Uso da análise computacional de imagens ultra-sonográficas na avaliação da função luteal em bovinos

• Ano de defesa: 2007
• Autor(a) principal: Siqueira, Luiz Gustavo Bruno
• Orientador(a): Torres, Ciro Alexandre Alves
• Banca de defesa: Carvalho, Giovanni Ribeiro de , Costa, Eduardo Paulino da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Programa de Pós-Graduação: Mestrado em Zootecnia
• Departamento: Genética e Melhoramento de Animais Domésticos; Nutrição e Alimentação Animal; Pastagens e Forragicul
• País: BR
• Palavras-chave em Português: Ultra-sonografia; Corpo lúteo; Bovino
• Palavras-chave em Inglês: Ultrasonography; Corpus luteum; Bovine
• Área do conhecimento CNPq: CNPQ::CIENCIAS AGRARIAS::ZOOTECNIA::PRODUCAO ANIMAL
• Link de acesso: http://locus.ufv.br/handle/123456789/5843

Ultrasound image attributes, quantified by computer-assisted analysis, are a good indication of corpora lutea (CL) physiological and functional status. However, a standard method to capture and processing images has not been established for bovine. The purposes of this research were to establish a methodology for capture, processing and computer-assisted analysis; to evaluate CL morphological and echotexture changes throughout a natural estrous cycle or a luteolysis induced estrous cycle; and to determine whether computer-assisted analysis is efficient or not for embryo recipient selection. In the first study, evaluations were made of methods to capture images, systems of capture and the minimum representative sample area. For transrectal (TR) images, the minimum established area was 0.25 cm² (4.42% of the major area). A correlation was found in mean pixel values (r = 0.9930; P<0.001) and heterogeneity (r = 0.9861; P<0.001) between images videotaped or direct digitized. In a second study, two experiments were developed: natural estrous cycle and induced luteolysis (day 10) estrous cycle. Echotexture parameters of the CL, in the expt. 1, were analyzed after data adjustment to the onset of luteolysis. In both experiments, mean pixel values did not differ throughout the days of the cycle (P>0.10) and there was no correlation between plasma progesterone and mean pixel value (r = -0.39 and r = -0.28; P>0.10). Pixel heterogeneity varied through the days in a cycle with natural luteolysis (P<0.01), but did not in a luteolysis induced cycle (P>0.10). Correlations were observed between pixel heterogeneity and plasma P4 in both of the experiments (r = - 0.69 and r = -0.48; P<0.05). In a third study, embryo recipients (n=259) were examined by transrectal ultrasonography on the embryo transfer day (ET). The same type of analysis was done to assess image attributes, as previously tested in the 1st and 2nd studies. Good quality embryos (grades 1 and 2) collected after conventional superovulatory treatment (MOET; n=94) or in vitro produced (IVP; 88) were transferred to the recipients. Pregnancy rates were 56.4% and 30.2% for MOET and IVP, respectively. Plasma progesterone concentration values were greater in the recipients later diagnosed as pregnant, than in non-pregnant (5.88±0.77 vs 3.98±0.48 ng/mL, respectively; P<0.05). There was no difference in plasma P4 between pregnant and non pregnant IVP recipients (3.97±0.57 vs 3.52±0.23; P>0.10). Corpus luteum and luteal tissue area did not differ between pregnant and non-pregnant animals. In addition, no difference in mean pixel values (71.80±1.29 vs 71.19±1.14; P>0.10; scale of 0 to 255) and heterogeneity (14.80 vs 14.52; P>0.10) was observed, between pregnant and non pregnant MOET or IVP recipients. Echotexture parameters (mean pixel value and heterogeneity) showed correlations with plasma P4 (0.17 and 0.18, respectively; P<0.01). In conclusion, differences between capture, processing and image analysis methods should be considered in luteal echotexture evaluation; if on one hand, pixel heterogeneity in ultrasound images is a strong indicative of functional status and steroidogenic capacity of luteal gland, on the other hand, mean pixel values are a poor indicator of these characteristics. Finally, in this approach computer-assisted analysis of ultrasound images was not efficient on embryo recipient selection, probably due to other variables which strongly influence overall pregnancy rate.