Early Detection And Root 3D Architecture Of Oilseed Rape Clubroot Based On Low Field Nuclear Magnetic Resonance

The root system of a crop is an important organ that provides nutrients and water to the plant.If the root system is infected with diseases,it can cause plant death and crop yield reduction.Since root disease development is in the subsurface soil environment,it is more difficult to detect in a timely manner than crop diseases that infest above-ground plants.In this thesis,we used magnetic resonance T2 relaxation spectrum and imaging techniques for the early nondestructive detection of rapeseed clubroot using oilseed rape as the target.The main research contents and findings are summarized as follows:(1)The effects of soil moisture content,soil volume and data collection height on the T2 relaxation spectrum of rapeseed root system were investigated.A discriminant analysis model based on the root T2 relaxation spectra was established to enable the detection of oilseed rape clubroot.Soil water content had a significant effect on the magnetic resonance T2 relaxation spectra of the root samples,while soil volume and measurement height,did not have a significant effect on the root T2 relaxation spectrum.The T2 relaxation time of diseased rape roots was greater than the relaxation time of healthy roots,and the peak area of diseased roots was greater than the peak area of healthy roots.Based on the root relaxation spectrum data,both the gradient boosting tree(GBDT)and random forest model(RF)could achieve the accuracy of 87.5% in discriminating oilseed rape clubroot.The RF model after further feature fusion of root relaxation spectrum and root fresh weight could improve the accuracy to 93% and above in both the validation and test sets.(2)The effects of different soil types,soil moisture content and imaging setup parameters on MRI of rape roots were investigated.The soil moisture content favorable for MRI of roots,as well as the optimal imaging mode and imaging parameters for MRI of rape roots were determined.The recognizability of the root system in the MRI images decreases significantly when the soil water content exceeds 198%.Considering that the root system needs a certain amount of water to meet the growth and development needs,the soil water content to achieve root MRI should not exceed 130%.The effects of five key imaging parameters on root imaging were systematically investigated using SNR,CNR and imaging time,and the optimal imaging mode and combination of imaging parameters(proton density imaging: long TR=3000ms,short TE=18.14ms)for MRI of rape root were determined.Using the optimized imaging parameters,the minimum diameter of the rape root system that could be detected was 1 mm.(3)The root system histogram features of magnetic resonance sequence images were used to establish a discriminative model of rape clubroot disease based on internal root tissue information.Combined with computer graphics techniques,3D visualization model of root system was achieved and interactive measurement techniques were used to obtain three-dimensional root system architecture parameters.Moreover,quantitative analysis of the change pattern of three-dimensional root system conformation parameters was performed.The established discriminative model of rape clubroot based on external architecture parameters can achieve early and accurate detection of rape clubroot disease.The histogram features of rape root system extracted from magnetic resonance images were used to establish a RF model of rape clubroot based on internal root tissue information,and more than 90% accuracy was achieved in both the validation and test sets.The 3D visualization model of rapeseed root system was established and the three-dimensional root system architecture parameters were measured in situ using interactive measurement techniques to quantitatively analyze the change patterns of three-dimensional root system conformation parameters of rapeseed under clubroot disease stress,and.The.The results showed that the parameters of root system volume,maximum diameter of root system,total number of root branches and maximum depth of root system can effectively reflect the condition of root system attacked by clubroot disease,and can be used for early detection and early warning of rape clubroot disease.The RF model of rape clubroot based on external morphological architecture established based on the above-mentioned architecture parameters reached more than95% in both the validation and test sets.