High-throughput three-dimensional visualization of root system architecture of rice using X-ray computed tomography [post]

2019 unpublished
Plants adjust their root system architecture (RSA) against changing environments to optimize their growth. Nondestructive phenotyping of roots beneath the soil not only reveal the response of RSA against environmental stimuli but also allow for designing an ideal RSA for crop cultivation. Generally, roots beneath the soil surface are three-dimensionally visualized using X-ray computed tomography (CT). However, root isolation from X-ray CT images involves a longer time; in addition, CT scanning
more » ... ition, CT scanning and reconstruction processes require longer periods. For large-scale root phenotyping, a shorter image acquisition time is required. Thus, the objective of this study is to develop a high-throughput pipeline to visualize rice RSA consisting of radicle and crown roots in the soil, from X-ray CT images. Results: We performed the following three processes to develop the pipeline. First, we used calcined clay with uniform soil particle size as the soil substrate. The size of voids between the soil particles was less than the scanning resolution, resulting in a clear root shape in the CT images. Second, we optimized the parameters for rapid X-ray CT scanning. Higher tube voltage and current produced the highest root-to-soil contrast images. Third, we used a 3-D median filter to reduce noise, and an edge detection alogism to isolate the root segments. The detection limits of the root diameters of the pots of diameters 16 cm and 20 cm were 0.2 mm and 0.3 mm, respectively. Because the crown root diameter of rice is generally higher than 0.2 mm, almost all crown roots could be visualized. Our condition allows for simultaneously performing CT scanning and reconstruction by a high-performance computing technology. Consequently, our pipeline visualizes rice RSA in the soil, requiring less than 10 min (33 s, if a rough image is acceptable) for CT scanning and reconstruction, and 2 min for image processing to visualize rice RSA. We scanned the roots of the upland rice (considered in this study) daily, and our pipeline successfully visualized the root development dynamics over three weeks. Conclusions: We developed a rapid three-dimensional visualization method to visualize rice RSA in the soil using X-ray CT and a fully automated-image processing method known as RSAvis3D. Our methodology allows for high-throughput measuring and requires no manual operators in image processing, thereby providing a potentially efficient large-scale root phenotyping.
doi:10.21203/rs.2.18397/v1 fatcat:einvvsfgjfgnddoenfm4cvyq2y