TY - GEN
T1 - A complete modelling of Local Binary Pattern for detection of diabetic retinopathy
AU - Sarwinda, Devvi
AU - Bustamam, Alhadi
AU - Wibisono, Ari
N1 - Publisher Copyright:
© 2017 IEEE.
PY - 2017/10/1
Y1 - 2017/10/1
N2 - Diabetic Retinopathy (DR) is one of the eye diseases which is caused by Diabetes Mellitus, its effect can make blindness. DR can be detected from retinal image with various approaches such as area, color, statistic, and texture. In this study, we propose detection of DR by using texture feature characteristic from STARE database. A complete modelling of Local Binary pattern (CLBP) presented as feature extraction method of texture. Utilization of sign, magnitude and mean value are applied to this feature extraction approach. We have used Expectation Maximization-Principal Component Analysis (EM-PCA) as feature selection method and KNN as a classifier. The experimental results (combination of CLBP sign and mean value, and combination of CLBP sign and magnitude) show better accuracy compare to another method. CLBP-SC (CLBP sign and mean value) has similar accuracy with CLBP-SM (CLBP sign and magnitude), where it is 97.16%. For sensitivity and specificity performance, the higher value is 98% and is 97% respectively. In addition, we also do running time comparison of five approaches. CLBP-SM gives good performance with smaller running time. These results suggest that our proposed method in this paper can be used in aid system diagnosis for diabetic retinopathy.
AB - Diabetic Retinopathy (DR) is one of the eye diseases which is caused by Diabetes Mellitus, its effect can make blindness. DR can be detected from retinal image with various approaches such as area, color, statistic, and texture. In this study, we propose detection of DR by using texture feature characteristic from STARE database. A complete modelling of Local Binary pattern (CLBP) presented as feature extraction method of texture. Utilization of sign, magnitude and mean value are applied to this feature extraction approach. We have used Expectation Maximization-Principal Component Analysis (EM-PCA) as feature selection method and KNN as a classifier. The experimental results (combination of CLBP sign and mean value, and combination of CLBP sign and magnitude) show better accuracy compare to another method. CLBP-SC (CLBP sign and mean value) has similar accuracy with CLBP-SM (CLBP sign and magnitude), where it is 97.16%. For sensitivity and specificity performance, the higher value is 98% and is 97% respectively. In addition, we also do running time comparison of five approaches. CLBP-SM gives good performance with smaller running time. These results suggest that our proposed method in this paper can be used in aid system diagnosis for diabetic retinopathy.
KW - Diabetic Retinopathy (DR)
KW - LBP
KW - PCA
KW - texture feature
UR - http://www.scopus.com/inward/record.url?scp=85049753192&partnerID=8YFLogxK
U2 - 10.1109/ICICOS.2017.8276329
DO - 10.1109/ICICOS.2017.8276329
M3 - Conference contribution
AN - SCOPUS:85049753192
T3 - Proceedings - 2017 1st International Conference on Informatics and Computational Sciences, ICICoS 2017
SP - 7
EP - 10
BT - Proceedings - 2017 1st International Conference on Informatics and Computational Sciences, ICICoS 2017
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 1st International Conference on Informatics and Computational Sciences, ICICoS 2017
Y2 - 15 November 2017 through 16 November 2017
ER -