| Rice is one of the most important food crops both in China and worldwide.Improving the yield and quality of rice is related to the global social and economic development and stability.Therefore,how to cultivate rice varieties with higher yield and good quality has been one of hot spots of rice cultivation and breeding experts.In recent years,rice functional genomics and molecular breeding have developed rapidly,with hundreds of different rice varieties waiting to be screened every day.Rice phenotypic traits with high-throughput will provide powerful technical support for rice variety screening and identification.Rice stem plays a role of nutrient transport and support to the plants,and the selection of rice varieties with lodging resistance will directly related to the improvement of rice yield and quality.Therefore,high-throughput,accurate and non-destructive measurement of rice stem tiller internal structure phenotypic traits is of great significance for rice genetic improvement breeding and cultivation research.At present,traditional rice stem tiller phenotypic traits acquisition mainly rely on manual work.Measuring stem tiller-related phenotypic traits using vernier calipers and protractor measurement tools has disadvantages of destructive,low efficiency and poor repeatability.It is not suitable for high-throughput acquirement and identification of rice varieties,and it has become a bottleneck that restricts the development of rice functional genome and rice improvement breeding.With the development of optical imaging technology,computer science technology and automated integrated control technology,it is possible to measure rice stem tiller phenotypic traits with accurate,non-destructive and high-throughput.Thus,it can provide corresponding phenotypic traits support for rice breeding and cultivation experts.In this paper,a high-throughput non-destructive measurement device based on Micro-CT imaging system and its measurement method are proposed for the reconstruction and application of rice stem tiller internal structure.The main research contents are as follows:(1)Micro-CT system with high-throughput and accurate non-destructive was developed to solve the problem that rice stem tiller traits depended on manual measurement.Based on Lab VIEW software,PLC controller,driver and other control platforms,Micro-CT system realizes the automatic collection of rice samples.In order to ensure the high quality of the acquired images,a step-by-step rotating acquisition mode is proposed.At the same time,the exposure of the flat panel detector is synchronized with the motion of the rotation platform.(2)Before the rice samples were tested,the accuracy of the Micro-CT system was evaluated by using 8 round plastic pipes and 35 rice plants.Then image acquisition,image processing and phenotypic traits were acquired for 234 rice plants.High-throughput non-destructive acquisition of high-resolution(97μm)rice stem tiller transverse section images and 75 rice tiller traits were acquired based on micro-CT system.A total of 739 phenotypic traits were acquired at 9 time points.Based on accelerated algorithms such as GPU and ROI,approximately 278s were required to complete the CT inspection of each plant,and about 310 rice plants can be completed per day.In addition,the self-developed image processing algorithm can also obtain a large number of new traits which can not be quantified artificially,such as tiller area,tiller area growth rate,average tiller angle and other new traits.(3)Based on Micro-CT system,rice tillering and elongation process can be obtained by continuous dynamic measurement of rice plant.Based on TTA and TPA,a logical growth model was established,which can accurately predict and quantify the dynamic growth process of rice stem tiller.A yield prediction model was established based on the phenotypic traits by Micro-CT system.When 10 independent variables were selected,the R~2 value of yield prediction model was 0.48,which could explain 48%of the yield variation of rice.By analyzing the relationship between changes in the tiller angle and yield,it was found that the higher-yield accessions had a larger average tiller angle in the early tillering stage and a smaller average tiller angle in the late tillering stage.In addition,the implications of tillering senescence for yield and drought resistance were also analyzed.The lower tillering senescence accessions had a higher yield and a better drought tolerance(a higher GPAR and leaf moisture content,and a lower TBR).The higher tillering senescence accessions had a lower yield and a poor drought tolerance.A total of 402 loci that significantly associated with rice tiller and yield were detected by genome-wide association analysis.One locus on chromosome 9 was detected at all 9 time points,including a known gene locus TAC1 associated with tiller angle.In addition,2new gene loci related to rice tiller and yield were found,which will be beneficial in breeding for high yield.(4)Based on Micro-CT system,104 rice stems were acquired and reconstructed transverse section images.Then the corresponding models were trained by deep learning algorithm and the rice stems were segmented.Finally,the three-dimensional images were obtained and 24 rice stem traits were acquired.The Micro-CT system measurement values of rice stem tiller were compared with the manual measurement.(5)24 phenotypic traits of rice stem obtained by Micro-CT system were used to establish a bending stress prediction model.When 2 independent variables were selected,the R~2 value of bending stress prediction model was 0.722.Based on the two independent variables selected by the prediction model,three methods,including SDA,SVM and RF,were used to classify the high and low bending stresses,the final leave-one-out cross validation classification accuracy were 87.5%,83.7%and 80.8%respectively.We also analyzed the relationship between the bending stress and the 24 rice stem traits.There was a higher positive correlation between the bending stress and 8 morphological traits of rice stem,while the tiller number had a higher negative correlation with the bending stress and 8 morphological traits of rice stem.By comparing the accuracy of CT image measurement and RGB side view image measurement for rice stem diameter,the results show that the accuracy of rice stem diameter measurement based on CT image is higher than that based on RGB side view image.The implications of bending stress for dry weight,stem density and drought-related traits were analyzed.Rice plants with larger bending stress had a larger dry weight,a larger stem density and a larger stem area,but a poor drought resistance.The Micro-CT system can obtain the stem area under each height quantitatively and non-destructive,and more accurate quantify the change trend of stem area in each internode.(6)Research on reconstruction of wheat stem internal structure.Micro-CT system was used for image acquisition of wheat stem,after image processing,4 wheat tiller-related traits were acquired,including the tiller number,tiller angle,tiller wall thickness and tiller diameter,and then compared with the manual measurements.Finally,comparing measurement method for wheat stem tiller by Micro-CT system with the three existing methods,it is concluded that the Micro-CT system has the advantages of high-throughput non-destructive measurement and comprehensive measurement of the phenotypic traits of the wheat tiller,which provides a new technical support for wheat breeding and cultivation. |
PDF Full Text Request