Related Experiment Video
Updated: Jul 24, 2025
08:14
Sample Preparation Strategies for Mass Spectrometry Imaging of 3D Cell Culture Models
Published on: December 5, 2014
18.3K
DeepS: Accelerating 3D Mass Spectrometry Imaging via a Deep Neural Network
Dan Li1, Yao Qian1, Haiming Yao1
1Department of Precision Instrument, Tsinghua University, Beijing 100084, China.
Analytical Chemistry
|July 10, 2023
Summary
DeepS accelerates three-dimensional mass spectrometry imaging (3D MSI) using a novel sparse sampling network. This method reconstructs 3D MS images from sparse data, achieving high accuracy and significantly reducing analysis time for complex biological samples.
Area of Science:
- Biomedical imaging
- Analytical chemistry
- Computational biology
Background:
- Mass spectrometry imaging (MSI) visualizes biomolecule distribution in tissues.
- Three-dimensional (3D) MSI offers enhanced spatial mapping of complex biological structures.
- Traditional 3D MSI is time-consuming due to serial 2D section analysis.
Purpose of the Study:
- To develop a significantly faster 3D MSI workflow.
- To enable high-resolution 3D biomolecule mapping with reduced data acquisition.
- To validate the workflow on diverse biological models.
Main Methods:
- A 3D sparse sampling network (3D-SSNet) was developed for image reconstruction.
- A sparse sampling strategy was employed to reduce data acquisition.
- Transfer learning was utilized to adapt the model for heterogeneous samples.
Main Results:
- The DeepS workflow achieved results comparable to full sampling MSI at 20-30% sampling ratios.
- Successful 3D MSI analysis was demonstrated on mouse brains with Alzheimer's disease and glioblastoma.
- The method was also applied to 3D imaging of a mouse kidney.
Conclusions:
- DeepS substantially accelerates 3D MSI analysis through sparse sampling and deep learning.
- The workflow provides accurate biomolecular mapping in complex biological tissues.
- This approach enhances the utility of 3D MSI for disease and organ studies.
