Image Process Tool an Alternative Method for Quantitative Assessment of Mammary Gland Structure in Mohabadi Goat

Document Type : Research Articles

Authors

1 Department of Animal Science, Faculty of Agriculture, University of Tabriz, Tabriz, Iran

2 Animal Breeding and Genetics Group, Ghiam Dairy Complex, Isfahan 833111-4761, Iran

Abstract

Milking ability in goat farming is not particularly new, however, it has frameworks for receiving economic benefits from marketing of milk. For this reason, investigation of varied methods plays an important role to evaluate mammary gland structure in dairy animal. Hence, the overall objective of current work was to test the image processing tools as an alternative method for quantitative assessment of mammary gland structure in Mohabadi goat. Under this circumstance, survey data was collected from 72 Mohabadi does from research station of Department of Animal Science, University of Tehran. The fixed effects for each candidate animal consists of pedigree, birth date, milk characteristics as well as monitoring the history of record during assessment was monitored. Image process tools was based on two stages: First, three different camera photos were taken from different angels of mammary gland structure from both sides of each animal and next step, Digimizzer image processing software was employed for further analysis and estimation of seven morphological dimensions (udder length, teat length and teat angle at both sides and as well as udder height). To investigate this statistically, the descriptive statistics, correlation, different regression analysis, principal component analysis and as well as association study between linear mammary gland traits and milk were analyzed using SAS 9.3 software. The highest correlation was observed in the length of right and left udder (r=0.67 P<0.05). The results revealed that the milk yield was significantly influenced by the length of the right teat (P<0.05). In conclusion, digital image processing may be considered as alternative tools for linear morphological characterization that would provide more accurate observation and measurements on the indigenous goat population.

Keywords


INTRODUCTION

In dairy goat farming, milk production is an increasingly marked area due to a wide range of advantages and wealth of human health (Abd-Allah et al. 2011). Milking ability in goat farming is not particularly new, however, it has frameworks to receive economic benefits of milk marketing (El-Awady and Oudah, 2011). For this reason, the investigation of varied methods plays an important role to evaluate mammary gland structure in dairy animals (Capote et al. 2006). Traditionally, to recognize high potential milk producer animals, farmers generally focus on visible morphological appearance of mammary gland structure as the main organ of milk secretion using visual assessments. This kind of measurement is complex problem and to simplify it is required to understand key factors affecting the parameters which in summary included nutrition and feed formulation, lactation stage (Knight and Wild, 1993), breed (Kominakis et al. 2009), age of does (Kor et al. 2005), kidding type (Ketto et al. 2014), parity (Adegoke et al. 2016), offspring sex (Carnicella et al. 2008), year and season of kidding (Hilali and Jesry, 2009), milking frequency (Ishag et al. 2012) and type of milking system (Fatima et al. 2019), lactation rate (James and Osinowo, 2004) and milking time (Katanos et al. 2005), udder size are important (Berry et al. 2004). The strong association between the anatomy of mammary gland structure, udder attachments, and milk production of dairy livestock have been found in several recent studies (Altincekis and Koyucu, 2011; Sadeghi et al. 2013). For this reason, understanding the key factors influencing mammary gland structure is the appropriate scenario for research and several previous researches focused on the estimation of genetic parameters and the heritability coefficient for linear mammary gland structure in different farm animals (Amao et al. 2003). It is critical to have initial phenotypic records in the breeding of any candidate traits during the production stage (Carta et al. 2009). Traditionally, the most of traits have been performed using conventional tools such as scales, meters, laboratory measurements or precision volumetric tools (Fernandez et al. 1995). The summary of previous reports of heritability coefficient for udder characteristics showed the low to medium level as including: udder depth (0.16), udder attachment (0.17), teat placement (0.24), teat size (0.18) and udder shape (0.24) were similar to those for dairy cattle which reflect a positive genetic response to direct selection (Fernandez et al. 1997; McLaren et al. 2016). Genetic correlations among udder traits were generally favorable, implying that selection for improvement of one trait would result in improvement of others (Martinez et al. 2011). However, the process of recording is not accurate, time-consuming, laborious, and costly. One of the major drawbacks of biometric measurements in animal farming is that the unstable nature of the animal and its aggressive behavior as well as unwanted mobility which often make measurements wrong and not trustable; therefore, alternative methods are required to minimize contract and touch of animal (Torres et al. 2012). Image processing is an alternative technology inside machine vision science used to extract information from a simple image and to create accurate inputs for further interpretation (Sadeghi et al. 2016). Image processing is also one of the dependent tools of artificial intelligence and it is defined as the science for using the properties of constant or moving objective to identify and make decisions about different people or subjects (Sadeghi et al. 2014). As detail, in the present technology, instead of being directly measured by humans, the image is first taken from the subject or object, and then it is analyzed using various computer and software features and finally by appropriate algorithm design. In recent decades, the use of image processing has been developed in many disciplines. Main application is namely, agriculture-related industries and nowadays researchers have investigated the various methods to use image processing techniques to facilitate and enhance the accuracy of assessment and decision-making systems in various agricultural sciences. Animal processing technology today has many applications such as carcass quality, body and breast biometrics studies, skin quality determination, egg quality determination, live animal weight estimation, sperm motility and viability studies in reproductive physiology and animal behavioral issues (Khojastekey and Dianat, 2016). Mahabadi goat is one of the indigenous goats of Iran and is considered for its meat, milk, and fiber production. Geographically, this goat was distributed in the south of west Azerbaijan province and in Mahabad, Piranshahr and Sardasht cities. Marketing live body dependence of the sex of animal is between 50 above 80 kg. Twining percentage is estimated at 70-80% and 216- liter milk during 120 day of the lactation period. The overall goal of this work was to test the image process tool an alternative method for quantitative assessment of mammary gland structure in Mohabadi goats.

 

MATERIALS AND METHODS

Animal and geographical location

The present study started in the first middle of July- the end of November 2019. Under this circumstance, survey data was collected from 72 Mohabadi does from research station of Department of Animal Science, University of Tehran. Fixed effects for each candidate animal in this work consisted of pedigree, birth date, milk characteristics as well as monitoring the history of record during assessment. Image process tools were based on two stages:

First, three different camera photos were taken from different angels of mammary gland structure of each animal which in turn, seven morphologic measurements including udder length, teat length and teat angle from both side and also udder height were measured. These parameters were measured by the protocol as described by Labussière et al. (1981). Figure 1 illustrated the detail of taking photo and position of each candidate animal for investigation. 

Next step, Digimizzer image processing software was used for further analysis and estimation of morphological dimensions. The Descriptive statistics for Digimizzer image processing in this study are shown in Figure 2.

 

Figure 1 Phenotype of Mohabadi does and identification of investigated animal

 

Figure 2 Measurements and understanding of udder morphological characteristic

1: left teat length (cm); 2: right teat length (cm); 3: height udder (cm); 4: right udder length (cm); 5: left udder length (cm); 6: left teat angle (degree) and 7: right teat angle (degree)

 

Statistical analysis

To investigate this statistically, the descriptive statistics, correlation, regression, principal component analysis and association study between linear mammary gland trait and milk were calculated using SAS 9.4 software (SAS, 2004). Before enter to Generalized Linear model (GLM) analysis, Shapiro-wilk test was used for residual values of each certain measurement. Table 1 shows the condition for classification of investigated traits based on outputs of descriptive statistics. Principal component analysis (PCA) according to correlation outputs was done and Kaiser-Meyer-Olkin Measure (KMO) and Bartlett's Test were employed for the power of sample size and suitability from raw data volume.

 

Table 1 Condition for classification of investigated traits based on outputs of descriptive statistics

R: right and L: left.

 

Value above 70 for KMO may demonstrate input file has to permit for further interpretations (Keskin et al. 2005; Keskin et al. 2007). Eventually, the regression equation of daily milk yield on udder morphological characteristic was determined using the stepwise selection.

 

RESULTS AND DISCUSSION

From the above-mentioned short review, key findings emerge that the descriptive statistics for Digimizzer image processing in this study are shown in Table 2. As can be seen, results show a reasonable variation for measured parameters within the investigated population. Notably, teat angels and teat length showed the highest and lowest variation among parameters. Linear regression equation of effective morphological factors and as well as lactation period on milk production was done and stepwise regression linear equation for estimation of daily milk production for Mahabadi breed was as follows:

Y= 799.07 + 58 X1 - 2.35 X2

Where:

Y: variable (daily milk yield).

X1: right udder length.

X2: lactation day.

To justify this equation, it can be stated that since the animals in this study have passed the lactation peak (4-6 weeks), the amount of milk production and lactation day will be negatively correlated. As correlation analysis of udder characteristics provided a low to a medium correlation between studied parameters overall (Figure 3). The present findings confirmed that the length of the right teat was showed the significant factor affecting milk yield during lactation. Regression outputs also revealed the mean milk yield decreased as a result of increased right teat length. Other measured parameters showed no significant response during statistical interpretation. Figure 3 clearly indicated Rteat has negative significant correlation with most measured parameters and also Rhight and Lhight has positively correlated in the analysis. Furthermore, the results of PCA to condense the information for udder morphological characteristics with minimal loss of information were reported in Table 3. The first 5 components (left teat length; right teat length; height udder; right udder length; left udder length) and the second 2 components (left and right teat angles) consisting of positive coefficients explained 37.46% and 21.84% of the total variance respectively, which is considered sufficient to capture most of the variation for udder morphological traits. Figure 4 illustrated the significance statue between milk and udder morphological structure. When lactation period increased the dimension of mammary gland and teat have tendency to change. Udder characteristics in dairy animals are directly involved in economic production (Emediato et al. 2008). Udder shape and its related traits are effective on milk production and easy milking (Perez-Cabal et al. 2013; Sadeghi et al. 2014), and high correlation of mammalian traits with quantitative and qualitative traits of milk is an appropriate factor and an accessible way to select dairy cattle by breeders (Eyduran et al. 2013). It has been reported that nutritional status (Morand-Fehr et al. 2007), animal genotyping (Sezenler et al. 2016), pregnancy (Rogers and Spencer, 1999), lactation, lactation stage (Skapetas et al. 2005), weight, physical condition (Merkhan, 2014), animal health status (Milerski et al. 2006) and breeding systems (Mckusick et al. 1999) are the most prominent factors affecting udder morphology. We hypothesized that the pattern and variability of udder morphology in Mohabadi goat could influence milk yield and can to play as criteria for identification does with high potential for milk production. The study results provided some interesting findings regarding the correlations between udder and milk. The key outputs of our study demonstrated that the right udder length is a significantly associated with higher milk production.

 

Table 2 Descriptive statistics for Digimizzer image processing in current study

R: right and L: left.

 

Figure 3 Correlation outputs and significant statue of seven investigated udder characteristics

Left teat length (Lengle); Right teat length (Rengle); Height udder (high); Right udder length (Rhight); Left udder length (Lhight); Left teat angle (Lteat): Right teat angle (Rteat)

Dark blue color is the highest positive correlation and dark red color also representative of the highest negative correlation pattern

 

To understand this finding, due to the existence of milk in the right side of the dairy animal, or by the development of more alveoli and blood vessels in the right side of the udder, or by greater sucking of the right teat in the animals studied by the offspring. Those that release more oxytocin and thus facilitate milk excretion from the right udder. Here, we compared the finding of the proposed image processing method with those of the traditional methods. In line with our highlights, Sadeghi et al. (2013) emphasized a significant an association between right teat length and milk yield during lactation period. Contrary to our findings, some similar reports highlighted the association between other udder morphology on milk and in addition to, authors pointed different pattern of correlation in their goat population. Table 4 showed a wide board of similar literatures on morphological characteristics of goat and sheep with traditional method of measurement. As reasonable justification, this inconsistency may be due to the multiple possibilities and invisible different between the condition of study designs, including nutrition and feed formulation, lactation stage, breed, age of does, kidding type parity, offspring sex, year and season of kidding, milking frequency and type of milking system lactation rate and milking time, udder size.

 

Table 3 Principal component (PC) analysis (Rteat, Lhigh, Rhigh, Lteat, high, Lengle, Rengel) outputs and extracted component and related variance values in this study

1 Extraction Method: Principal component analysis.

 

Figure 4 Results of significance statue between milk and udder morphological structure class; Rteat length (cm), Lteat length (cm), Height udder (cm), R udder length (cm), L udder length (cm), Rangle (degree) and Langle (degree)

 

However, this study is restricted with several limitations, such as sample size and history of milking, availability of fixed effects are not sufficient and therefore, it is not surprising to see that low significance during the interpretation of data. Therefore, further validation will be required using high inputs and goat breeds. There are some limitations of this study. For example, sample size and history of milking, availability of fixed effects are not sufficient and therefore, it is not surprising to see that low significance during interpretation of data. Therefore, further validation will be required using the high inputs and goat breeds.

 

Table 4 Summary for similar pieces of literature on morphological characteristics

 

CONCLUSION

Udder morphological traits are important parameters traditionally used in dairy goat production. They are essential for accurately monitoring milk ability of herds. On this basis, we conclude that the length of right and left udder showed the highest correlation. The results demonstrated that the length of the right teat significantly influenced milk yield. As a final message, digital image processing may be considered as alternative tool for linear morphological characterization that would provide more accurate observation and measurements on indigenous goat population.

 

ACKNOWLEDGEMENT

We extend our thanks to both Dr Mohammad and Hossein Moradi Shar-e-Babak for giving permission for recording of his project animals and providing opportunity for collaboration with university of Tehran. The authors appreciate the unknown referee’s valuable and profound comments.

Abd-Allah M., Abass S.F. and Allam F.M. (2011). Factors affecting the milk yield and composition of Rahmani and Chios sheep. Int. J. Livest. Prod. 2, 24-30.
Adegoke E.O., Machebe N.S., Ezekwe A.G. and Agaviezor O.B. (2016). Effect of parity on changes in udder traits, milk yield and composition of West African dwarf sheep during lactation. Anim. Reprod. Sci. 57(6), 1047-1057.
Akbaş A.A., Elmaz Ö., Sari M. and Saatci M. (2019). Assessment of some udder and teat traits of Honamlı goats in terms of dairy characters. J. Res. Vet. Med. 38(2), 57-64.
Altincekis S.O. and Koyucu M. (2011). Relationship between udder measurements and the linear scores for udder morphology traits in Kivircik, Tahirova and Karacabey Merino ewes. Kafkas Univ. Vet. Fac. Derg. 17, 71-76.
Amao O.A., Osinowo O.A., Onwuka C.F.I., Abiola S.S. and Dipeolu M.A. (2003). Evaluation of udder traits in West African Dwarf goats. Nigerian J. Anim. Prod. 30, 246-252.
Berry D.P., Buckley F., Dillon P., Evans R.D. and Veerkamp R.F. (2004). Genetic relationships among linear type traits, milk yield, body weight, fertility and somatic cell count in primiparous dairy cows. Iranian J. Agric. Food Res. 43, 161-176.
Capote J., Arguello A., Castro N., Lopez L.J. and Caja G. (2006). Correlations between udder morphology, milk yield, and milking ability with different milking frequencies in dairy goats. J. Dairy Sci. 89, 2076-2079.
Carnicella D., Dario M., Ayres M.C.C., Laudadio V. and Dario C. (2008). The effect of diet, parity, year and number of kids on milk yield composition in Maltese goat. Small Rumin. Res. 77, 71-74.
Carta A., Casu S. and Salaris S. (2009). Current state of genetic improvement in dairy sheep. J. Dairy Sci. 92, 5814-5833.
El-Awady H.G. and Oudah E.Z.M. (2011). Genetic and economic analysis for the relationship between udder health and milk production traits in Friesian cows. J. Anim. Sci. 24, 1514-1524.
Emediato R.M.S., Siqueira E.R., Stradiotto M.M., Maesta S.A. and Fernandes S. (2008). Relationship between udder measurements and milk yield in Bergamasca ewes in Breazil. Small Rumin. Res. 75, 232-235.
Eyduran E., Yilmaz I., Kaygisiz A. and Aktas Z.M. (2013). An investigation on relationship between lactation milk yield¸ somatic cell count and udder traits in first lactation Turkish Saanen goat using different statistical techniques. J. Anim. Plant Sci. 23, 956-963.
Fernandez G., Alvarez P., San Primitivo F. and De La Fuente L.F. (1995). Factors affecting variation of udder traits of dairy ewes. J. Dairy Sci. 78, 342-849.
Fernandez G., Baro J.A., De La Fuente L.F. and San Primitivo F. (1997). Genetic parameters for linear udder traits of dairy ewes. J. Dairy Sci. 80, 601-605.
Hilali D.M.L. and Jesry T.G. (2009). Udder measurements and milk production in two Awassi sheep genotypes and their crosses. J. Dairy Sci. 92(9), 4613-4620.
Iñiguez L. and Hilali M. (2009). Evaluation of Awassi genotypes for improved milk production in Syria. Livest. Sci. 120, 232-239.
Ishag I.A., Abdalla S.A. and Ahmed M.K.A. (2012). Factors affecting milk production traits of Saanen goat raised under Sudan-semi arid conditions. J. Anim. Feed Res. 5, 435-438.
James I.J. and Osinowo O.A. (2004). Changes in udder size and live weight of West African Dwarf, Red Sokoto and Sahel goats during lactation and their phenotypic 88 relationship with partial daily milk yield. Nigerian J. Anim. Prod. 31, 119-129.
Katanos J., Skapetas B. and Laga V. (2005). Machine milking ability and milk composition of some imported dairy goat breeds and some crosses in Greece. Czech J. Anim. Sci.
50, 394-401.
Keskin S., Kor A. and Karaca S. (2007). Use of factor analysis scores in multiple linear regression model for determining relationships between milk yield and some udder traits in goats. J. Appl. Anim. Res. 31, 185-188.
Keskin S., Kor A., Karaca S. and Mirtagioglu H. (2005). A study of relationships between milk yield and some udder traits by using of path analysis in Akkeci goats. J. Anim. Sci. 4, 547-550.
Ketto I.A., Massawe I. and Kifaro G.C. (2014). Effects of supplementation, birth type, age and stage of lactation on milk yield and composition of Norwegian × Small East African goats in Morogoro, Tanzania. Livest. Res. Rural Dev. Available at: http://www.lrrd.org/lrrd26/12/kett26234.html.
Khojastekey M. and Dianat R. (2016). The image processing technology and over review of some result of its use in animal husbandry. J. Appl. Anim. Res. 17, 35-44.
Knight C.H. and Wild C.J. (1993). Mammary cell changes during pregnancy and lactation. Livest. Prod. Sci. 35, 3-19.
Kominakis A.P., Papavasiliou D. and Rogdakis E. (2009). Relationships among udder characteristics, milk yield and, non-yield traits in Frizarta dairy sheep. Small Rumin. Res. 84, 82-88.
Kor A., Karaca S., Mirtagioglu H. and Keskin S. (2005). Study of relationships between milk yield and some udder traits by using of path analysis in Akkeci goats. J. Anim. Vet. Adv. 5, 547-550.
Labussière J., Dotchewski D. and Combaud J.F. (1981). Caractéristiques morphologiques de la mamelle des brebis Lacaune. Méthodologie pour l’obtention des donneés. Relations avec l’aptitud à la traite. Ann. Zootech. 30, 115-136.
Martinez M.E., Calderon C., De La Barra R., De La Fuente F. and Gonzalo C. (2011). Udder morphological traits and milk yield of Chilota and Suffolk sheep breeds. Chilian J. Agric. Res. 71, 90-95.
Mckusick B., Berger Y. and Thomas D. (1999). Effects of udder morphology on commercial milk production of East Friesian crossbred ewes. Pp. 55-559 in Proc. 5th Great Lakes Dairy Sheep Symp., Great Lakes, USA.
McLaren A., Mucha S., Mrode R., Coffey M. and Conington J. (2016). Genetic parameters of linear conformation type traits and their relationship with milk yield throughout lactation in mixed-breed dairy goats. J. Dairy Sci. 99, 5516-5525.
Merkhan K.Y. (2014). Milk traits and their relationship with udder measurements in Awassi ewes. Iranian J. Appl. Anim. Sci. 4, 521-526.
Milerski M., Margetín M., Čapistrák A., Apolen D., Špánik J. and Oravcová M. (2006). Relationships between external and internal udder measurements and the linear scores for udder morphology traits in dairy sheep. Czech J. Anim. Sci. 51(9), 383-390.
Morand-Fehr P., Fedele V., Decandia M. and Le Frileux Y. (2007). Influence of farming and feeding systems on composition and quality of goat and sheep milk. Small Rumin. Res. 68, 20-34.
Perez-Cabal M.A., Legaz E., Cervantes I., De La Fuente L.F., Martinez R., Goyache F. and Gutierrez J.P. (2013). Association between body and udder morphological traits and dairy performance in Spanish Assaf sheep. Arch. Tierz. 42, 430-442.
Rogers G.W. and Spencer S.B. (1991). Relationships among udder and teat morphology and milking characteristics. J. Dairy Sci. 74, 4189-4194.
Sadeghi S., Rafat A. and Bohlouli M. (2014). Effect of crossbreeding on linear udder scores and their phenotypic relationships in Iranian fat-tailed ewe's. Biotechnol. Anim. Husb. 30(1), 61-77.
Sadeghi S., Rafat A., Ghaderi Zefrei M., Khaligh F., Rostami K., Bohlouli M., Bahrani Behzadi M. and Mohaghegh M. (2013). Factors affecting external and internal mammary morphology traits and assessment of their interrelationships with milk yield in Lori Bakhtiari breed ewes. Livest. Res. Rural Dev. Available at: http://www.lrrd.org/lrrd25/3/sade25037.htm.
Sadeghi S., Rafat S.A., Moghadam G. and Janmohammadi H.M. (2016). Measuring of udder morphological characteristics among crossbred and pure sheep breeds. Iranian J. Appl. Anim. Sci. 6(2), 343-348.
SAS Institute. (2004). SAS®/STAT Software, Release 9.4. SAS Institute, Inc., Cary, NC. USA.
Sezenler T., Ceyhan A., Yuksel M.A., Koncagul S., Soysal D. and Yildirir M. (2016). Influence of year, parity and birth type on milk yield and milk components of Bandirma sheep (German Black Head Mutton x Kivircik). J. Agric. Sci. 22, 89-98.
Skapetas B., Mazaraki K., Katanos I., Laga V. and Matara H. (2005). Effect of lactation and prolificacy on milkability, composition and somatic cell count of milk of crossbred goats Saanen × native Greek breed. Epeirotike Hestia. 33, 35-51.
Suárez-Trujillo A., Capote J., Argüello A., Castro N., Morales-DelaNuez A. and Torres A. (2013). Effects of breed and milking frequency on udder histological structures in dairy goats. J. Appl. Anim. Res. 41, 2.1-7.
Torres A., Castro N., Hernández-Castellano L.E., Argüello A. and Capote J. (2012). Short communication: Effects of milking frequency on udder morphology, milk partitioning, and milk quality in 3 dairy goat breeds. J. Dairy Sci. 96, 1071-1074.
Türkyılmaz D., Özyürek S., Esenbuğa N. and Yaprak M. (2017). Correlation between various udder measurements and milk components in Morkaraman, Tuj and Awassi sheep. Pakistan J. Zool. 50(5), 1921-1927.