Thesis 1432
  • José Manuel Pintor Freire
Read Center
  • Universidade de Vigo - Escola Superior de Enxeñaría Informática
Doctoral Program
  • Sistemas Software Inteligentes y Adaptables (SSIA)

Diseño y desarrollo de una herramienta para el análisis cuantitativo automático de imágenes microscópicas: aplicación a la gestión de recursos marinos

To estimate productivity of a fish stock, the precise determination of fish fecundity is essential. The stereological method accurately estimates fecundity from histological images of a fish gonad. For that purpose, a hexagonal grid is overlaid on the histological image and the number of grid points associated to each oocyte (reproductive cells) category and the number of oocytes in each category is counted. This process is done manually often using off-the-shelf software, but it is very time-consuming, requires specialized technicians, and does not allow to review the calculations. In this paper, we describe and evaluate the software Govocitos, which offers an easy and automatic way to estimate fecundity using the stereological method. Govocitos contains a module to automatically detects the matured oocytes in the slice (nearly 80\% of oocytes are correctly detected) and a module to automatically classify the oocytes according to the presence/absence of nucleus (with 84% of accuracy) and to three development stages (with 87% of accuracy). It also provides a user friendly GUI that allows the experts to modify the outlines and classifications of oocytes, to calculate diameters, areas and roundness, to build diameter frequency histograms, to count the points and objects inside the grid, to estimate partial and potential fecundity and to export the data to files and into a database. In addition, Govocitos provides the possibility of varying grid characteristics, it can be trained to work with different species and it allows to check and supervise the calculations whenever needed including in a later point in time. Govocitos is a free software that can be downloaded from or
Keywords: Histological image, fish fecundity, computer vision, edge detection, object recognition, texture analysis, classification
Canonical link