Sagittal Plane Registration Of Mouse Brain Mass Spectrum Image

Mass spectrometry imaging(MSI)can detect the species and spatial distribution of molecules in the sample at the same time.Meanwhile,it can simultaneously image hundreds of biomolecules without the need for markers,such as proteins,peptides,lipids,metabolites.Recently,MSI technology has played a huge role in clinical diagnosis and drug research,such as the study of biomarker molecules of disease,detection of the distribution of a variety of metabolites in specific tissues,determination of tumor tissue margins.However,after the MSI experiment,the MSI data needs to be reconstructed.Histological staining of adjacent tissue sections is needed to confirm the accuracy of spatial information,to promote the fusion of MSI data with histological structure.Nevertheless,accurate anatomical annotation of stained sections often requires biological experts,and manual histological subdivision of stained sections is a potential risk of introducing human bias and it is time-consuming.Registered the MSI data to the reference brain slices in the standard spectrogram,the researchers can achieve automatic anatomical interpretation of the MSI data,and facilitate subsequent automated data processing.Because the coronal plane of the brain is roughly symmetrical from left to right,and contains a small number of brain regions,each brain region has a large area,which is easy to automatic registration.Nevertheless,a large number of the registration works were focused on the coronal slices of mice.However,the sagittal slices can provide more biological information about the relationship between MSI data and the structure of brain regions in a single histological section.Which is more useful for studying biomolecular changes in specific anatomical structures caused by disease.Nevertheless compared with the coronal plane,the sagittal plane sections of mice contain more brain region structures,and the area of each brain region is relatively small,and the brain regions of different mice were more severely deviated,so registration is more difficult.Currently,there is a lack of automatic registration methods for the sagittal plane of the brain.In order to study the automatic registration of MSI data in the sagittal plane,it is necessary to solve the problems of preprocessing MSI data and improving the accuracy of registration.Based on these problems,this paper has done some research on the registration of MSI data in the sagittal plane,mainly involving the following two aspects.Firstly,in the aspect of MSI data pretreatment,we preprocessed the original MSI data by combining the Matlab programming language and the deep learning method.To more accurately look for biomarker molecules associated with disease or specific stimuli,which requires us to screen all the available data collected individually.We found that there were a lot of noise signals in the original MSI data,and manual screening is time-consuming and laborious,and cannot guarantee accuracy.Therefore,based on the characteristics of these noise signals,we used the matlab programming on the noise signal for a preliminary screeningthe.However,some noise signals that are difficult to be screened out by simple programming statements,and deep learning method is adopted for further noise reduction.Through the above two steps,we realize fast noise reduction of MSI data,thus laying a foundation for the subsequent screening of biomarkers and registration.Then,we introduced a strategy based on auxiliary line is proposed to register MSI data in sagittal plane.The auxiliary line was added at the brain regions of CC and CP as the landmarks for registration,MSI data before and after registration in multiple brain regions were analyzed.We found that the addition of auxiliary lines not only improved the registration accuracy of two brain regions,CC and CP,but also improved the registration accuracy of several adjacent brain regions and even the whole sagittal plane brain tissue.Meanwhile,we studied the registration of multiple sagittal slices with different staining conditions and deformation degrees.The matching effect of the brain image after registration and the geometry of the reference brain image in the standard spectral image is improved significantly.Through further analysis of the MSI data of these brain slices with and without auxiliary line after registration,it can be found that the accuracy of the MSI data of all brain slices with auxiliary line registration has been significantly improved,which proves that this registration method has good universality.