Premise of the study: X-ray microcomputed tomography (microCT) can be used to measure 3D leaf internal anatomy, providing a holistic view of tissue organisation. Previously, the substantial time needed for segmenting multiple tissues limited this technique to small datasets, restricting its utility for phenotyping experiments and limiting our confidence in the conclusion of these studies due to low replication numbers.
Methods and Results:We present a Python codebase for random-forest machine learning segmentation and 3D leaf anatomical traits quantification which dramatically reduces the time required to process leaf microCT scans. By training the model on 6 hand segmented image slices out of >1500 in the full dataset, it achieves >90% accuracy in background and tissue segmentation, including veins and bundle sheaths grouped together, but not when taken separately.Conclusion: Overall, this 3D segmentation and quantification pipeline can reduce one of the major barriers to using microCT imaging in high-throughput plant phenotyping.
KEY WORDS (3-6): plant leaf anatomy, plant phenotyping, random-forest classification, microCT, image segmentationWe would like to thank Klara Voggeneder for hand labelling the microCT test scan and for improving the hand labelling method, Santiago Trueba for providing the images for Pinus pungens, and Goran Lovric at the Swiss Light Source for assistance during beamtime. GTR was supported by a FWF Austrian Science Fund Lise-Meitner Fellowship (project M2245-B32).
AUTHORS CONTRIBUTIONSJME and MRJ conceptualized and programmed the random forest segmentation program, with assistance from GTR. GTR programmed the leaf traits analysis and the segmentation testing, with assistance from MRJ. GTR acquired and processed the microCT test scan at the SLS. EJF tested and commented on the segmentation program during the development. GTR, MRJ, and JME wrote the paper, with revisions from all authors. AM and CRB provided funding and access to data.
DATA ACCESSIBILITY STATEMENTThe code for analysis suite is available online at github.com/plant-microct-tools/leaftraits-microct, and future updates will be integrated to this repository.