基于空间漫反射光的母兔早期妊娠诊断装置研制

妊娠诊断是肉兔繁殖管理中的重要环节，尤其是早期妊娠诊断，可提前未妊娠母兔的再授精时间，提高种兔利用率，缩短繁殖周期。针对人工摸胎诊断法存在对母兔腹内胚胎产生机械损伤、母兔应激反应大及对工人经验要求高等缺陷，该研究基于妊娠母兔与未妊娠母兔腹内有无孕囊组织所引起的光学特性差异，提出一种基于空间漫反射光的母兔妊娠诊断方法，研制了快速、无侵入式的便携诊断装置。该装置由具有2个红外发光二极管（Light Emitting Diode，LED，LED发光波长分别为850和930 nm）、3个硅基光电二极管和外围电路的传感探头和信号处理主机组成。利用该便携诊断装置采集130只人工授精14 d后的母兔（包括63只妊娠母兔和67只未妊娠母兔）腹部漫反射光强度数据，将采集的数据按照7∶3的比例划分为训练集和测试集，训练集数据分别用于建立偏最小二乘判别分析（Partial Least Squares Discrimination Analysis, PLS-DA）模型和支持向量机（Support Vector Machine, SVM）分类模型，测试集数据用于模型性能测试，并对比两种模型的分类性能。同时，利用PLS-DA对采样数据进行有监督的主成分分析和变量重要性分析，结果表明妊娠母兔与未妊娠母兔的采样数据之间存在差异，可以被较好的分类。对比两种分类模型的分类性能发现SVM对妊娠母兔和未妊娠母兔的分类性能均比PLS-DA好，对测试集数据的灵敏度、特异性和准确率分别为80.95%、83.33%和82.05%。研究结果表明，该研究提出的光学妊娠诊断方法可行，研制的诊断装置可对授精14 d后的母兔进行妊娠诊断，对提高兔产业体系智能化装备水平有积极的促进作用。

Development of the early pregnancy diagnosis device for female rabbits based on spatial diffuse light

Abstract: Pregnancy diagnosis is one of the most important links in the reproductive management of meat rabbits. Especially, the early identification of non-pregnant rabbits can advance the re-insemination time, further improving the service rate of breeding rabbits for a shorter reproductive cycle in commercial productions. However, the manual touch diagnosis can present some stress responses of rabbits at present. A non-invasive approach can be an alternative for the early pregnancy diagnosis in the future large-scale production of rabbits. In this study, a non-invasive portable diagnosis device was developed to rapidly and accurately identify the early pregnancy of rabbits using spatial diffuse light. The optical characteristics also presented differently from the gestational sac tissue in the abdomen of pregnant and non-pregnant rabbits. The portable diagnostic device consisted of a sensing probe and a signal processing host with two infrared LEDs (the light-emitting wavelengths of 850 and 930 nm), three silicon-based photodiodes, and peripheral circuits. 130 rabbits were selected to test after 14 days of artificial insemination, including 63 pregnant rabbits and 67 non-pregnant rabbits. The collected data was then divided into the training set and test set, according to the ratio of 7:3. Specifically, the training set was used to establish the partial least squares discriminant analysis (PLS-DA) and support vector machine (SVM) classification model. The test set was used to evaluate the classification performance of the established model. At the same time, the PLS-DA was conducted for the supervised principal component analysis (PCA) and variable importance analysis (VIA) on the sampling data. The results showed that there were great differences between the sampling datasets of pregnant and non-pregnant rabbits, indicating a better classification. Furthermore, it was found that the SVM presented a better classification performance than that of the PLS-DA for the pregnant and non-pregnant rabbits, where the sensitivity, specificity, and accuracy of the test set data were 80.95%, 83.33%, and 82.05%, respectively. Consequently, the developed diagnosis device can be widely expected to rapidly identify the pregnancy of rabbits within 14 days after insemination, indicating a feasible optical pregnancy diagnosis. The finding can provide a strong reference to the early pregnancy diagnosis of rabbits, further enhancing the intelligent level of industrial equipment in rabbit production.