sin θ+y cos θ. From (1) and (2), the Gabor elementary function can be rewritten as centre frequency σx and σy are the spread of the Gaussian in x and y directions
GABOR FILTER Frequency domain  Assuming σx and σy are the same u'=u cos θ +v sin θ and v'=u sin θ+v cos θ v (U,V) can decision (U0 ,θ) V φ U0 θ u U
GABOR FILTER  Sample Original (a) f = 0.2,θ = 0 0 (b) f = 0.2,θ = 45 0 (c) f = 0.2,θ = 90 0 (d) f = 0.2,θ = 135 0
COMPARE TEMPLATE ï‚¢ Compare each pixel with the templates. ï‚¢ Supervised Training : generated templates ï‚— 24 distinctive Gabor filters are used to generate a feature vector for each pixel and its neighbors. background basal stroma normal abnormal cells cells 500 pixels 500 pixels 500 pixels 500 pixels 500 pixels average average average average average
SEGMENTED IMAGE & K-MEAN CLUSTERING ï‚¢ Segmentation ï‚— After each pixel compare with the five feature vector templates. ï‚¢ blue : background , yellow : basal , white : stroma, ï‚¢ green : normal cell , red : abnormal cell ï‚¢ K-Means Clustering ï‚— Based on the color. ï‚— Quantify the normal nuclei and abnormal nuclei.
SEGMENTED IMAGE & K-MEAN CLUSTERING
CALCULATE THE RATIO AND GRADING ï‚¢ How to classify the image into categories ? ï‚— Use the ratio of number of normal and abnormal cells. Benign the number of abnormal cells < 5 CIN 1 ratio between abnormal and normal cells < 1/3 CIN 2 ratio between abnormal and normal cells between 1/3 ~ 2/3 CIN 3 ratio between abnormal and normal cells > 2/3 or full Malignant ratio between abnormal and normal cells > CIN 3
CALCULATE THE RATIO AND GRADING ï‚¢ Table 1 shows the sample of the ratio between abnormal and normal cell.
CALCULATE THE RATIO AND GRADING ï‚¢ Table 2 shows the confusion matrix of the Gabor filter hybrid with K-means clustering. ï‚¢ The sensitivity of normal is 87%, CIN 1 is 86%, CIN 2 82 %,CIN 3 84% and malignant is 89%. ï‚¢ The percentage of specificity of this system is 85%. (52/60)=0.87 (60/70)=0.86 (41/50)=0.82 (42/50)=0.84 (219/245)=0.89
COMPARED WITH SERVAL METHOD ï‚¢ Gray level Features , color K-mean and incremental thresholding.
CONCLUSION ï‚¢ A methodology of Gabor filter bank with hybrid K- means clustering algorithm has been proposed. ï‚¢ Designing Gabor filter bank with the optimum selection parameters and different classification method can improve performance using this algorithm.

More Related Content

Cervical cancer classification using gabor filters 1026

  • 1. CERVICAL CANCER CLASSIFICATION USING GABOR FILTERS 2011 First IEEE International Conference on Healthcare Informatics, Imaging and Systems Biology Advisor : Yin-Fu Huang Student : Chen-Ju Lai
  • 2. OUTLINE ï‚¢ INTRODUCTION ï‚¢ DATA COLLECTION ï‚¢ METHODOLOGY AND RESULT ï‚¢ CONCLUSION
  • 3. INTRODUCTION ï‚¢ Cervical cancer ï‚¢ Biopsy test ï‚¢ Cervical intraepithelial neoplasia (CIN) ï‚¢ Input : histology images Feature extraction : texture , using Gabor filter Classification method : K-Means Clustering Output : Normal/CIN1/CIN2/CIN3/Malignant Pre-cancer
  • 4. DATA COLLECTION ï‚¢ Pathology anatomy laboratory of Saiful Anwar hospital ï‚¢ Biopsy images : resolution 4080 x 3072 pixels (categorized by an expert pathologist) ï‚¢ 475 labelled images are used in this study Normal CIN1 CIN2 CIN3 Malignant 60 70 50 50 245
  • 7. GABOR FILTER Spatial domain Gabor elementary function 2D Gaussian function x'=x cos θ +y sin θ and y'=-x sin θ+y cos θ. From (1) and (2), the Gabor elementary function can be rewritten as centre frequency σx and σy are the spread of the Gaussian in x and y directions
  • 8. GABOR FILTER Frequency domain ï‚¢ Assuming σx and σy are the same u'=u cos θ +v sin θ and v'=u sin θ+v cos θ v (U,V) can decision (U0 ,θ) V φ U0 θ u U
  • 9. GABOR FILTER ï‚¢ Sample Original (a) f = 0.2,θ = 0 0 (b) f = 0.2,θ = 45 0 (c) f = 0.2,θ = 90 0 (d) f = 0.2,θ = 135 0
  • 10. COMPARE TEMPLATE ï‚¢ Compare each pixel with the templates. ï‚¢ Supervised Training : generated templates ï‚— 24 distinctive Gabor filters are used to generate a feature vector for each pixel and its neighbors. background basal stroma normal abnormal cells cells 500 pixels 500 pixels 500 pixels 500 pixels 500 pixels average average average average average
  • 11. SEGMENTED IMAGE & K-MEAN CLUSTERING ï‚¢ Segmentation ï‚— After each pixel compare with the five feature vector templates. ï‚¢ blue : background , yellow : basal , white : stroma, ï‚¢ green : normal cell , red : abnormal cell ï‚¢ K-Means Clustering ï‚— Based on the color. ï‚— Quantify the normal nuclei and abnormal nuclei.
  • 12. SEGMENTED IMAGE & K-MEAN CLUSTERING
  • 13. CALCULATE THE RATIO AND GRADING ï‚¢ How to classify the image into categories ? ï‚— Use the ratio of number of normal and abnormal cells. Benign the number of abnormal cells < 5 CIN 1 ratio between abnormal and normal cells < 1/3 CIN 2 ratio between abnormal and normal cells between 1/3 ~ 2/3 CIN 3 ratio between abnormal and normal cells > 2/3 or full Malignant ratio between abnormal and normal cells > CIN 3
  • 14. CALCULATE THE RATIO AND GRADING ï‚¢ Table 1 shows the sample of the ratio between abnormal and normal cell.
  • 15. CALCULATE THE RATIO AND GRADING ï‚¢ Table 2 shows the confusion matrix of the Gabor filter hybrid with K-means clustering. ï‚¢ The sensitivity of normal is 87%, CIN 1 is 86%, CIN 2 82 %,CIN 3 84% and malignant is 89%. ï‚¢ The percentage of specificity of this system is 85%. (52/60)=0.87 (60/70)=0.86 (41/50)=0.82 (42/50)=0.84 (219/245)=0.89
  • 16. COMPARED WITH SERVAL METHOD ï‚¢ Gray level Features , color K-mean and incremental thresholding.
  • 17. CONCLUSION ï‚¢ A methodology of Gabor filter bank with hybrid K- means clustering algorithm has been proposed. ï‚¢ Designing Gabor filter bank with the optimum selection parameters and different classification method can improve performance using this algorithm.