Pregnancy diagnosis in pigs: a field study comparing linear-array real-time ultrasound scanning and amplitude depth analysis

Assessment of diagnostic accuracy and effectiveness of trans-abdominal real-time ultrasound imaging for pregnancy diagnosis in breeding sows under intensive management

Early and accurate determination of pregnancy is critical to optimum reproductive performance in pigs and enables farmers to early rebreed or cull non-pregnant animals. Most of the conventional diagnostic methods are unsuitable for systematic application under practical conditions. The advent of real-time ultrasonography has made it possible to establish relatively more reliable pregnancy diagnosis. The present study was carried out to evaluate the diagnostic accuracy and effectiveness of trans-abdominal real-time ultrasound (RTU) imaging vis-à-vis pregnancy status in sows reared under intensive management. Trans-abdominal ultrasonographic examinations were performed using a mechanical sector array transducer and portable ultrasound system in crossbred sows from 20 days post-insemination for up to next 40 days. Animals were followed up for subsequent reproductive performance with farrowing data used as the definitive test for deriving predictive values. Accuracy for diagnosis was determined by diagnostic accuracy measures like sensitivity, specificity, predictive values, and likelihood ratios. Before 30 days of breeding, RTU imaging had 84.21% sensitivity and 75% specificity. Relatively higher false diagnosis rates were obtained in animals checked at or before 55 days after AI than in animals checked after 55 days (21.73% versus 9.09%). Negative pregnancy rate was low with 29.16% (7/24) false positives. Overall sensitivity and specificity, using farrowing history as the gold standard, were 94.74% and 70.83% respectively. The sensitivity of testing tended to be slightly lower in sows with litter size of less than 8 total born piglets, compared to sows with 8 or more piglets. Overall positive likelihood ratio was 3.25 while negative likelihood ratio was 0.07. The results indicate that pregnancy in swine herds can be reliably detected earlier in gestation by 30 days post-insemination using trans-abdominal RTU imaging. This non-invasive technique with portable imaging system can be used as an integral part of reproductive monitoring and sound management practices for profitable swine production systems.

Evaluation of a Numerical, Real-Time Ultrasound Imaging Model for the Prediction of Litter Size in Pregnant Sows, with Machine Learning

Simple Summary

The use of a numeric model to quantify real-time ultrasonographic (RTU) imaging is a prominent methodfor predicting the expected litter size by training an artificial neural network (ANN) to minimize the error of the prediction measured by metrics, such as root square mean error and mean absolute error. Time of the RTU application is a critical factor for such a prediction. Rated scale values (RSV) obtained from the RTU images relate to the accurate diagnosis of pregnancy and of litter size, suggesting the potential of a generalized use of the model in various farm conditions. Ultimately, the employment of the model in machine learning for an automated prediction of litter size can be used as a routine on-farm procedure for the efficient management of gestating sows.

Abstract

The present study aimed to evaluate the accuracy of a numerical model, quantifying real-time ultrasonographic (RTU) images of pregnant sows, to predict litter size. The time of the test with the least error was also considered. A number of 4165 pregnancies in Farm 1 and 438 in Farm 2 were diagnosed twice, with the quality of the RTU images translated into rated-scale values (RSV1 and RSV2). When a deep neural network (DNN) was trained, the evaluation of the method showed that the prediction of litter size can be performed with little error. Root square mean error (RMSE) for training, validation with data from Farm 1, and testing on the data from Farm 2 were 0.91, 0.97, and 1.05, respectively. Corresponding mean absolute errors (MAE) were 2.27, 2.41, and 2.58. Time appeared to be a critical factor for the accuracy of the model. The smallest MAE was achieved when the RTU was performed at days 20–22. It is concluded that a numerical, RTU imaging model is a prominent predictor of litter size, when a DNN is used. Therefore, early routinely evaluated RTU images of pregnant sows can predict litter size, with machine learning, in an automated manner and provide a useful tool for the efficient management of pregnant sows.

Relaxin: a hormonal aid to diagnose pregnancy status in wild mammalian species

In the beginning of 1960s, seminal studies characterizing circulating concentrations of immunoreactive relaxin in companion dogs and evaluating the differences in concentrations among pregnant, nonpregnant, and pseudopregnant bitches indicated the potential for relaxin to be applied clinically as a diagnostic aid to detect pregnancy status in wild animal species. A brief historical overview of the nature of relaxin and early work to develop and validate immunologic methods to analyze relaxin in the blood of rodents and pigs is initially discussed, which is followed by a summary of the development and validation of relaxin immunoassays to diagnose pregnancy in companion dogs and cats. Thereafter, observation of the pregnancy-specific increase in circulating concentrations of relaxin in laboratory, companion, and farm animal species leads to discussion on the application of radioimmunoassays, enzyme immunoassays, and a rapid immunomigration assay to diagnose pregnancy in wild terrestrial (e.g., wolves, lions, elephants, rhinoceros, panda) and marine (e.g., seals, dolphins) mammal species. A reference table is included with a comprehensive list of numerous species and essential reagents that have been used in various in-house and commercial immunoassays to successfully analyze relaxin quantitatively and qualitatively in blood (serum or plasma) and to some extent in urine. Although the detection of relaxin concentrations has the potential to aid in the diagnosis of pregnancy in many wild animal species, there are challenges in other species. Future efforts should focus on validation of nonradiolabeled relaxin immunoassays for broader application among species and improving techniques (e.g., extraction, purification) to analyze relaxin in samples other than blood (e.g., urine, feces, saliva, blow, skin, blubber) that can be collected in a less-invasive or -stressful manner and processed accordingly for basic and applied purposes, especially with application toward conservation of threatened or endangered species.

An update on the use of B-mode ultrasonography in female pig reproduction

After technological advances allowed for the adaptation of B-mode ultrasonography equipment for use in pig facilities, ultrasonography quickly established itself as an ideal diagnostic aid for determining pregnancy status in pigs. In recent years, B-mode ultrasonography has found increased application in its use for monitoring ovarian activity and in estimating time of ovulation in pigs. B-mode ultrasonography is also valuable in providing a detailed assessment of the sow's ovaries and uterus to determine if pathological conditions exist, which could be contributing to poor individual or herd reproductive performance. In its most recent application in pigs, the gilt genital tract has been characterized peripubertally by ultrasonography in order to detect onset of puberty. The purpose of this review is to provide an update on the current status of B-mode ultrasonography in pig reproduction, and how this technology can be of value when used in pig production medicine.

Integrating ultrasonography within the reproductive management of the collared peccary (Tayassu tajacu)

Ultrasound imaging has been used to elucidate certain aspects of the reproductive biology of wild or endangered species. However, to our knowledge, this tool has not been used for reproductive monitoring of the collared peccary (Tayassu tajacu). In this study, real-time ultrasonography was used in 16 collared peccary females to diagnose early pregnancy status and predict gestational age. Based on the detection of an embryo, the earliest pregnancy diagnosis was made on Day 18 after mating, with the mean time needed for diagnosis being 22 days. Overall accuracies on Days 22, 26 and 28 were 56, 93, and 100%, respectively. On Days 26 and 28, all pregnancy and non-pregnancy diagnoses, respectively, were correct. The fetal measurements that best correlated with gestational age were crown-rump-length (CRL) and the length and diameter of the thorax. CRL was considered the most practical measurement because, contrary to thoracic fetometry, it could be determined when the embryo was first detected. Our findings revealed real-time ultrasound scanning to be a very accurate method for early pregnancy diagnosis and prediction of gestational age in the collared peccary.

Ultrasonographic characterization of the ovaries in non-pregnant first served sows and gilts

This study was aimed firstly, to examine the ovaries in non-pregnant first served sows and gilts by transcutaneous ultrasonography and secondly, to evaluate the suitability for this procedure to be performed routinely on farms. Two thousand five hundred and twenty-three females on a 1250-sow unit, synchronized with Regumate (gilts only) and/or gonadotropins (sows and gilts) not detected as returned to estrus by daily boar contact prior to scanning were ultrasonographically tested for pregnancy between days 20 and 114 postinsemination (p.i.). Of 256 sows (S) and 130 gilts (G) found to be non-pregnant the ovaries were visualized in 87.1 and 80.0% of them, respectively. Ovarian findings were: corpora lutea (CL); follicles of 2-6mm (F(2-6)); peri-ovulatory ovarian structures (POS; comprising follicles of 7-8mm and corpora haemorrhagica); single cysts (SC); oligocystic ovarian degeneration (OOD) and polycystic ovarian degeneration (POD). Their incidence was: CL>F2-6>POS>POD ( P<0.05 ) in both S and G. POD and SC plus OOD were more frequently in S ( P<0.05 ). The ovarian findings were related to the intervals of regular (days 18-25 p.i. (R1), 38-46 p.i. (R2)) and irregular service returns (days 26-37 p.i. (IR1), 47-114 p.i. (IR2)). Comparison within intervals: CL tended to be more frequently with P<0.05 only at IR2 in S. Comparison among intervals (R1 to IR2): The percentage of females (1) with CL tended to increase (S and G) and (2) with F2-6 plus POS decreased significantly (S; P<0.05 ) or tendentiously (G). SC plus OOD was higher before R2, POD after IR1 (S and G; P<0.05 ). In conclusion, the results indicate a high heterogeneity of ovarian structures in non-pregnant first served sows and gilts up to day 114 after service and suggest CL as an important cause for a delayed and, rather than POD, a failed service return. The results further demonstrate that transcutaneous ultrasonography is an appropriate and recommended method for examining the ovaries on farm in female pigs with reproductive failures.

Early diagnosis of pregnancy in sows by ultrasound evaluation of embryo development and uterine echotexture

Two operators attempted to detect pregnancy ultrasonographically in 196 sows daily from 15 to 25 days after insemination; 20 unbred sows were also investigated. The probe was applied transcutaneously on the right abdominal wall near the last three mammary glands. During each examination, the embryos were visualised and their transverse and longitudinal dimensions were measured. Pregnancy was confirmed by an ultrasonographic detection of embryos five days after the first ultrasound diagnosis and finally 30 to 32 days after insemination. The accuracy of diagnosis was less than 83 per cent on days 15, 16 and 17 but improved to more than 90 per cent from day 18 onwards. The uterine echotexture was studied in seven sows at oestrus and 15, 17, 19, 21, 23 and 25 days after insemination. The echotexture was more homogeneous from days 15 to 25 after insemination than at oestrus.

Pregnancy diagnosis in sows: direct ELISA for estrone in feces and its prospects for an on-farm test, in comparison to ultrasonography

The usefulness of fecal estrone (E1) measurement as a tool for pregnancy diagnosis was investigated. Concentrations of E1 were measured in feces from pregnant and nonpregnant sows by a direct ELISA without extraction. Highly significant differences in E1 concentrations were found in feces from nonpregnant and pregnant sows (P = 0.016 to < 0.001). Pregnancy diagnosis on Days 26 to 32 after insemination, based both on fecal E1 concentrations as measured by ELISA and ultrasonography using a 5.0 MHz linear-array transducer, was performed in a group of 496 gilts and sows. The fecal E1 test had a sensitivity (correct diagnosis of pregnancy) of 96.5% and a specificity (correct diagnosis of nonpregnancy) of 93.6%, using 3.65 ng E1/g feces as a cut-off value. For ultrasonographic pregnancy diagnosis the test sensitivity and specificity were 99.3 and 92.5%, respectively. Although an increase of fecal E1 concentrations was noticed for increasing litter sizes, the results indicated that these concentrations could not be used to predict litter size. It is concluded that the distribution of fecal E1 concentrations in both nonpregnant and pregnant sows offers a suitable basis for the development of a simple, sow-side pregnancy test.

Evaluation of false ultrasonographic pregnancy diagnoses in sows by measuring the concentration of unconjugated estrogens in feces

On Days 26, 28, and 30 after AI, ultrasonographic pregnancy diagnoses were performed on 207 gilts and sows by using a 3.5 MHz linear-array transducer. Fecal samples were taken from the rectum after each ultrasonographic examination, and the concentrations of unconjugated estrogens in selected samples (n = 73) were measured by RIA. Fecal unconjugated estrogen concentration of 11.7 ng/g feces or higher was indicative of pregnancy. The overall sensitivity and specificity of the ultrasonographic test was 99% for farrowing sows and 73.1% for nonfarrowing sows. With one exception, sows with a false negative diagnosis by ultrasonography on Day 26 were correctly diagnosed pregnant by elevated fecal unconjugated estrogens or repeated ultrasonographic examinations on Days 28 or 30. Return to estrus around the sampling period may cause false positive results in the unconjugated estrogen assay, while early embryonic mortality can result in false positive diagnoses in both the ultrasonographic test and estrogen assay. Although there was a positive correlation between the concentrations of unconjugated estrogens in the feces and litter size at farrowing in the selected sows, it seems very unlikely that fecal estrogens can provide an accurate tool for predicting litter size.

Pregnancy diagnosis in swine by palpation and by amplitude-depth ultrasound scanning

The relative accuracy of two pregnancy testing methods for swine were compared in a field study. The procedures used were manual palpation and amplitude-depth ultrasonic scanning. A total of 369 sows were examined by both methods. Seven additional gilts were examined by ultrasound only and 46 sows by palpation per rectum only. The number of correct positive and negative diagnoses made by both methods were calculated, and determination of accuracy as well as comparison between the tests were made on this basis. The relative accuracy was 97.6% for the manual method and 96.8% for the ultrasound method. Both tests had a high sensitivity, 99.2 and 98.9%, respectively. The ability of the tests to detect the non-pregnant animals was not as high, which is reflected by a lower specificity. No significant differences were noted between the two methods. A lower specificity and a lower negative predictive value were provided by ultrasound scanning as compared with those acquired by manual palpation. Both procedures were considered to be quick and convenient to perform. It was concluded that in spite of the new pregnancy testing methods introduced in the swine industry, manual palpation remains the most practical in terms of its accuracy, ease, and the minimal requirement for equipment. In gilts, palpation is unsuitable and ultrasonography currently remains the best choice for the diagnosis of pregnancy.

Pregnancy diagnosis in cows with linear-array real-time ultrasound scanning: a preliminary note

Pregnancy diagnoses were performed under farm conditions in 201 cows (Hungarian Red Pied and crossbreds with Holstein) between 21 and 70 days after last insemination date, using a linear-array real-time ultrasound scanner with a 3.0 MHz rectal transducer. Identification of (a) sharply demarcated black area(s) and/or an embryo within the uterus were used as criterions for pregnancy. Pregnancy was confirmed by rectal examination at 6 and/or 8 weeks after A.I. From the total number of correct positive diagnoses (n = 129), incorrect positive diagnoses (n = 3, made on days 36, 40 and 44 after A I), correct negative diagnoses (n = 62) and incorrect negative diagnoses (n = 7, made on days 25, 28, 28, 29, 30, 31, and 33 after A I), a sensitivity of 94.8%, a specificity of 95,3%, a positive predictive value of 97.7% and a negative predictive value of 89.8% were calculated. From these preliminary results it was concluded that real-time ultrasound scanning is a useful and reliable technique for early pregnancy diagnosis in cows. It is suggested that a combination of milk/plasma progesterone estimation between days 21 and 24 and real-time ultrasound scanning between days 25 and 45 would allow the confirmation and quantification of (late) embryonic mortality in the cow.