Related Experiment Video
Updated: May 21, 2026
09:53
Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography
Published on: August 16, 2020
Support Vector Machine on fluorescence landscapes for breast cancer diagnostics
Tatjana Dramićanin1, Lea Lenhardt, Ivana Zeković
1Vinča Institute of Nuclear Sciences, University of Belgrade, P.O. Box 522, 11001 Belgrade, Serbia.
Journal of Fluorescence
|June 9, 2012
Summary
Support Vector Machine (SVM) analysis of breast tissue fluorescence spectra shows synchronous fluorescence spectra (SFS) data achieve 100% accuracy for breast cancer diagnosis, outperforming excitation-emission matrices (EEM). Minimal data input combinations for SFS are identified.
Area of Science:
- Biomedical Engineering
- Medical Diagnostics
- Spectroscopy
Background:
- Breast cancer diagnosis relies on accurate and sensitive methods.
- Current diagnostic tools can be invasive or lack optimal specificity.
- Fluorescence spectroscopy offers a potential non-invasive approach for tissue analysis.
Purpose of the Study:
- To develop and optimize a Support Vector Machine (SVM) based diagnostic tool for breast cancer detection.
- To evaluate the efficacy of excitation-emission matrices (EEM) and total synchronous fluorescence spectra (SFS) as input data for SVM.
- To identify optimal data input combinations for high sensitivity and specificity in breast cancer diagnostics.
Main Methods:
- Collected UV-VIS fluorescence spectra (EEM and SFS) from normal and malignant breast tissue specimens.
- Utilized Support Vector Machine (SVM) algorithms for classification of tissue types.
- Tested various combinations of spectral data inputs to assess classification accuracy against histopathology.
- Determined diagnostic sensitivity and specificity for different data input strategies.
Main Results:
- SVM analysis using excitation-emission matrices (EEM) yielded 67% sensitivity and 62% specificity.
- SVM analysis using total synchronous fluorescence spectra (SFS) achieved 100% sensitivity and specificity for certain input combinations.
- Identified specific SFS data combinations that provide maximal diagnostic performance with minimal data requirements.
Conclusions:
- Total synchronous fluorescence spectra (SFS) demonstrate superior performance over EEM for SVM-based breast cancer diagnostics.
- SVM models utilizing SFS data can achieve highly accurate and specific breast cancer detection.
- Optimized SFS data input strategies offer a promising, efficient approach for developing advanced breast cancer diagnostic tools.
