Publications
2013
1.
Alegre, Enrique; Biehl, Michael; Petkov, Nicolai; Sánchez-González, Lidia
Assessment of acrosome state in boar spermatozoa heads using n-contours descriptor and RLVQ Artículo de revista
En: Computer methods and programs in biomedicine, vol. 111, no 3, pp. 525–536, 2013, (Publisher: Elsevier).
Resumen | Enlaces | BibTeX | Etiquetas: digital image processing, machine learning, Sperm Analysis, veterinary science
@article{alegre_assessment_2013,
title = {Assessment of acrosome state in boar spermatozoa heads using n-contours descriptor and RLVQ},
author = {Enrique Alegre and Michael Biehl and Nicolai Petkov and Lidia Sánchez-González},
url = {https://www.sciencedirect.com/science/article/pii/S0169260713001478},
year = {2013},
date = {2013-01-01},
journal = {Computer methods and programs in biomedicine},
volume = {111},
number = {3},
pages = {525–536},
abstract = {This paper presents a digital image processing method to assess the acrosome state of boar spermatozoa heads. Using grayscale images labeled with fluorescent data, the sperm heads are segmented, and multiple inner contours are generated using a logarithmic distance function. Local texture features are computed for these contours, and classification performance is evaluated using Relevance Learning Vector Quantization, class conditional means, and KNN with cross-validation. The best results are achieved with gradient magnitude data, yielding a test error of only 1%, outperforming previous methods and demonstrating the potential for automated veterinary applications.},
note = {Publisher: Elsevier},
keywords = {digital image processing, machine learning, Sperm Analysis, veterinary science},
pubstate = {published},
tppubtype = {article}
}
This paper presents a digital image processing method to assess the acrosome state of boar spermatozoa heads. Using grayscale images labeled with fluorescent data, the sperm heads are segmented, and multiple inner contours are generated using a logarithmic distance function. Local texture features are computed for these contours, and classification performance is evaluated using Relevance Learning Vector Quantization, class conditional means, and KNN with cross-validation. The best results are achieved with gradient magnitude data, yielding a test error of only 1%, outperforming previous methods and demonstrating the potential for automated veterinary applications.
2011
2.
Fernández-Robles, Laura; García-Ordás, Maite; García-Ordás, Diego; Olivera, Óscar García-Olalla; Alegre, Enrique
Acrosome evaluation of spermatozoa cells using sift and classical texture descriptors Artículo de revista
En: Actas de las XXXII Jornadas de Automática, Escuela Técnica Superior de Ingeniería, Univesidad de Sevilla: Sevilla, 7 al 9 de septiembre de 2011, pp. 84, 2011, (Publisher: Universidad de Sevilla).
Resumen | Enlaces | BibTeX | Etiquetas: acrosome integrity, image processing, SIFT, Sperm Analysis, veterinary science
@article{fernandez-robles_acrosome_2011,
title = {Acrosome evaluation of spermatozoa cells using sift and classical texture descriptors},
author = {Laura Fernández-Robles and Maite García-Ordás and Diego García-Ordás and Óscar García-Olalla Olivera and Enrique Alegre},
url = {https://portalcientifico.unileon.es/documentos/6660aac4241b8f26a79c807a},
year = {2011},
date = {2011-01-01},
journal = {Actas de las XXXII Jornadas de Automática, Escuela Técnica Superior de Ingeniería, Univesidad de Sevilla: Sevilla, 7 al 9 de septiembre de 2011},
pages = {84},
abstract = {Automatic assessment of sperm quality is an important
challenge in the veterinary field. In this
paper, we explore how to best describe the acrosomes
of boar spermatozoa using image analysis
to automatically classify them as intact or damaged.
Our proposal is to characterize the acrosomes
in terms of their membrane integrity using
texture descriptors and compare them with descriptors
based on local invariant features, particularly,
Scale Invariant Feature Transform (SIFT)
method. On the one hand, we use Zernike moments
and Haralick features extracted from the
original image and from the coefficients of the Discrete
Wavelet Transform. On the other hand, the
heads’ features are distinctively described by SIFT,
a method based on detecting local points of interest.
Classification using kNN shows that the best
results were obtained by SIFT, with an overall hit
rate of 84.64% and, what is more important, a
higher hit rate in the damaged (92.96%) than in
the intact class (76.15%). These results make this
descriptor very attractive for the veterinary community.},
note = {Publisher: Universidad de Sevilla},
keywords = {acrosome integrity, image processing, SIFT, Sperm Analysis, veterinary science},
pubstate = {published},
tppubtype = {article}
}
Automatic assessment of sperm quality is an important
challenge in the veterinary field. In this
paper, we explore how to best describe the acrosomes
of boar spermatozoa using image analysis
to automatically classify them as intact or damaged.
Our proposal is to characterize the acrosomes
in terms of their membrane integrity using
texture descriptors and compare them with descriptors
based on local invariant features, particularly,
Scale Invariant Feature Transform (SIFT)
method. On the one hand, we use Zernike moments
and Haralick features extracted from the
original image and from the coefficients of the Discrete
Wavelet Transform. On the other hand, the
heads’ features are distinctively described by SIFT,
a method based on detecting local points of interest.
Classification using kNN shows that the best
results were obtained by SIFT, with an overall hit
rate of 84.64% and, what is more important, a
higher hit rate in the damaged (92.96%) than in
the intact class (76.15%). These results make this
descriptor very attractive for the veterinary community.
challenge in the veterinary field. In this
paper, we explore how to best describe the acrosomes
of boar spermatozoa using image analysis
to automatically classify them as intact or damaged.
Our proposal is to characterize the acrosomes
in terms of their membrane integrity using
texture descriptors and compare them with descriptors
based on local invariant features, particularly,
Scale Invariant Feature Transform (SIFT)
method. On the one hand, we use Zernike moments
and Haralick features extracted from the
original image and from the coefficients of the Discrete
Wavelet Transform. On the other hand, the
heads’ features are distinctively described by SIFT,
a method based on detecting local points of interest.
Classification using kNN shows that the best
results were obtained by SIFT, with an overall hit
rate of 84.64% and, what is more important, a
higher hit rate in the damaged (92.96%) than in
the intact class (76.15%). These results make this
descriptor very attractive for the veterinary community.