Related Experiment Video
Updated: Jun 9, 2025
05:16
Multiplexed Fluorescent Immunohistochemical Staining of Four Endometrial Immune Cell Types in Recurrent Miscarriage
Published on: August 4, 2021
3.3K
Enhancing endometrial cancer detection: Blood serum intrinsic fluorescence data processing and machine learning
Monika Švecová1, Linda Blahová2, Jozef Kostolný2
1Department of Medical and Clinical Biochemistry, Faculty of Medicine, Pavol Jozef Šafárik University in Košice, Tr. SNP 1, 040 01, Košice, Slovakia.
Talanta
|October 29, 2024
Summary
This study presents a new, cost-effective blood test using fluorescence and machine learning for early endometrial cancer (EC) detection. The method shows high accuracy, potentially improving patient outcomes.
Area of Science:
- Biochemistry
- Medical Diagnostics
- Computational Biology
Background:
- Endometrial cancer (EC) is the most common gynecologic malignancy in developed nations.
- Current early detection of EC is limited due to the absence of a standard laboratory screening test.
- Timely diagnosis is crucial for improving patient prognosis and survival rates.
Purpose of the Study:
- To develop and validate a novel, minimally invasive, and cost-effective method for early endometrial cancer detection.
- To utilize intrinsic fluorescence analysis of blood serum combined with machine learning for enhanced EC diagnosis.
- To identify reliable fluorescent spectral markers for improved diagnostic efficacy.
Main Methods:
- Blood serum samples were analyzed using a luminescence spectrophotometer to capture intrinsic fluorescence spectra.
- Synchronous excitation spectra were visualized using wavelength contour matrices.
- Machine learning algorithms (Random Forest, Logistic Regression, SVM, SGD) and dimensionality reduction techniques (PCA, PLS-DA) were employed for data processing.
Main Results:
- Fluorescence ratios R300/330 and R360/490 were identified as key metabolic biomarkers with high diagnostic efficacy (AUC 0.88 and 0.91).
- Logistic Regression (LR) and Random Forest (RF) algorithms demonstrated high sensitivity, specificity, PPV, and NPV.
- Optimized LR achieved a sensitivity of 0.94, specificity of 0.89, and an AUC of 0.94, indicating significant diagnostic potential.
Conclusions:
- The developed fluorescence-based method combined with machine learning offers a promising approach for early endometrial cancer detection.
- This innovative technique has the potential to significantly enhance laboratory diagnostics and improve clinical outcomes for EC patients.
- The identified fluorescent spectral markers represent valuable biomarkers for non-invasive EC screening.
