Find us on Facebook   WebLab News RSS-Feed
Part-financed by the European Union
(European Regional Development Fund)
South Baltic WebLab - Header Grafic

South Baltic WebLab

Learning from Dolphins - Blog

Dolphin voices but also lame horses

15.12.12, 17:53 (comments: 0)

I had to go to the veterinarian today with my dog. She has started to limp on one or possibly both of her front legs. Even the vet couldn't say. This problem reminded me of my Master's Thesis project. Determining the degree of lameness  in horses is also a very difficult thing. The method applied today on most clinics just relies completely on the veterinarians' abilities and experience to see how much pain the animal is experiencing during walk and trot. Normally, nothing is measured. The vet just uses her/his eyes to do this. Obviously, this opens up for misinterpretations and in the worse case scenario, maltreatment if the animal is diagnosed with the wrong thing or if an injury is completely missed. For my Master's Thesis project I wanted to make the method better and more objective by simply measuring some parameters of the movement of the horses. I managed to build a measurement system that worked. However, taking it further to a product or testing it on clinics proved to be much more difficult. I never got that far. Perhaps, if my research funding for doing dolphin research completely ends, I might pick up that loose end with the lameness measurement system. Clearly, it should be very useful also for dogs, as I experienced today.


Horses and dogs aside. It is dolphins we are talking about. If you understand Swedish and want to watch me talk about my research you can visit the following website and look at the video clip there. It is 20 minutes long.


Perhaps you are wondering how it is possible to study individual dolphin behaviour while there are multiple individuals swimming around in the same group. That was something that I also questioned myself when I was starting to analyze the data I had collected. It is tricky. The echolocation signals , or "clicks" from each animals comes in long series of clicks, one after the other, "called click trains". If you have multiple individuals that are echolocation concurrently, you will end up having overlapping click trains from different individuals in your data. In order to analyze the echolocation clicks on an individual level you must figure out how to separate these click trains in the data. That is something that I have developed an algorithm for. The word "algorithm" sounds fancy, but it is nothing more than just step-by-step instructions to how to perform the separation procedure for the overlapping click trains. Like a recipe written in mathematics that describes the whole thing. The computer program basically works as a voice recognition program and compares all the clicks with each other. When it finds a click that matches another click they are assumed to originate from one and the same animal and they are grouped together.


Tomorrow is the last day of my blog. Maybe I'll show you some photos from another field trip where I recorded the echolocation sound of a beluga whale. Make sure you don't miss it ;-)


Go back

Add a comment

Comments are being moderated. It might take up to a day before a comment is authorized.