Trichomes are unicellular or multicellular hair-like appendages developed on the aerial plant epidermis of most plant species that act as a protective barrier against natural hazards. For this reason, evaluating the density of trichomes is a valuable approach for elucidating plant defence responses to a continuous challenging environment. However, previous methods for trichome counting, although reliable, require the use of specialised equipment, software or previous manipulation steps of the plant tissue, which poses a complicated hurdle for many laboratories. Here, we propose a new fast, accessible and user-friendly method to quantify trichomes that overcomes all these drawbacks and makes trichome quantification a reachable option for the scientific community. Particularly, this new method is based on the use of machine learning as a reliable tool for quantifying trichomes, following an Ilastik-Fiji tandem approach directly performed on 2D images. Our method shows high reliability and efficacy on trichome quantification in Arabidopsis thaliana by comparing manual and automated results in Arabidopsis accessions with diverse trichome densities. Due to the plasticity that machine learning provides, this method also showed adaptability to other plant species, demonstrating the ability of the method to spread its scope to a greater scientific community.
Keywords: Arabidopsis thaliana; Fiji; Ilastik; Machine learning; Photoshop; Plant trichomes.
Copyright © 2022 The Authors. Published by Elsevier B.V. All rights reserved.