Share Email Print
cover

Proceedings Paper

Skin cancer texture analysis of OCT images based on Haralick, fractal dimension and the complex directional field features
Author(s): Dmitry S. Raupov; Oleg O. Myakinin; Ivan A. Bratchenko; Dmitry V. Kornilin; Valery P. Zakharov; Alexander G. Khramov
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Optical coherence tomography (OCT) is usually employed for the measurement of tumor topology, which reflects structural changes of a tissue. We investigated the possibility of OCT in detecting changes using a computer texture analysis method based on Haralick texture features, fractal dimension and the complex directional field method from different tissues. These features were used to identify special spatial characteristics, which differ healthy tissue from various skin cancers in cross-section OCT images (B-scans). Speckle reduction is an important pre-processing stage for OCT image processing. In this paper, an interval type-II fuzzy anisotropic diffusion algorithm for speckle noise reduction in OCT images was used. The Haralick texture feature set includes contrast, correlation, energy, and homogeneity evaluated in different directions. A box-counting method is applied to compute fractal dimension of investigated tissues. Additionally, we used the complex directional field calculated by the local gradient methodology to increase of the assessment quality of the diagnosis method. The complex directional field (as well as the “classical” directional field) can help describe an image as set of directions. Considering to a fact that malignant tissue grows anisotropically, some principal grooves may be observed on dermoscopic images, which mean possible existence of principal directions on OCT images. Our results suggest that described texture features may provide useful information to differentiate pathological from healthy patients. The problem of recognition melanoma from nevi is decided in this work due to the big quantity of experimental data (143 OCT-images include tumors as Basal Cell Carcinoma (BCC), Malignant Melanoma (MM) and Nevi). We have sensitivity about 90% and specificity about 85%. Further research is warranted to determine how this approach may be used to select the regions of interest automatically.

Paper Details

Date Published: 27 April 2016
PDF: 10 pages
Proc. SPIE 9887, Biophotonics: Photonic Solutions for Better Health Care V, 98873F (27 April 2016); doi: 10.1117/12.2227896
Show Author Affiliations
Dmitry S. Raupov, Samara State Aerospace Univ. (Russian Federation)
Oleg O. Myakinin, Samara State Aerospace Univ. (Russian Federation)
Ivan A. Bratchenko, Samara State Aerospace Univ. (Russian Federation)
Dmitry V. Kornilin, Samara State Aerospace Univ. (Russian Federation)
Valery P. Zakharov, Samara State Aerospace Univ. (Russian Federation)
Alexander G. Khramov, Samara State Aerospace Univ. (Russian Federation)


Published in SPIE Proceedings Vol. 9887:
Biophotonics: Photonic Solutions for Better Health Care V
Jürgen Popp; Valery V. Tuchin; Dennis L. Matthews; Francesco Saverio Pavone, Editor(s)

© SPIE. Terms of Use
Back to Top