Automatic Recognition Method Of Mouse Optomotor Response Based On Image Technology

In ophthalmic experiments,researchers often need to judge the vision of mice.However,mice cannot express their subjective feelings,the visual acuity of mice often needs to be measured using specific experiments.This article introduces a detail method for automatically measuring the optomotor response of mice to solve the problem of measuring the visual acuity of mice in ophthalmic experiments.In the process of automatically measuring the mouse,the camera collects the images recording the mouse movement and transmits the image to the computer.The computer first recognizes the tracking of the mouse's head from the picture,and then detects the number and duration of the mouse's optomotor response from the tracking.Researchers then can judge the vision of the mice.This article first introduces the device used to measure the optomotor response of mice.The hardware part provides the basic experimental environment required for the optomotor response,while the software part can control the progress of the entire optomotor reaction experiment.The combination of the hardware part and the software part greatly improves the efficiency of the optomotor reaction experiment.The hardware part of optomotor response matched with software functions can greatly improve the optomotor response detection efficiency.Secondly,this article introduces the process of acquiring the mouse head orientation from the mouse motion images collected by the camera.First,the area threshold is used to filter the contours in the binarized image.Then the characteristics of the mouse ear and tail are used to initially identify the position of the mouse's snout.In order to obtain a more accurate mouse snout position,a technique of snout repositioning is used to reposition the mouse snout position,and then the head orientation of the mouse is calculated from the snout position.In order to solve the problem that the position of the snout fails to be recognized due to the uncertain movement of the mouse,this paper additionally proposes a method to identify the position of the snout of the mouse based on the neural network.The accuracy reaches 91.4% for those whose distance between the prediction result of the finally trained model network and the position of the snout is less than 4.3mm.Next,this paper proposes an algorithm to automatically calculate the number of optomotor reactions and the duration of optokinetic reactions in mice.The algorithm establishes the motion curve of the mouse's optomotor response from the head position data of the mouse,and then uses the interval-based segmentation method to detect the optomotor response between each cell using the three characteristics of slope,correlation coefficient,and difference standard deviation.Situation and connect consecutive optomotor response intervals together.In order to improve the accuracy of the optomotor response detection algorithm,a difference-based acceleration algorithm is used to eliminate the curves that are obviously not the optomotor response,and the support vector machine is used to obtain the characteristic parameters to determine the optomotor response.Finally,based on different grating speeds,different grating spatial frequencies and different visual acuity mice,the optomotor response experiments verify the rationality and effectiveness of the proposed optomotor automatic detection algorithm.From the results of analysis of variance,it can be seen that the automatic detection of optomotor response proposed in this paper can be used to distinguish mice with different visions.Using the automatic detection method of optomotor response proposed in this paper can greatly save the time and energy of researchers in ophthalmic experiments,and improve the accuracy of the efficiency of optomotor response detection.Moreover,this method of automatically detecting the eyesight of mice can be used for the vision measurement of multiple mice at the same time,saving a lot of manpower and time costs.