The automated, computer-aided method for differentiation and classification of malignant (M) from benign (B) cases, by analyzing the UV/VIS spectra of pleural effusions is described. It was shown that by two independent objective features, the maximum of Katz fractal dimension (KFDmax) and the area under normalized UV/VIS absorbance curve (Area), highly reliable M-B classification is possible. In the Area-KFDmax space M and B samples are linearly separable permitting thus the use of linear support vector machine as a classification tool. By analyzing 104 samples of UV/VIS spectra of pleural effusions (88 M and 16 B) collected from patients at the Clinic for Lung Diseases and Tuberculosis, Military Medical Academy in Belgrade, the accuracy of 95.45% for M cases and 100% for B cases are obtained by using the proposed method. It was shown that by applying some modifications, which are suggested in the paper, the accuracy of 100% for M cases can be reached.
TY - JOUR
AU - Jevtic, Dubravka R.
AU - Avramov Ivić, Milka
AU - Reljin, Irini S.
AU - Reljin, Branimir D.
AU - Plavec, Goran I.
AU - Petrović, Slobodan D.
AU - Mijin, Dušan
PY - 2014
UR - http://cer.ihtm.bg.ac.rs/handle/123456789/1516
AB - The automated, computer-aided method for differentiation and classification of malignant (M) from benign (B) cases, by analyzing the UV/VIS spectra of pleural effusions is described. It was shown that by two independent objective features, the maximum of Katz fractal dimension (KFDmax) and the area under normalized UV/VIS absorbance curve (Area), highly reliable M-B classification is possible. In the Area-KFDmax space M and B samples are linearly separable permitting thus the use of linear support vector machine as a classification tool. By analyzing 104 samples of UV/VIS spectra of pleural effusions (88 M and 16 B) collected from patients at the Clinic for Lung Diseases and Tuberculosis, Military Medical Academy in Belgrade, the accuracy of 95.45% for M cases and 100% for B cases are obtained by using the proposed method. It was shown that by applying some modifications, which are suggested in the paper, the accuracy of 100% for M cases can be reached.
PB - Elsevier Science Bv, Amsterdam
T2 - Physica A: Statistical Mechanics and its Applications
T1 - Diagnostic spectroscopic and computer-aided evaluation of malignancy from UV/VIS spectra of clear pleural effusions
VL - 403
SP - 206
EP - 216
DO - 10.1016/j.physa.2014.02.026
ER -
@article{
author = "Jevtic, Dubravka R. and Avramov Ivić, Milka and Reljin, Irini S. and Reljin, Branimir D. and Plavec, Goran I. and Petrović, Slobodan D. and Mijin, Dušan",
year = "2014",
url = "http://cer.ihtm.bg.ac.rs/handle/123456789/1516",
abstract = "The automated, computer-aided method for differentiation and classification of malignant (M) from benign (B) cases, by analyzing the UV/VIS spectra of pleural effusions is described. It was shown that by two independent objective features, the maximum of Katz fractal dimension (KFDmax) and the area under normalized UV/VIS absorbance curve (Area), highly reliable M-B classification is possible. In the Area-KFDmax space M and B samples are linearly separable permitting thus the use of linear support vector machine as a classification tool. By analyzing 104 samples of UV/VIS spectra of pleural effusions (88 M and 16 B) collected from patients at the Clinic for Lung Diseases and Tuberculosis, Military Medical Academy in Belgrade, the accuracy of 95.45% for M cases and 100% for B cases are obtained by using the proposed method. It was shown that by applying some modifications, which are suggested in the paper, the accuracy of 100% for M cases can be reached.",
publisher = "Elsevier Science Bv, Amsterdam",
journal = "Physica A: Statistical Mechanics and its Applications",
title = "Diagnostic spectroscopic and computer-aided evaluation of malignancy from UV/VIS spectra of clear pleural effusions",
volume = "403",
pages = "206-216",
doi = "10.1016/j.physa.2014.02.026"
}
Jevtic, D. R., Avramov Ivić, M., Reljin, I. S., Reljin, B. D., Plavec, G. I., Petrović, S. D.,& Mijin, D. (2014). Diagnostic spectroscopic and computer-aided evaluation of malignancy from UV/VIS spectra of clear pleural effusions.
Physica A: Statistical Mechanics and its Applications
Elsevier Science Bv, Amsterdam., 403, 206-216.
https://doi.org/10.1016/j.physa.2014.02.026
Jevtic DR, Avramov Ivić M, Reljin IS, Reljin BD, Plavec GI, Petrović SD, Mijin D. Diagnostic spectroscopic and computer-aided evaluation of malignancy from UV/VIS spectra of clear pleural effusions. Physica A: Statistical Mechanics and its Applications. 2014;403:206-216
Jevtic Dubravka R., Avramov Ivić Milka, Reljin Irini S., Reljin Branimir D., Plavec Goran I., Petrović Slobodan D., Mijin Dušan, "Diagnostic spectroscopic and computer-aided evaluation of malignancy from UV/VIS spectra of clear pleural effusions" Physica A: Statistical Mechanics and its Applications, 403 (2014):206-216,
https://doi.org/10.1016/j.physa.2014.02.026 .
