|RAJA, PRANAV - University Of California, Davis
|EARLES, J - University Of California, Davis
|BUCHKO, ALEXANDER - California Polytechnic State University
|MOMAYYEZI, MINA - University Of California, Davis
|DUONG, FIONA - San Francisco State University
|PARKINSON, DILWORTH - Lawrence Berkeley National Laboratory
|FORRESTEL, ELIZABETH - University Of California, Davis
|SHACKEL, KENNETH - University Of California, Davis
Submitted to: ArXiv
Publication Type: Pre-print Publication
Publication Acceptance Date: 3/18/2022
Publication Date: 3/18/2022
Citation: Rippner, D.A., Raja, P., Earles, J., Buchko, A., Momayyezi, M., Duong, F., Parkinson, D., Gupta, L., Forrestel, E., McElrone, A.J. 2022. A workflow for segmenting soil and plant X-ray CT images with deep learning in Google’s Colaboratory. ArXiv. https://doi.org/10.48550/arXiv.2203.09674.
Interpretive Summary: X-ray microcomputed tomography (microCT) based imaging is widely used to study soils and plants. Despite the widespread use, processing the data generated by microCT imaging is challenging, often taking hours to days to complete. Recent advances in computing can help researchers speed up the process of analyzing microCT image data. We have developed a work flow using Google's Colaboratory based computing resources to analyze microCT image data using artificial neural networks, reducing analysis time to minutes. To demonstrate the power of the workflow, we show examples of analysis done on microCT image data collected from walnut leaves, an almond bud, and a soil aggregate.
Technical Abstract: X-ray micro-computed tomography has enabled the characterization of the properties and processes that take place in plants and soils at the micron scale. Despite the widespread use of this advanced technique, major limitations in both hardware and software limit the speed and accuracy of image processing and data analysis. Recent advances in machine learning, specifically the application of convolutional neural networks to image analysis have enabled rapid and accurate segmentation of image data. Challenges remain to applying convolutional neural networks to the analysis of environmentally and agriculturally relevant images. Specifically, there is a disconnect between the computer scientists and engineers, who build these AI/ML tools, and the potential end users in agricultural research, who may be unsure of how to apply these tools in their work. Additionally, the computing resources required for training and applying deep learning models are unique, more common to computer gaming systems or graphics design work, rather than traditional computational systems. To navigate these challenges, we developed a modular workflow for applying neural networks to X-ray µCT images, using low-cost resources in Google’s Colaboratory. Here we present the results of the workflow, illustrating how parameters can be optimized to achieve best results using example scans from walnut leaves, almond flower buds, and a soil aggregate.