Ohio State University Research/Extension Bulletin

Animal Sciences Research and Reviews

Special Circular 156

Summary of National Swine Improvement Federation Real-Time Ultrasound Certification Program

S.J. Moeller
Department of Animal Sciences

Summary

Eighteen swine ultrasound technicians participated in a national program of training and certification for the measurement of backfat and loin muscle area in live swine at Ohio State University on May 24 to 25, 1995. Criteria to achieve certification status were based on the accuracy (Standard Deviation of Prediction; SDP), repeatability (Standard Deviation of Difference; SDD), and bias (BIAS) associated with ultrasonic measurements of tenth rib backfat and loin muscle area on the live pig when compared to measurements taken on the carcass. Ten of the eighteen technicians received National Swine Improvement Federation (NSIF) certification for measurement of tenth rib backfat and loin muscle area, with three additional technicians granted certification for tenth rib backfat only.

Introduction

In response to expanded interest in and utilization of ultrasound to monitor carcass composition in breeding swine throughout the United States, the National Swine Improvement Federation (NSIF) has initiated an ultrasound certification program designed to establish guidelines for the use of ultrasound in measuring the economically important traits of backfat and loin muscle area. The goal of the NSIF certification program is to provide a basis for the assessment of the accuracy, repeatability, and bias of an ultrasonic technician's measurements on a live pig when compared with measurements taken on the carcass. The ultrasonic certification program is designed to standardize the gathering of backfat and loin muscle area information on pigs that are being performance tested on breeding herds throughout the United States. Through this program, carcass data collected will be more accurate, allowing genetic differences among animals to be more fully expressed. Ultrasonic technicians completing and passing the certification criteria are awarded ultrasonic certification for a period of two years.

Ohio State University was chosen as a host site for the NSIF certification clinic held on May 24 to 25, 1995. Eighteen participants from 10 states attended the 2-day event held in the Animal Sciences Building in Columbus.

Materials and Methods

The NSIF ultrasound certification clinic consisted of an educational program, followed by a hands-on scanning practicum. The educational program consisted of sessions on principles of ultrasound, statistical methodology, and review of literature related to assessment of accuracy, anatomy of the pig, ultrasound settings, probe placement on the pig, carcass evaluation procedures, and hands-on demonstration of ultrasonic machines.

The scanning practicum was designed to evaluate the technicians on their ability to measure tenth rib backfat (BF10) and loin muscle area (LMA) and was the basis for granting ultrasonic certification. Technicians were required to scan 50 market weight pigs representative of the variation in live weight, backfat depth, and loin muscle area present in market and breeding swine in the current U. S. swine population. Based on these targets, pigs were selected from a local marketing station in Ohio targeting a standard deviation of 0.25 inch for BF10 and 0.75 inch² for LMA.

The 50 pigs were identified and funneled through a working maze designed to allow each technician access to every pig for collection of BF10 and LMA measurements independently from other participants. Following collection of the scan information on each pig, the pigs were renumbered and reordered in the maze for collection of a second set of BF10 and LMA measurements used to assess the repeatability. Following the collection of two measurements on each pig, technicians were required to independently interpret and submit a written copy of their results on each pig for comparison with the carcass measurements.

Carcass measurements were collected under commercial slaughter conditions at Ohio Packing Company, Columbus, OH. Two meat scientists were utilized as carcass officials and given the task of independently measuring BF10 and LMA on the 50 chilled carcasses. Independent measurements were required to be within 0.05 inch for BF10 and within 0.20 inch² for LMA when establishing the official carcass measurements for each pig. When independent measurements were outside these criteria, the two officials were required to make a consensus judgment on the carcass measurement. Official carcass measurements were the average of the two independent measurements on each pig.

Following collection and input of all scan and carcass data, each individual technician was evaluated based upon three criteria. The first statistic evaluated was the standard deviation of prediction (SDP), which evaluates the bias adjusted average error of ultrasonic measurements when compared to the carcass. The second statistic was the standard deviation of the difference (SDD), which measures the average error between repeated ultrasonic scans on the same pig. The final statistic calculated was bias (BIAS), which is the mean difference between the ultrasonic measurements of BF10 and LMA across all 50 pigs and the mean of the carcass measurements on all 50 pigs. The criteria for meeting NSIF certification standards for BF10 were the 0.16 inches for SDP, SDD, and BIAS. For LMA, the certification criteria were 0.60 square inches for SDP, SDD, and BIAS. Technicians were required to meet the standards for each statistic to achieve certification. Certification technicians could achieve certification for either BF10, LMA alone, or both.

Results and Conclusions

The results of the ultrasound certification clinic held in Ohio are presented in the Table 1. Individual technicians' results are coded to retain confidentiality, with results presented for each statistic based on this code. Of the eighteen participants in the clinic, 10 technicians met the required standards for both BF10 and LMA. Three additional technicians received certification for the measurement of BF10 only on the live pig.

The results of the clinic showed that, on average, the technicians tended to underestimate BF10 with ultrasound (0.059 inch), with all technicians but one underestimating the fat on the pigs and all technicians within the standard.

Standard Deviation of Prediction and SDD for BF10 were consistent within technician, with four of the five technicians who did not meet the standard being over the established criteria for both accuracy (SDP) and repeatability (SDD). Technicians tended to overestimate LMA (0.34 inch²) with ultrasound in contrast to measurement of BF10. The eight technicians failing to achieve certification status for LMA were over the standard for both SDP and SDD, indicating inconsistent results for accuracy and a general inability to be repeatable on the same animal.

Technicians participating in the ultrasound clinic had a wide range in experience and expertise with ultrasound equipment and measuring live animals. In a questionnaire answered by the technicians, the number of pigs scanned annually ranged from less than 100 to more than 10,000. The statistics were very indicative of the influence experience had on the final results and the number of participants achieving certification. Experienced technicians performed better than those who had little experience prior to the program.

Table 1. Results and summary statistics for participants in the NSIF ultrasound certification program.

BF10 LMA

Code SDP SDD BIAS CERT SDP SDD BIAS CERT

T10 0.132 0.093 -0.090 YES 0.382 0.405 0.155 YES

T11 0.111 0.071 -0.099 YES 0.444 0.481 0.224 YES

T12 0.146 0.121 -0.030 YES 0.552 0.634 0.313 NO

T13 0.129 0.112 -0.085 YES 0.444 0.516 0.559 YES

T14 0.238 0.197 -0.142 NO NO

T15 0.124 0.123 -0.075 YES 0.478 0.495 0.159 YES

T16 0.117 0.092 -0.020 YES 0.569 0.664 0.649 NO

T17 0.205 0.130 -0.144 NO

T18 0.186 0.211 -0.046 NO 0.607 0.727 0.174 NO

T19 0.142 0.098 -0.152 YES 0.652 0.659 0.647 NO

T20 0.117 0.117 -0.094 YES 0.404 0.582 0.559 YES

T21 0.133 0.146 0.010 YES 0.485 0.316 0.112 YES

T22 0.119 0.113 -0.041 YES 0.520 0.545 0.241 YES

T23 0.183 0.211 -0.021 NO 0.625 0.708 0.146 NO

T24 0.142 0.170 -0.073 NO 0.515 0.641 0.764 NO

T25 0.129 0.09 -0.075 YES 0.454 0.562 0.264 YES

T26 0.123 0.086 -0.084 YES 0.511 0.441 0.217 YES

T27 0.110 0.119 -0.083 YES 0.472 0.382 0.254 YES

0.1437 0.1278 -0.0747 0.5071 0.5474 0.3398

SDP = Standard Deviation of Prediction { · _i·_k(Scan_{i j k} - Carcass_i - Bias_j)²/(N_i-1)}^1/2 .
SDD = Standard Deviation of Difference {·_i·_k(Scan2_{i j} - Scan_{i j})²/( N _i -1)}^1/2 .
Bias = ·_i·_k(Scan_{i j k} - Carcass_i)²/ N _j .
where · is the summation symbol.
Scan_{i j k} is the kth scan taken by the jth technician on the ith pig.
Carcass_i is the carcass measurement on the ith pig.
Bias_j is the bias for the jth technician.
N_i is the total number of pigs scanned.
N _i is the number of unique pigs scanned.
N _j the number of unique pigs scanned by the jth technician.

Table of Contents, Special Circular 156 | Go to Ohioline

Table 1. Results and summary statistics for participants in the NSIF ultrasound certification program.
	BF10				LMA
Code	SDP	SDD	BIAS	CERT	SDP	SDD	BIAS	CERT
T10	0.132	0.093	-0.090	YES	0.382	0.405	0.155	YES
T11	0.111	0.071	-0.099	YES	0.444	0.481	0.224	YES
T12	0.146	0.121	-0.030	YES	0.552	0.634	0.313	NO
T13	0.129	0.112	-0.085	YES	0.444	0.516	0.559	YES
T14	0.238	0.197	-0.142	NO				NO
T15	0.124	0.123	-0.075	YES	0.478	0.495	0.159	YES
T16	0.117	0.092	-0.020	YES	0.569	0.664	0.649	NO
T17	0.205	0.130	-0.144	NO
T18	0.186	0.211	-0.046	NO	0.607	0.727	0.174	NO
T19	0.142	0.098	-0.152	YES	0.652	0.659	0.647	NO
T20	0.117	0.117	-0.094	YES	0.404	0.582	0.559	YES
T21	0.133	0.146	0.010	YES	0.485	0.316	0.112	YES
T22	0.119	0.113	-0.041	YES	0.520	0.545	0.241	YES
T23	0.183	0.211	-0.021	NO	0.625	0.708	0.146	NO
T24	0.142	0.170	-0.073	NO	0.515	0.641	0.764	NO
T25	0.129	0.09	-0.075	YES	0.454	0.562	0.264	YES
T26	0.123	0.086	-0.084	YES	0.511	0.441	0.217	YES
T27	0.110	0.119	-0.083	YES	0.472	0.382	0.254	YES
	0.1437	0.1278	-0.0747		0.5071	0.5474	0.3398
SDP = Standard Deviation of Prediction { · _i·_k(Scan_{i j k} - Carcass_i - Bias_j)²/(N_i-1)}^1/2 . SDD = Standard Deviation of Difference {·_i·_k(Scan2_{i j} - Scan_{i j})²/( N _i -1)}^1/2 . Bias = ·_i·_k(Scan_{i j k} - Carcass_i)²/ N _j . where · is the summation symbol. Scan_{i j k} is the kth scan taken by the jth technician on the ith pig. Carcass_i is the carcass measurement on the ith pig. Bias_j is the bias for the jth technician. N_i is the total number of pigs scanned. N _i is the number of unique pigs scanned. N _j the number of unique pigs scanned by the jth technician.