Determinação do rendimento em peças e tecidos de carcaças de caprinos por análise de imagens obtidas por tomografia computorizada
Data
2018-11-07
Autores
Título da revista
ISSN da revista
Título do Volume
Editora
Resumo
Neste trabalho foram utilizadas as técnicas não destrutivas tomografia computorizada (TC) e análise de imagem vídeo (VIA) para avaliar a composição de carcaças leves de cabritos da raça Bravia. Foram utilizadas dez carcaças de cabritos da raça Bravia com 5,2 ± 0,81 kg de peso da carcaça fria. As carcaças foram embaladas a vácuo e transportadas para a sala experimental de abate da UTAD para avaliação da composição TC e VIA. Foi determinada a composição da carcaça por dissecação em músculo (M), gordura subcutânea (GS), gordura intermuscular (GI) e osso (O). Para a TC as carcaças foram examinadas num equipamento de GE Brivo 325. Para a determinação do volume dos tecidos foi utilizado o método de segmentação (thresholding) tendo em consideração as unidades Hounsfield (HU) características de cada tecido. Assim, o tecido muscular foi segmentado utilizando uma gama de -22 a 146 HU; o tecido adiposo foi segmentado através de uma gama de -194 e - 23 HU e para o osso foi considerado um valor de HU entre 146 e 1000. Para o método VIA foram obtidas imagens das carcaças nas vistas dorsal e lateral com uma câmara digital (Nikon D3100) sempre nas mesmas condições de luz e de posição. Foram realizadas 25 medidas com recurso ao programa ImageJ 1.42q. Para a análise da composição da carcaça e das peças foi realizada uma estatística descritiva com a determinação da média, desvio padrão, máximo, mínimo e coeficiente de variação. Relativamente à análise da TC também foram realizadas análises descritivas e uma variância simples para peso dos tecidos entre a TC e a dissecação. Foi feita uma análise de correlação e de regressão entre as variáveis de composição determinada por dissecação e as respetivas variáveis obtidas por TC. Para a análise do VIA para além das análises descritivas foram realizados modelos stepwise para a estimativa da composição da carcaça a partir de medidas VIA e determinados os coeficientes de determinação (R2 ) e o desvio padrão dos resíduos. Todas as análises estatísticas foram realizadas com o programa JMP (versão 13; SAS Institute, Cary, NC, EUA). Para a TC os valores de correlação entre os tecidos determinados por dissecação e por TC variaram entre 0,68 e 0,98 (P<0,01). Foi possível explicar 95% da variação de músculo da carcaça com a TC. Relativamente à VIA também foi possível explicar de forma significativa a variação dos tecidos da carcaça dos cabritos (R2 entre 0,85 e 0,99; P<0,01) Os resultados encontrados mostram que a TC e a VIA apresentam capacidade para estimar de forma não destrutiva a composição das carcaças ligeiras de cabritos da raça Bravia.
In this work, non-destructive computerized tomography (CT) and video image analysis (VIA) was used to evaluate the composition of light carcasses of Bravia goat breed. Ten carcasses with 5.2 ± 0.81 kg of weight were used. The carcases were vacuum packed and transported to the UTAD for carcass composition determination, evaluation of TC composition and VIA. The carcass composition was determined by dissection in muscle (M), subcutaneous fat (GS), intermuscular fat (GI) and bone (O). For CT, the carcasses were examined in a GE Brivo 325 equipment. A thresholding method was used to determine the volume of tissues, taking into account the Hounsfield units (HU) characteristic of each tissue. Thus for muscle tissue was segmented using a range of -22 to 146 HU; the fat tissue was segmented through a range of -194 and -23 HU and for bone was considered an HU value between 146 and 1000. Regarding VIA, the carcasses images were obtained in the dorsal and lateral views with a digital camera (Nikon D3100) always in the same light and position conditions. Twenty-five measurements were performed using the ImageJ 1.42q program. For carcass composition, a descriptive statistic was presented with the determination of the mean, standard deviation, maximum, minimum and coefficient of variation. Regarding the CT analysis, descriptive analyses and a one-way ANOVA were performed for tissue weight between CT and dissection. A correlation and regression analysis were performed between the composition variables determined by dissection and the respective variables obtained by CT. For the analysis of the VIA in addition to the descriptive analyses, stepwise models were performed to estimate the carcass composition from VIA measurements, and the determination coefficient (R2 ) and the residual standard deviation were determined. All statistical analyses were performed with the JMP program (version 13; SAS Institute, Cary, NC, USA). For CT, the correlation values between tissues determined by dissection and CT varied between 0.68 and 0.98 (P <0.01) It was possible to explain 95% of the variation of carcass muscle with CT. Concerning to VIA, it was also possible to explain the variation of goat carcass tissues (R2 between 0.85 and 0.99, P <0.01) significantly. The results show that the TC and the VIA can non-destructively predict the composition of the Bravia goats lightweight carcasses.
In this work, non-destructive computerized tomography (CT) and video image analysis (VIA) was used to evaluate the composition of light carcasses of Bravia goat breed. Ten carcasses with 5.2 ± 0.81 kg of weight were used. The carcases were vacuum packed and transported to the UTAD for carcass composition determination, evaluation of TC composition and VIA. The carcass composition was determined by dissection in muscle (M), subcutaneous fat (GS), intermuscular fat (GI) and bone (O). For CT, the carcasses were examined in a GE Brivo 325 equipment. A thresholding method was used to determine the volume of tissues, taking into account the Hounsfield units (HU) characteristic of each tissue. Thus for muscle tissue was segmented using a range of -22 to 146 HU; the fat tissue was segmented through a range of -194 and -23 HU and for bone was considered an HU value between 146 and 1000. Regarding VIA, the carcasses images were obtained in the dorsal and lateral views with a digital camera (Nikon D3100) always in the same light and position conditions. Twenty-five measurements were performed using the ImageJ 1.42q program. For carcass composition, a descriptive statistic was presented with the determination of the mean, standard deviation, maximum, minimum and coefficient of variation. Regarding the CT analysis, descriptive analyses and a one-way ANOVA were performed for tissue weight between CT and dissection. A correlation and regression analysis were performed between the composition variables determined by dissection and the respective variables obtained by CT. For the analysis of the VIA in addition to the descriptive analyses, stepwise models were performed to estimate the carcass composition from VIA measurements, and the determination coefficient (R2 ) and the residual standard deviation were determined. All statistical analyses were performed with the JMP program (version 13; SAS Institute, Cary, NC, USA). For CT, the correlation values between tissues determined by dissection and CT varied between 0.68 and 0.98 (P <0.01) It was possible to explain 95% of the variation of carcass muscle with CT. Concerning to VIA, it was also possible to explain the variation of goat carcass tissues (R2 between 0.85 and 0.99, P <0.01) significantly. The results show that the TC and the VIA can non-destructively predict the composition of the Bravia goats lightweight carcasses.
Descrição
Dissertação de Mestrado em Engenharia Zootécnica apresentada à Universidade de Trás-os-Montes e Alto Douro
Palavras-chave
análise de imagem vídeo , cabritos , carcaça , tomografia computorizada