![Author ORCID: We display the ORCID iD icon alongside authors names on our website to acknowledge that the ORCiD has been authenticated when entered by the user. To view the users ORCiD record click the icon. [opens in a new tab]](https://chemrxiv.org/engage/assets/public/chemrxiv/images/logos/orcid.png)
Abstract
Super-resolution reconstruction has attracted increasing research interest in mass spectrometry imaging (MSI), but it remains a challenging ill-posed problem. In the present study, we proposed a Deep learning based SUper-REsolution model called DeepSURE, where the hematoxylin and eosin (H&E) stain microscopy image is used to pose constrains in the process of super-resolution reconstruction to alleviate the ill-poseness. A novel model architecture is designed to achieve multi-task optimization by incorporating multi-modal image registration and fusion in a mutual reinforced framework. The present results demonstrated that the DeepSURE method is able to produce super-resolution reconstruction image with rich chemical information and detailed structure both on visual inspection and quantitative evaluation. In addition, the method was found able to improve the delimitation of boundary between cancerous and para-cancerous regions in MSI image. Furthermore, the reconstruction of low-resolution spatial transcriptomics data demonstrates that the developed DeepSURE method may find wider applications in biomedical fields.
Supplementary materials
Title
Multimodal Image Registration and Fusion Offer Better Spatial Resolution for Mass Spectrometry Imaging
Description
Super-resolution reconstruction has attracted increasing research interest in mass spectrometry imaging (MSI), but it remains a challenging ill-posed problem. In the present study, we proposed a Deep learning based SUper-REsolution model called DeepSURE, where the hematoxylin and eosin (H&E) stain microscopy image is used to pose constrains in the process of super-resolution reconstruction to alleviate the ill-poseness. A novel model architecture is designed to achieve multi-task optimization by incorporating multi-modal image registration and fusion in a mutual reinforced framework. The present results demonstrated that the DeepSURE method is able to produce super-resolution reconstruction image with rich chemical information and detailed structure both on visual inspection and quantitative evaluation. In addition, the method was found able to improve the delimitation of boundary between cancerous and para-cancerous regions in MSI image. Furthermore, the reconstruction of low-resolution spatial transcriptomics data demonstrates that the developed DeepSURE method may find wider applications in biomedical fields.
Actions
