Device and method of using rodent vocalizations for automatic classification of animal behavior
The field of this invention is classifying animal behaviors. In particular the fields of this invention include using animals in vivariums, such as rodents, particularly mice. Ultrasonic vocalizations of mice in a vivarium are compared to a predetermined set of positive phenotypes and to a predetermined set of negative phenotypes. The output of those comparisons are then compared to determine a metric of distance to classify vocalizations as closer to or more distant from elements of the positive or negative sets.
1. A device for automatically classifying animal behaviors comprising:
a study animal type;
a vivarium comprising a set of study animals, of the study animal type, in a plurality of cages;
a non-transitory first memory comprising a first set of positive video and audio behaviors;
a non-transitory second memory comprising a second set of negative video and audio behaviors;
at least one ultrasonic audio sensor proximal to an each cage in the plurality of cages;
at least one video camera proximal to the each cage in the plurality of cages;
wherein at least one ultrasonic sensor and at least one video sensor detect and automatically communicate first communicated behaviors comprising one or more audio and video behaviors, from the at least one cage in the plurality of cages, respectively;
a first processor adapted to identify a first selected set of behaviors derived from the first communicated behaviors, wherein each behavior in the first selected set of behaviors is consistent with one or both of at least one behavior in the first set of positive video and audio behaviors; or at least one behavior in the second set of negative video and audio behaviors;
a second processor adapted to classify the first selected set of behaviors responsive to at least both of: (a) a least difference between the first selected set of behaviors and the first set of positive video and audio behaviors, and (b) a largest difference between the first selected set of behaviors and the second set of negative video and audio behaviors;
wherein the first and second processors are configurable to operate as a same processor;
wherein each of the automatically classified animal behaviors by the device is quantified;
wherein the at least a subset of the automatically classified animal behaviors comprising classified ultrasonic vocalization is then used as accepted dispositive animal behavior.
2. The device of claim 1 , wherein:
the classified ultrasonic vocalizations comprise two or more vocalizations from the Markush group of accent; upward; downward; flat; chevron; U-shaped; and complex.
3. The device of claim 1 , wherein:
the classified ultrasonic vocalizations comprise two or more vocalizations from the Markush group of shape of pitch v. time; intensity; number of times a vocalization occurs within a predetermined time window; change in pitch measured as Hz/time; and simultaneous vocalizations below 25 KHz.
4. The device of claim 1 , wherein:
either the first set of positive video and audio behaviors, or the second set of negative video and audio behaviors, or both, comprise both at least one video behavior and at least one audio behavior.
5. The device of claim 1 , wherein:
the classifying by the second processor comprises:
at least one substitution of a video behavior with an audio behavior or a substitution of an audio behavior with a video behavior, or both, prior to computation of the least difference and prior to the computation of the largest difference.
6. The device of claim 1 , wherein:
the first set of positive video and audio behaviors and the second set of negative video and audio behaviors are generated using a same vivarium environment and a same animal type as the study animal type.
7. The device of claim 1 , wherein:
the classified ultrasonic vocalizations comprise identification of vocalization in one of two predetermined frequency bands, wherein a range of the two predetermined frequency bands are responsive to the study animal type.
8. The device of claim 1 , wherein:
at least a subset of audio behaviors comprise classified non-vocalization sounds.
9. The device of claim 1 , wherein:
the plurality of cages in the vivarium are free of electronic penetrations and are home cages of their respective animals.
10. The device of claim 1 , wherein:
the at least one ultrasonic audio sensor is mechanically independent of the proximal cage such that the at least one ultrasonic audio sensor or the proximal cage is replaceable without mechanically moving the proximal cage or the at least one ultrasonic audio sensor, respectively.
11. The device of claim 1 , wherein:
the device is free of manually observed behavior of the study animals, and free of manually communicated behavior of the study animals, and free of manual classifying of sets of behaviors.
12. The device of claim 1 , wherein:
the detection and automatic communication of one or more audio and video behaviors is continuous.
13. The device of claim 1 , wherein:
the at least one video sensor is mechanically independent of the proximal cage such that the at least one video sensor or the proximal cage are configurable to be replaced without mechanically moving the proximal cage or the at least one video sensor, respectively.
14. A device for automatically classifying animal behaviors comprising:
a study animal type;
a vivarium comprising a set of study animals, of the study animal type, in a plurality of cages;
a non-transitory first memory comprising a first set of positive video and audio behaviors;
a non-transitory second memory comprising a second set of negative video and audio behaviors;
at least one ultrasonic audio sensor proximal to an each cage in the plurality of cages;
at least one video camera proximal to the each cage in the plurality of cages;
wherein at least one ultrasonic sensor and at least one video sensor detect automatically and communicate first communicated behaviors comprising one or more audio and video behaviors, from the at least one cage in the plurality of cages, respectively;
a first processor adapted to identify a first selected set of behaviors derived from the first communicated behaviors, wherein each behavior in the first selected set of behaviors is consistent with one or both at least one behavior in the first set of positive video and audio behaviors; or at least one behavior in the second set of negative video and audio behaviors;
a second processor adapted to classify the first selected set of behaviors responsive to at least both of: (a) a least difference between the first selected set of behaviors and the first set of positive video and audio behaviors, and (b) a largest difference between the first selected set of behaviors and the second set of negative video and audio behaviors;
wherein the first and second processors are configurable to operate as a the same processor;
wherein each of the automatically classified animal behaviors is quantified;
wherein at least a subset of the automatically classified animal behaviors by the device is quantified;
wherein a study using the set of study animals is terminated, responsive to the classified ultrasonic vocalizations.
15. A method for automatically classifying animal behaviors comprising:
a study animal type;
a vivarium comprising a set of study animals, of the study animal type, in a plurality of cages;
a non-transitory first memory comprising a first set of positive video and audio behaviors;
a non-transitory second memory comprising a second set of negative video and audio behaviors;
at least one ultrasonic audio sensor proximal to an each cage in the plurality of cages;
at least one video camera proximal to the each cage in the plurality of cages;
wherein the method comprises the steps:
communicating automatically first communicated behaviors detected by at least one ultrasonic sensor and at least one video sensor wherein the first communicated behaviors comprise one or more audio and video behaviors, from the at least one cage in the plurality of cages;
identifying by a first processor a first selected set of behaviors derived from the first communicated behaviors, wherein each behavior in the first selected set of behaviors is consistent with one or both of at least one behavior in the first set of positive video and audio behaviors; or at least one behavior in the second set of negative video and audio behaviors;
classifying by a second processor the first selected set of behaviors responsive to at least both of: (a) a least difference between the first selected set of behaviors and the first set of positive video and audio behaviors, and (b) a largest difference between the first selected set of behaviors and the second set of negative video and audio behaviors;
wherein the first and second processors are configurable to operate as a same processor;
quantifying by the method each of the automatically classified animal behaviors;
accepting as dispositive animal behavior of at least a subset of the automatically classified animal behaviors comprising classified ultrasonic vocalization.
16. A system for automatically classifying animal behaviors comprising:
a study animal type;
a vivarium comprising a set of study animals, of the study animal type, in a plurality of cages;
a non-transitory first memory comprising a first set of positive video and audio behaviors;
a non-transitory second memory comprising a second set of negative video and audio behaviors;
at least one ultrasonic audio sensor proximal to an each cage in the plurality of cages;
at least one video camera proximal to the each cage in the plurality of cages;
wherein at least one ultrasonic sensor and at least one video sensor detect and automatically communicate first communicated behaviors comprising one or more audio and video behaviors, from the at least one cage in the plurality of cages, respectively;
a first processor adapted to identify a first selected set of behaviors derived from the first communicated behaviors, wherein each behavior in the first selected set of behaviors is consistent with one or both of at least one behavior in the first set of positive video and audio behaviors; or at least one behavior in the second set of negative video and audio behaviors;
a second processor adapted to classify the first selected set of behaviors responsive to at least both of: (a) a least difference between the first selected set of behaviors and the first set of positive video and audio behaviors and (b) a largest difference between the first selected set of behaviors and the second set of negative video and audio behaviors;
wherein the first and second processors are configurable to operate as a same processor;
wherein each of the automatically classified animal behaviors by the system is quantified;
wherein the at least a subset of the automatically classified animal behaviors comprising classified ultrasonic vocalization is then used as accepted dispositive animal behavior.