J APPL POULT RES 2006. 15:274-279
© 2006 Poultry Science Association
Assessing Feather Cover of Laying Hens by Infrared Thermography
N. J. Cook*,1,
A. B. Smykot
,
D. E. Holm,
G. Fasenko and
J. S. Church*
* Alberta Agriculture Food and Rural Development, Lacombe Research Centre, Lacombe, Alberta, Canada; and University of Alberta, Edmonton, Alberta, Canada
1 Corresponding author: nigel.cook{at}gov.ab.ca
 |
SUMMARY
|
|---|
Brown Leghorn hens (n = 98) were assessed for feather cover using a feather scoring (FS) scale and by infrared thermography. Mean image temperature provided a real-time assessment of feather cover that was significantly related to FS. The closest approximation to FS was achieved by considering the area of the image within specific temperature ranges. The percentage of the image area within a temperature range of 17 to 24°C was positively associated with feather cover, whereas the area within a temperature range of 28 to 31°C was negatively associated with FS and indicative of a lack of cover (bare skin). Infrared thermography provided a fast, objective, and accurate assessment of feather cover on a continuous variable.
Key Words: laying hen • feather score • infrared thermography
|
DESCRIPTION OF PROBLEM
|
|---|
Feather losses in laying hens are predominantly due to age, wear, and feather pecking, a behavior that stems from a complex interaction of environmental, social, and genetic factors. The assessment of feather cover is an important factor for assessing bird welfare and, by extension, welfare-friendly production practices. Currently, the most commonly used method is feather scoring, which is a visual assessment of the feather cover at several anatomical locations, followed by allocation of a subjective score. The subjectivity of feather scoring may result in higher variation and lower repeatability than objective assessment on a continuous variable such as radiated temperature. In addition, feather scoring does not provide a measure of actual feather cover. Infrared thermography has the potential to be an objective, accurate, and repeatable method of assessing feather cover with obvious advantages over current subjective assessments. However, to assess infrared thermography (IRT) temperature data, it was necessary to compare these data to an "independent" method of assessing plumage. Thus, the independent measure in this study was a rating of feather scoring (FS), using the method described by Tauson et al. [1].
Each pixel in an infrared image has an associated temperature (°C); thus, an IRT image is effectively a temperature map of the object. Radiated temperature is more responsive to change than core body temperature and provides a sensitive measure of thermoregulatory fluctuations that occur in response to disease and stress [2, 3, 4]. Infrared thermography has been used in poultry to study the effects of heat stress and methods of cooling birds; for example, water spraying [5]. Feathers have a primary role in insulating birds, and therefore bare skin radiates higher temperatures in the infrared spectrum than surrounding feathers. The objective of the present experiment was to assess the use of IRT for the assessment of feather cover, or lack of cover (areas of bare skin). The principal approach of this study was to compare temperature measures among populations of birds independently classified by FS. Two IRT measurement parameters were tested for significance against FS categories. The first parameter to be statistically tested was the mean image temperature. This provided an evaluation of the IRT method that could be used in real time. The second approach was to define temperature ranges that specifically evaluated areas of feather cover and bare skin.
|
MATERIALS AND METHODS
|
|---|
The University of Alberta Faculty of Agriculture, Forestry and Home Economics Animal Care Committee approved the research protocol according to Guidelines of the Canadian Council on Animal Care.
Ninety-eight (n = 98) Brown Leghorn hens at the University of Alberta Poultry Research Center were assessed for feather cover. The hens were housed in colony cages consisting of 2 tiers of 12 cages. Cages provided 580 cm2 of floor space per bird and were fitted with nest boxes (60 x 20 cm) and a wooden perch. Birds were fed a standard laying hen ration and were 60 wk of age at the time of evaluation. Live weights were recorded using a hanging scale and infrared images were taken from a fixed distance (1.5 m) while the birds were suspended. At this point, each bird was independently assigned an FS by an experienced rater on 6 areas of the body (neck, back, wings, tail, front, and vent) using an established 4-point scale [1]. Good feather cover was rated as 4 and very poor cover as 1. Infrared images were not rated for FS.
Infrared images were recorded of the front, dorsal, and vent aspects using a S60 Therma-CAM infrared camera [6] and analyzed for temperature information using ThermaCAM Researcher Pro (v2.7) software [7]. Temperature data provided by the IRT camera during operation include the minimum, maximum, range, and mean image temperatures. Of these parameters, only the mean image temperature was likely to be related to feather cover. However, the mean of the total image area is heavily influenced by extraneous temperatures; that is, outside the periphery of the bird. Consequently, before capturing temperature data, background and extraneous temperature information were excluded by defining an area of interest using a drawing tool provided with the image analysis software. The defined area encompassed the surface of the bird on which the FS ratings were made. Thus, the neck, head, legs, background, and handler were excluded. Within the defined area of the image, 99.9% was within a temperature range of 17 to 38°C. Statistical analyses were performed on temperature and area parameters from data derived from the defined area. However, much of the temperature information on an IRT image is not related to either feather or skin temperature but contributes to the mean temperature. Furthermore, the mean image temperature does not convey any information about the distribution of temperatures and therefore of feathers or bare skin. Because temperature distribution was likely to more accurately reflect feather cover, the second part of the statistical analyses examined the relationship between temperature per unit of area and FS categories. It was necessary to define temperature ranges that corresponded to feathers and skin. The image analysis software represented the temperature distribution as the number of pixels within temperature intervals of 1°C. The numbers of pixels within each interval was expressed as the percentage of the total number of pixels in the defined area. Each interval (% area) was tested against FS categories using 1-way ANOVA. From these analyses, all areas within a temperature range of 17.25 to 24.25°C were found to be statistically positively associated with FS categories. Thus, the sum of the areas within the 17.25 to 24.35°C range was taken to represent feather cover. Conversely, areas within the range of 28.25 to 31.25°C were statistically negatively associated with FS categories. The sum of areas within the range 28.25 to 31.25°C was used to represent lack of feather cover and tested for significance among FS categories. The data presented are from the dorsal view. Similar results were obtained for the front and vent views but are not presented in this paper.
Statistical analyses [8] examined the relationship among and to FS categories for the mean image temperature (Table 1
and 2) and for the percentage area within the temperature range of 28.25 to 31.25°C (Table 3). The assessment of the mean image temperature was used to verify a real-time measure of feather cover. Statistical analyses compared mean values among FS categories and correlation between temperature and area parameters across FS categories. The Kruskal-Wallis test is a nonparametric comparison of the rankings achieved by converting the range of mean temperatures into ranks based on the ordered differences among means and comparing these ranks to those of the FS scale.
View this table:
[in this window]
[in a new window]
|
Table 3. Mean (SE) areas of the dorsal view within the 28.25 to 31.25°C temperature range for feather score (FS) categories
|
|
|
RESULTS AND DISCUSSION
|
|---|
Figure 1 shows representative IRT images of the dorsal area of birds rated in FS categories 1 to 4. The mean temperature within each FS category and the results of a Student’s t-test for all means are given in Table 1
. In all image views, the mean image temperature declined with increasing feather cover. The largest and statistically significant difference among mean image temperatures occurred for FS1 and FS2 categories and was consistent across all image views (dorsal, front, and vent). However, there were less statistical differences in mean temperature among FS3 and FS4 categories.
View larger version (115K):
[in this window]
[in a new window]
|
Figure 1. Representative infrared images of laying hens in feather score (FS) categories FS1 to FS4. Mean temperatures: FS1 = 30.6°C; FS2 = 25.6°C; FS3 = 23.5°C; FS4 = 20.4°C. Feather score was evaluated on a 4-point scale, in which good feather cover was rated as 4 and very poor cover as 1.
|
|
Table 2 gives the results of 1-way ANOVA (r-value and F-ratio) and nonparametric test (Kruskal-Wallis) of the relationship between mean temperatures and FS scores. Statistical analyses revealed a highly significant relationship between mean image temperature and FS categories for all views (Table 2). However, the relationships were statistically stronger for the dorsal and vent views relative to the front view. These data demonstrated that the mean image temperature could provide a real-time assessment of feather cover. It was noteworthy that the Pearson correlation demonstrated significant relationships (P < 0.01) for the mean image temperature among different views. Correlation coefficients were r = 0.56, 0.43, and 0.50 for front vs. dorsal, front vs. vent, and dorsal vs. vent, respectively.
Figure 2 shows the temperature distribution profiles of the front, dorsal, and vent views. Each curve represented 99.9% of the total defined image area. In all views, the temperature distributions across the population of birds were skewed toward the higher temperature ranges. The interaction of temperature and area was represented as the percentage of the total image area within temperature intervals of 1°C. The percentage area within each 1°C interval was tested for significant correlations against FS scoring using 1-way ANOVA. These analyses were conducted for all image views but only those data derived from analysis of the dorsal images are presented. From ANOVA, 2 sets of areas were significantly associated with FS. Summing areas of statistical significance identified a temperature range of 17.25 to 24.25°C that was positively associated with FS; i.e., feather cover. Areas within the range 24.25 to 27.25°C were not correlated with FS. Note that the areas in this temperature range contributed to the mean temperature but were irrelevant for assessing feather cover. Areas representing 1°C intervals from 28.25 to 31.25°C were significantly negatively associated with FS; i.e., bare skin. Table 3 gives the mean (SE) and 95% confidence intervals for the area within the range 28.25 to 31.25°C for the dorsal views. The areas within the higher temperature range (28.25 to 31.25°C) represented between 2.1 to 15.2% of the total defined image areas across FS categories. The means of the percentage areas within the defined temperature range (28.24 to 31.25°C) were significantly different among all FS categories, with very little overlap among FS categories for 95% confidence intervals. Note that statistical separation of all FS categories did not occur for the mean image temperature (Table 1). Thus, although similar statistical relationships were observed for the mean temperature and the percentage area within the range 28.25 to 31.25°C, the latter parameter was superior to the mean temperature in distinguishing among FS categories. Figure 2 shows the temperature distributions of each of the views for all birds. The distributions among views did not precisely overlap but all were skewed toward the higher end of the temperature scale. In addition, Pearson correlations of mean temperature among views were statistically significant. Consequently, it is not surprising that similar results were found with front and vent images, but with slightly different temperature ranges representative of either feather cover or bare skin. Significant correlations among views indicate that in most cases an accurate assessment of the whole bird could be made from one view.
The study demonstrated a highly statistically significant relationship between radiated temperature and feather cover, as judged by FS. A real-time measure of feather cover can be obtained from the mean image temperature. Future refinements in the measurement of feather cover by IRT will include capturing the image against a cooler background. The camera software can be used to automatically subtract background temperatures leaving an outline of the bird from which the mean image temperature could be used for a real-time assessment. However, a more detailed assessment, particularly for research purposes, could be achieved by expressing the area of the image within specified temperature ranges. Thus, for the dorsal view, the area of the image in the 17 to 24°C range gave a measure of feather cover and within a range of 28 to 31°C provided a measure of bare skin. Feather scoring is the accepted method of assessing feather cover and as such provided an independent variable to which the infrared data were compared. However, the data derived from infrared thermography provide a continuous variable that more accurately reflects either actual feather cover or areas of bare skin.
|
CONCLUSIONS AND APPLICATIONS
|
|---|
- The loss of insulation resulting from poor feather cover is directly measurable as heat loss in the infrared spectrum and quantified by IRT.
- Real-time assessment of feather cover by IRT is achievable using the mean image temperature, although care should be taken to reduce or eliminate temperature information from extraneous sources.
- Quantitative measurements of feather cover, or bare skin, can be made from comparative analysis of areas of the image within specified temperature ranges.
|
ACKNOWLEDGMENTS
|
|---|
The authors gratefully acknowledge the technical assistance provided by Sigrid Lohmann and Pierre Lepage of Agriculture and AgriFood Canada, Lacombe Research Centre, Alberta, Canada.
|
REFERENCES AND NOTES
|
|---|
- Tauson, R., T. Ambrosen, and K. Elwinger. 1984. Evaluation of procedures for scoring the integument of laying hens—Independent scoring of plumage condition. Acta Agric. Scand. 34:400–408.
- Berry, R. J., A. D. Kennedy, S. L. Scott, B. L. Kyle, and A. L. Schaefer. 2003. Daily variation in the udder surface temperature of dairy cows measured by infrared thermography: Potential for mastitis detection. Can. J. Anim. Sci. 83:687–693.
- Schaefer, A. L., N. J. Cook, S. V. Tessaro, D. Deregt, G. Desroches, P. L. Dubeski, A. K. W. Tong, and D. L. Godson. 2004. Early detection of infection using infrared thermography. Can. J. Anim. Sci. 84:73–80.
- Cook, N. J., J. S. Church, A. L. Schaefer, J. R. Webster, L. R. Matthews, and J. M. Suttie. 2005. Stress and pain assessment of velvet antler removal from elk (Cervus elaphus canadensis) and reindeer (Rangifer tarandus). Online J. Vet. Res. 9:13–25.
- Chepete, H. J., and H. Xin. 2000. Cooling laying hens by intermittent partial surface sprinkling. Trans. ASAE 43:965–971.
- Infrared Camera S60 ThermaCAM, Flir Systems Inc., Santa Barbara, CA.
- Infrared image software ThermaCAM Pro (v2.7), Flir Systems Inc., Santa Barbara, CA.
- Statistical Analysis Software, JMP (v5.1), SAS Institute, Inc., Cary, NC.
This article has been cited by other articles:
|
|
|
|
 
A. Colak, B. Polat, Z. Okumus, M. Kaya, L. E. Yanmaz, and A. Hayirli
Short Communication: Early Detection of Mastitis Using Infrared Thermography in Dairy Cows
J Dairy Sci,
November 1, 2008;
91(11):
4244 - 4248.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
SUMMARY
DESCRIPTION OF PROBLEM
