Related Experiment Video
Updated: Sep 21, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Automated Cancer Diagnostics via Analysis of Optical and Chemical Images by Deep and Shallow Learning
Olof Gerdur Isberg1,2,3, Valentina Giunchiglia1, James S McKenzie1
1Department of Metabolism, Digestion and Reproduction, Faculty of Medicine, Imperial College London, London SW7 2AZ, UK.
Metabolites
|May 28, 2022
Summary
Digital pathology with deep learning and mass spectrometry imaging (MSI) with shallow learning both achieve over 90% F1-scores for automated breast cancer diagnostics. These advanced techniques offer efficient alternatives to traditional manual analysis of tissue samples.
Area of Science:
- Oncology
- Computational Pathology
- Biomedical Imaging
Background:
- Traditional optical microscopy for cancer diagnostics is labor-intensive and time-consuming.
- Increasing complexity in molecular diagnostics necessitates advanced analytical approaches.
- Digital pathology and mass spectrometry imaging (MSI) offer potential improvements to current workflows.
Purpose of the Study:
- To compare the diagnostic performance of digital pathology using deep learning against MSI with shallow learning.
- To evaluate automated tissue recognition and annotation in breast cancer samples.
- To assess the utility of biochemical information from MSI for further analysis.
Main Methods:
- Utilized formalin-fixed and paraffin-embedded (FFPE) breast cancer tissue microarrays (TMAs).
- Applied deep learning algorithms for tissue recognition on conventional optical images.
- Employed shallow learning techniques for annotation using mass spectrometry imaging (MSI) data.
Main Results:
- Both deep learning on optical images and shallow learning on MSI achieved automated diagnostics with F1-scores exceeding 90%.
- MSI provides intrinsic biochemical information valuable for subsequent analyses.
- Automated diagnostic capabilities were demonstrated for both approaches.
Conclusions:
- Digital pathology and MSI are effective automated diagnostic tools for breast cancer.
- Both deep learning and shallow learning approaches demonstrate high diagnostic accuracy.
- MSI offers additional benefits through inherent biochemical data for deeper insights.
