Share Email Print
cover

Proceedings Paper

Development of an acquisition protocol and a segmentation algortihm for wounds of cutaneous Leishmaniasis in digital images
Author(s): Kristians Diaz; Benjamín Castañeda; César Miranda; Roberto Lavarello; Alejandro Llanos
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

We developed a protocol for the acquisition of digital images and an algorithm for a color-based automatic segmentation of cutaneous lesions of Leishmaniasis. The protocol for image acquisition provides control over the working environment to manipulate brightness, lighting and undesirable shadows on the injury using indirect lighting. Also, this protocol was used to accurately calculate the area of the lesion expressed in mm2 even in curved surfaces by combining the information from two consecutive images. Different color spaces were analyzed and compared using ROC curves in order to determine the color layer with the highest contrast between the background and the wound. The proposed algorithm is composed of three stages: (1) Location of the wound determined by threshold and mathematical morphology techniques to the H layer of the HSV color space, (2) Determination of the boundaries of the wound by analyzing the color characteristics in the YIQ space based on masks (for the wound and the background) estimated from the first stage, and (3) Refinement of the calculations obtained on the previous stages by using the discrete dynamic contours algorithm. The segmented regions obtained with the algorithm were compared with manual segmentations made by a medical specialist. Broadly speaking, our results support that color provides useful information during segmentation and measurement of wounds of cutaneous Leishmaniasis. Results from ten images showed 99% specificity, 89% sensitivity, and 98% accuracy.

Paper Details

Date Published: 13 March 2010
PDF: 11 pages
Proc. SPIE 7623, Medical Imaging 2010: Image Processing, 76234L (13 March 2010); doi: 10.1117/12.844534
Show Author Affiliations
Kristians Diaz, Pontificia Univ. Católica del Perú (Peru)
Benjamín Castañeda, Pontificia Univ. Católica del Perú (Peru)
César Miranda, Univ. Peruana Cayetano Heredia (Peru)
Roberto Lavarello, Univ. of Illinois at Urbana-Champaign (United States)
Alejandro Llanos, Univ. Peruana Cayetano Heredia (Peru)


Published in SPIE Proceedings Vol. 7623:
Medical Imaging 2010: Image Processing
Benoit M. Dawant; David R. Haynor, Editor(s)

© SPIE. Terms of Use
Back to Top