Scholarlink

<p>This study integrates real-time ultrasound (RTU) imaging for live animal assessment with RGB (Red, Green, and Blue) imaging for post-slaughter ribeye analysis, aiming to bridge the gap between pre-slaughter prediction and carcass evaluation in high-lean beef production. The primary objective was to develop a predictive model for ribeye traits using RTU-derived measurements from crossbred Brahman cattle. A total of 74 cattle were studied, of which 15 ribeye steaks were available for RGB imaging and model development. Independent variables including backfat thicknesses (BF), ribeye area (REA), ribeye perimeter (REP), and ribeye depth (RED) were extracted from RTU images (denoted as u) using a portable ultrasound device (3.5-5.0&#xa0;MHz). Post-slaughter RGB images of ribeye steaks (s) were captured using an iPhone 13 Pro, and analysed via Image-J software. Regression models were developed using SPSS 25.0 and validated using XLStat. Significant differences (<i>p</i> &lt; 0.05) were observed between REA_u and REA_s yet a strong positive correlation (<i>r</i> = 0.960) confirmed agreement between two imaging technique. The resulting predictive equation for RGB-derived ribeye area was: REA_s = 38.815 + (0.932 × REA_u) with R<sup>2</sup> = 0.921, adjusted R<sup>2</sup> = 0.915 and the lowest standard error (SE = 1.350) among all the regression model used in this study. Model performance metrics further supported its reliability (RMSE = 1.350, MAPE = 1.049, MSE = 1.822). These results confirmed that REA_u is a reliable non-destructive predictor of ribeye trait, a valuable proof-of-concept tool in high-lean beef production systems. However, the model may be prone to overfitting, and further validation using a larger independent dataset is recommended in future studies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Real-time ultrasound and RGB imaging integration for non-invasive prediction of ribeye area in Tropical Crossbred Brahman Cattle

  • 期刊论文

  • 发表日期：

    2025年12月24日

查看全文
  • Nurul Nuraliya Shahrai,
  • Salma Mohamad Yusop,
  • Maimunah Mohd Ali,
  • Mohd Hanis Mohd Saim,
  • Ahmad Rodzian Pok Abdul Aziz,
  • Muhamad Hidayat Mohd Hashim,
  • Shuji Ueda

摘要

This study integrates real-time ultrasound (RTU) imaging for live animal assessment with RGB (Red, Green, and Blue) imaging for post-slaughter ribeye analysis, aiming to bridge the gap between pre-slaughter prediction and carcass evaluation in high-lean beef production. The primary objective was to develop a predictive model for ribeye traits using RTU-derived measurements from crossbred Brahman cattle. A total of 74 cattle were studied, of which 15 ribeye steaks were available for RGB imaging and model development. Independent variables including backfat thicknesses (BF), ribeye area (REA), ribeye perimeter (REP), and ribeye depth (RED) were extracted from RTU images (denoted as u) using a portable ultrasound device (3.5-5.0 MHz). Post-slaughter RGB images of ribeye steaks (s) were captured using an iPhone 13 Pro, and analysed via Image-J software. Regression models were developed using SPSS 25.0 and validated using XLStat. Significant differences (p < 0.05) were observed between REA_u and REA_s yet a strong positive correlation (r = 0.960) confirmed agreement between two imaging technique. The resulting predictive equation for RGB-derived ribeye area was: REA_s = 38.815 + (0.932 × REA_u) with R2 = 0.921, adjusted R2 = 0.915 and the lowest standard error (SE = 1.350) among all the regression model used in this study. Model performance metrics further supported its reliability (RMSE = 1.350, MAPE = 1.049, MSE = 1.822). These results confirmed that REA_u is a reliable non-destructive predictor of ribeye trait, a valuable proof-of-concept tool in high-lean beef production systems. However, the model may be prone to overfitting, and further validation using a larger independent dataset is recommended in future studies.