Gait analysis of acceleration sensor data recorded from lameness and sound dairy cows

Introduction
There are few studies that objectively evaluate the effect of improving lameness after trimming and treatment of hoof disease in daily cow [1]. An easy-to-use, objective method of diagnosing lameness in dairy cows is also required. In this study, a wireless three-dimensional acceleration sensor (WAS) was used to analyze gait of lameness status in dairy cows. We also tried to compare periodicity of the gait between the lameness and normal cows in order to applied a new diagnose method of the lame. 

Materials and Methods
The 20 normal and 28 lame Holstein cows were used. They kept in the free-stool barns within the jurisdiction of the NOSAI Hyogo Veterinary Clinic Division and Experimental Farm of Iwate University. The wireless acceleration sensors (WAS : Bycen and Sumitomo Electric) were fixed on the dorsal side of thoracic vertebrae (T12). All cows walked voluntarily on a straight pathway for approximately 20 sec. The sampling frequency of the WAS set to 200 Hz and the length of a stored data was fixed at 1024 and/or 2048 points (Fig.1.). Each row data was calculated the autocorrelation value during 512 points (approximately 5 sec) of walking. The data of using calculation were shifted 64 points each and draw on the highlighting contour graph (Fig.2.). All data of acceleration signals were analyzed by the MATLAB and EXCEL software. 

Results
The averaged scalar value for lame cows was significantly increased compared to normal cows. In cases accompanied by remarkable pain, such as digital dermatitis (DD), the waveform of acceleration of lameness displayed a large amplitude wave on dorso-ventral direction (Fig.1AB). Autocorrelation coefficients of walking in normal cattle showed a continually stable gait periodicity (Fig.2A). However, the autocorrelation value of the DD cattle was low and indicated an unstable during walking (Fig2B). 

Conclusions
The WAS proved suitable for evaluating gait and for monitoring walking. Autocorrelation coefficient methods was useful to applied for gait analysis of the daily cow. This application will be suitable for detection of diseased legs and other parts, prognosis judgment for locomotor disability after medical treatment, development and assessment of new cures, and evaluation of lameness scores.