Efficient imaging and computer vision detection of two cell shapes in young cotton fibers

Authors: GRAHAM, BENJAMIN - North Carolina State University; PARK, JEREMY - North Carolina State University; BILLINGS, GRANT - North Carolina State University; Hulse-Kemp, Amanda; HAIGLER, CANDACE - North Carolina State University; LOBATON, EDGAR - North Carolina State University

Submitted to: Applications in Plant Sciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/18/2022
Publication Date: 11/26/2022
Citation: Graham, B.P., Park, J., Billings, G.T., Hulse-Kemp, A.M., Haigler, C.H., Lobaton, E. 2022. Efficient imaging and computer vision detection of two cells in young cotton fibers. Applications in Plant Sciences. https://doi.org/10.1002/aps3.11503.
DOI: https://doi.org/10.1002/aps3.11503

Interpretive Summary: Our most abundant renewable textile material, cotton, is made up of numerous individual cotton fibers. It has been recently shown that developing cotton fibers have distinct tip shapes at a very young age. These single cells each exhibit one of two fiber tip shapes, either "tapered" (narrow) or "hemisphere" (wide). Cotton lines have different proportions of each of the two fiber tip shapes. Microscopically imaging and analyzing the shape of very young cotton fibers has historically been very time-consuming, preventing scientists from studying these features in large numbers of cotton lines. In this project, an interdisciplinary group of molecular and computational biologists together with engineers developed a semi-automated microscope imaging method and a machine learning model capable of detecting and classifying thousands of fibers per day. The computer is given images of young cotton fibers and then the software model identifies and quantifies the proportion of the two fiber tip shapes in each image. This advancement has allowed us to speed up fiber shape analysis over 8-fold compared to manual methods, opening up new possibilities for future research in this area, including testing the relationship of early fiber shape variations to mature fiber quality.

Technical Abstract: Introduction: The shape of young cotton fibers varies within and between cotton (Gossypium) species as shown so far by detailed analysis of two cultivars, one from G. hirsutum and one from G. barbadense. Both narrow and wide fibers exist in G. hirsutum cv. Deltapine 90, which may impact the quality of our most abundant renewable textile material. More efficient cellular phenotyping methods were needed to empower future research. Methods: We developed semi-automated methods to image young cotton fibers and a novel machine learning algorithm for rapid detection of tapered (narrow) or hemisphere (wide) cotton fibers in homogeneous or mixed populations. Results: The new methods were accurate for diverse accessions of G. hirsutum and G. barbadense and eight times more efficient than manual methods. Narrow fibers dominated in three accessions of G. barbadense, whereas three G. hirsutum accessions showed a bimodal distribution of tapered and hemisphere fibers in varying proportions. Discussion: The use or adaptation of these improved methods will facilitate higher throughput experiments to understand the controls of variable shapes of young cotton fibers or other elongating single cells. This research also enables exploring links between early cell shape and mature cotton fiber quality in diverse field-grown cotton accessions.