Automated analysis of postoperative behaviour: assessment of HomeCageScan as a novel method to rapidly identify pain and analgesic effects in mice

Abstract
This study evaluated whether the automated behaviour recognition software ‘HomeCageScan’ (HCS) could detect behaviour changes and any positive analgesic effects in two mouse strains undergoing vasectomy (C3H/HeNCrl and C57BL/6). Another objective was to test the effectiveness of HCS in differentiating between the effects of each treatment relative to conventional manual analysis. Each control (unoperated) group consisted of four mice of each strain. They were either untreated mice, mice given meloxicam alone (10 mg/kg) or mice given either saline or meloxicam (10 mg/kg) 30 min prior to isoflurane anaesthesia. The vasectomized mice received either saline or meloxicam at 5, 10 or 20 mg/kg, again, 30 min prior to isoflurane anaesthesia. Filming began one hour following surgery. Each mouse was filmed for 6 min for the manual analysis and then for a further 20 min for analysis with HCS. In a time-matched test, HCS and the manual analysis produced activity data that generated identical conclusions regarding treatment effects and strain differences. Both HCS and the manual analysis found the C57BL/6 controls were overall more active, but not following vasectomy, when both types of analysis detected markedly reduced activity. Low-dose meloxicam (5 mg/kg) had a positive effect on postoperative mobility in the C3H/HeNCrl mice; however, increasing the dose rate progressively reduced this. These effects were also detected with the manual analysis. Overall, HCS provided a sufficiently accurate and rapid method of analysing mouse behaviour encouraging more prolonged assessments in the future. This capability and the possibility of training the software to recognize a greater range of behaviours, including pain-specific indicators, should be of considerable value for assessing postoperative behaviour in both mice and rats. This would allow analgesic requirements to be investigated in a greater range of rodent models than is currently feasible with conventional analysis methods.