Imperial College London

DrAndreBrown

Faculty of MedicineInstitute of Clinical Sciences

Reader in Behavioural Phenomics
 
 
 
//

Contact

 

+44 (0)20 3313 8218andre.brown

 
 
//

Location

 

4.15BLMS BuildingHammersmith Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Koren:2015:10.1371/journal.pone.0122326,
author = {Koren, Y and Sznitman, R and Arratia, PE and Carls, C and Krajacic, P and Brown, AEX and Sznitman, J},
doi = {10.1371/journal.pone.0122326},
journal = {PLOS One},
title = {Model-independent phenotyping of C. elegans locomotion using scale-invariant feature transform},
url = {http://dx.doi.org/10.1371/journal.pone.0122326},
volume = {10},
year = {2015}
}

RIS format (EndNote, RefMan)

TY  - JOUR
AB - To uncover the genetic basis of behavioral traits in the model organism C. elegans, a common strategy is to study locomotion defects in mutants. Despite efforts to introduce (semi-)automated phenotyping strategies, current methods overwhelmingly depend on worm-specific features that must be hand-crafted and as such are not generalizable for phenotyping motility in other animal models. Hence, there is an ongoing need for robust algorithms that can automatically analyze and classify motility phenotypes quantitatively. To this end, we have developed a fully-automated approach to characterize C. elegans’ phenotypes that does not require the definition of nematode-specific features. Rather, we make use of the popular computer vision Scale-Invariant Feature Transform (SIFT) from which we construct histograms of commonly-observed SIFT features to represent nematode motility. We first evaluated our method on a synthetic dataset simulating a range of nematode crawling gaits. Next, we evaluated our algorithm on two distinct datasets of crawling C. elegans with mutants affecting neuromuscular structure and function. Not only is our algorithm able to detect differences between strains, results capture similarities in locomotory phenotypes that lead to clustering that is consistent with expectations based on genetic relationships. Our proposed approach generalizes directly and should be applicable to other animal models. Such applicability holds promise for computational ethology as more groups collect high-resolution image data of animal behavior.
AU - Koren,Y
AU - Sznitman,R
AU - Arratia,PE
AU - Carls,C
AU - Krajacic,P
AU - Brown,AEX
AU - Sznitman,J
DO - 10.1371/journal.pone.0122326
PY - 2015///
SN - 1932-6203
TI - Model-independent phenotyping of C. elegans locomotion using scale-invariant feature transform
T2 - PLOS One
UR - http://dx.doi.org/10.1371/journal.pone.0122326
UR - http://hdl.handle.net/10044/1/32429
VL - 10
ER -