Object Recognition Test
The object recognition test (ORT) is a behavioral test that is widely used to examine animal memory performance. This test was first described by Ennaceur et al. (1988) and has been used in many different variations ever since. A major drawback of the ORT is that the scoring is done by manual scoring, which can be liable to subjectivity. Various attempts have been made to automate the scoring of the rats, but to our knowledge the use of specific software has not been published yet. Recently, researchers at Maastricht University developed and implemented an automatic scoring algorithm in LabVIEW that reliably tracks the nose of the rats.
VI Technologies used the algorithm as a starting point to create a professional application with several extra features and improvements to aid in automatic scoring.
The picture below depicts a schematic overview of the vision processing performed. At first an analog monochrome PAL camera was used in combination with a PCI-1411 analog framegrabber from National Instruments, later these where replaced by a GigE camera.
Nose detection algorithm
The detection algorithm of the nose is as follows. The maximal intercept of the rat is determined. Subsequently the Center of Mass (CM-1) is determined and the line of the maximal intercept is shifted so that the maximal intercept line crosses the Center of Mass. This results in three different X-Y coordinates: Center of Mass, intercept-body crossing front, intercept-body crossing back. Then, a perpendicular is placed on 50% of the intercept line. The perpendicular divides the body in two parts and the Center of Mass is again calculated for the front part (CM-2). The point where the maximal intercept crosses the perimeter of the body is regarded as the position of the nose (N-1, see Fig.1). Using this algorithm, which is executed in a cycle of 40 ms, the nose gradually shifts to the actual nose position (see shifts from N-1 (solid lines) to N-3 (dotted-dashed lines) in the figure below). This process is dynamic since the X-Y coordinates of the previous calculation are compared with the new input of X-Y coordinates. Thereby, this process continuously leads to the detection of the actual nose position.
The small movie below shows the realtime tracking of a rat.
Automated scoring of novel object recognition in rats publication on ScienceDirect.com
Imaging system automates memory recall analysis publicatie in Vision Systems Design
Automated scoring of novel object recognition in rats case study on ni.com
Minder fouten bij onderzoek objectherkenning door ratten publicatie in Vision & Robotics (in Dutch)