THE CONSTRUCTION OF THE FEATURE VECTOR IN THE DIAGNOSIS OF SARCOIDOSIS BASED ON THE FRACTAL ANALYSIS OF CT CHEST IMAGES
CT images corresponding to the cross-sections of the patients’ upper torso were analysed. The data set included the healthy class and 3 classes of cases affected by sarcoidosis. It was a state involving only the trachea – Sick(1), a state including trachea and lung parenchyma – Sick(2) and a state involving only lung parenchyma – Sick(3). Based on a fractal analysis and a feature selection by linear stepwise regression, 4 descriptors were obtained, which were later used in the classification process. These were 2 fractal dimensions calculated by the variation and box counting methods, lacunarity calculated also with the box counting method and the intercept parameter calculated using the power spectral density method. Two descriptors were obtained as a result of a gray image analysis, and 2 more were the effect of a binary image analysis. The effectiveness of the descriptors was verified using 8 popular classification methods. In the process of classifier testing, the overall classification accuracy was 90.97%, and the healthy cases were detected with the accuracy of 100%. In turn, the accuracy of recognition of the sick cases was: Sick(1) – 92.50%, Sick(2) – 87.50% and Sick(3) – 90.00%. In the classification process, the best results were obtained with the support vector machine and the naive Bayes classifier. The results of the research have shown the high efficiency of a fractal analysis as a tool for the feature vector extraction in the computer aided diagnosis of sarcoidosis.
fractals; sarcoidosis; computed tomography; image texture analysis
Arkema E. V., Cozier Y. C.: Epidemiology of sarcoidosis: current findings and future directions. Ther Adv Chronic Dis 9(11)/2018, 227–240.
Baughman R. P., Culver D. A., Judson, M. A.: A Concise Review of Pulmonary Sarcoidosis. Am J Respir Crit Care Med 183/2011, 573–581.
Bradley D., Roth G.: Adaptive Thresholding Using the Integral Image. http://www.scs.carleton.ca/~roth/iit-publications-iti/docs/gerh-50002.pdf (available: 19.05.2019).
Breiman L.: Bagging Predictors. Mach Learn 24/1996, 123–140.
Breiman L.: Random Forests. Mach Learn 45/2001, 5–32.
Breiman L., Friedman J., Olshen R., et al.: Classification and Regression Trees. CRC Press, London 1984.
Bonifazi M., Gasparini S., Alfieri V., Renzoni E. A.: Pulmonary Sarcoidosis. Semin Respir Crit Care Med 38/2017, 437–449.
Clarke K. C.: Computation of the fractal dimension of topographic surfaces using the triangular prism surface area method. Comp Geosci 12/1986, 713–722.
Criado E., Sánchez M., Ramírez J., Arguis P., de Caralt T. M., Perea R. J., et al.: Pulmonary Sarcoidosis: Typical and Atypical Manifestations at High-Resolution CT with Pathologic Correlation. RadioGraphics 30/2010, 1567–1586 [DOI: 10.1148/rg.306105512].
Dennis T. J., Dessipris N. G.: Fractal modelling in image texture analysis. IEEE Proc.-F. 136/1989, 227–235.
Enas G. G., Chai S. C.: Choice of the smoothing parameter and efficiency of the k-nearest neighbor classification. Comput Math Appl 12/1986, 235–244.
Freund Y., Schapire R. E.: A decision-theoretic generalization of on-line learning and an application to boosting. J Comput Syst Sci Int 55/1996, 119–139.
Gneiting T., Sevcikova H., Percival D.: Estimators of Fractal Dimension: Assessing the Roughness of Time Series and Spatial Data. Statistical Science 27(2)/2012, 247–277.
Hothorn T., Lausen B.: Bundling classifiers by bagging trees. Comput Stat Data An 49(4)/2005, 1068–1078.
Iftekharuddin K. M., Jia W., Marsh R.: Fractal analysis of tumor in brain MR images. Mach Vision Appl 13(5-6)/2003, 352–362.
Liao S. H., Chu P. H., Hsiao P. Y.: Data mining techniques and applications – A decade review from 2000 to 2011. Expert Syst Appl 39/2012, 11303–11311.
Mandelbrot B.: The fractal geometry of nature. W. H. Freeman and Company, New York, (1983).
Omiotek Z.: Improvement of the classification quality in detection of Hashimoto’s disease with a combined classifier approach. P I Mech Eng H 231(8)/2017, 774–782.
Omiotek Z.: Fractal analysis of the grey and binary images in diagnosis of Hashimoto’s thyroiditis. Biocybern Biomed Eng 37(4)/2017, 655–665.
Omiotek Z., Wójcik W.: Zastosowanie metody Hellwiga do redukcji wymiaru przestrzeni cech obrazów USG tarczycy. Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska – IAPGOS 3/2014, 14–17 [DOI: 10.5604/20830157.1121333].
Omiotek Z., Wójcik W.: An efficient method for analyzing measurement results on the example of thyroid ultrasound images. Przeglad Elektrotechniczny 11/2016, 15–18.
Omiotek Z., Stepanchenko O., Wójcik W., Legieć W., Szatkowska M.: The use of the Hellwig’s method for feature selection in the detection of myeloma bone destruction based on radiographic images. Biocybern Biomed Eng 39(2)/2019, 328–338.
Otsu N.: A Threshold Selection Method from Gray-Level Histograms. IEEE Transactions on Systems, Man, and Cybernetics 9(1)/1979, 62–66.
Plotnick R. E., Gardner R. H., O'Neil R. V.: Lacunarity indices as measures of landscape texture. Landsc Ecol 8(3)/1993, 201–211.
Quinlan J. R.: Induction of decision trees. Mach Learn 1/1986, 81–106.
Sawicki D., Omiotek Z.: Evaluation of the possibility of using fractal analysis to study the flame in the co-firing process. Proc. SPIE 10808/2018.
Soto-Gomez N., Peters J. I., Nambiar A. M.: Diagnosis and Management of Sarcoidosis. American Family Physician 93(10)/2016, 840–848.
Super B. J., Bovik A. C.: Localized measurement of image fractal dimension using Gabor filters. J Visual Commun Image Represent 2(2)/1991, 114–128.
Venables W. N., Ripley B. D.: Modern Applied Statistics with S-PLUS. Springer, Berlin 1998.
Zhu Z., Stein M.: Parameter estimation for fractional Brownian surfaces. Statistica Sinica 12/2002, 863–883.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.