1. Computer-aided identification of allergenic species of Urticaceae pollen.
- Author
-
de Sá-Otero, M. Pilar, González, Amelia P., Rodríguez-Damián, M., and Cernadas, E.
- Subjects
URTICACEAE ,URTICALES ,POLLEN ,POLLINATION ,PALYNOLOGY ,BOTANY - Abstract
Plant species of the family Urticaceae have periods of extensive and overlapping flowering. Two genera Urtica ( U. urens L., U. membranacea Poir. in Lam) and Parietaria ( P. judaica L .) are abundant throughout the Iberian Peninsula. Both have substantial allergenic potential, but patient sensitivity is specific to each genus. As a result, interest from the medical profession in the identification of the pollen from these genera has arisen. It would be useful to be able to estimate the relative percentages of pollen of these genera in the air, which would allow clinicians to recommend the appropriate safety measures to patients. Unfortunately, palynologists are unable to distinguish between the species of genera ( Parietaria and Urtica ). Additionally, traditional methods of pollen counting are laborious and therefore not optimal for routine on-line analysis. Recent biometrical analysis enables us to discriminate between pollen species of the genera Parietaria and Urtica , but they fail to solve the problem of genera identification. Alternative methods for counting and distinguishing genera of Urticaceae have long been awaited by medical experts among others. In this paper we present a computational system to discriminate between genera by using image analysis. It is based on the definition and computation of digital shape measures on pollen images taken by an optical microscope. The system correctly classifies more than 86% of all experiments done. A set of shape parameters are tested and the subset with the highest discrimination power includes: area (A), diameter (DM), mean distance to centroide (D mean ) and roundness (R) of the pollen grain. [ABSTRACT FROM AUTHOR]
- Published
- 2004
- Full Text
- View/download PDF