Diagnosis of lameness via data mining algorithm by using thermal camera and image processing method in Brown Swiss...

Area of Science:

• Veterinary Medicine
• Animal Science
• Data Mining in Agriculture

Background:

• Lameness is a significant economic factor in cattle herd management, impacting animal health and welfare.
• Early detection of lameness is crucial for timely intervention and reducing economic losses.
• Traditional methods for lameness assessment can be subjective and time-consuming.

Purpose of the Study:

• To evaluate the efficacy of a data mining algorithm for early lameness detection in Brown Swiss cattle.
• To integrate lameness scores with thermal image parameters (Lab, HSB, RGB) for improved diagnostic accuracy.
• To identify key variables for classifying healthy versus lame animals using thermal imaging.

Main Methods:

• Thermal images of the fetlock joints of 33 Brown Swiss cattle were processed using ImageJ software.
• Data mining, specifically the Classification and Regression Tree (CART) algorithm, was employed for lameness diagnosis.
• Independent variables included skin surface temperatures and image parameters; lameness scores served as the binary response variable.

Main Results:

• The CART algorithm achieved high accuracy in classifying lame cattle (12 out of 13 correctly identified).
• Key predictors for lameness were maximum temperature (Tmax), green color mean, L (max) value, and animal age.
• The CART model demonstrated high sensitivity (92.31%), specificity (95%), and Area Under the ROC Curve (AUC) (93.7%).

Conclusions:

• The combination of CART algorithm, thermal imaging, and image processing is a valuable tool for detecting lameness in cattle herds.
• This approach offers a promising method for early and objective lameness diagnosis.
• Further research with larger animal cohorts is recommended to validate these findings.