Label-free cervical cancer detection in tissue samples using Raman spectroscopy combined with machine learning...

Area of Science:

• Biomedical Optics
• Molecular Spectroscopy
• Oncology

Background:

• Cervical cancer is a leading malignancy in women, typically diagnosed via histopathology using hematoxylin and eosin (H&E) staining.
• H&E staining relies on subjective pathologist interpretation and cannot reveal molecular alterations in tissues.
• There is a need for objective, molecular-based diagnostic methods for cervical cancer.

Purpose of the Study:

• To apply Raman spectroscopy for molecular characterization of cervical cancer tissues.
• To develop a machine learning model for classifying and staging cervical cancer based on spectral data.
• To explore Raman spectroscopy-based heat maps for visualizing tissue differences.

Main Methods:

• Tissue samples from cervical cancer patients were analyzed using Raman spectroscopy.
• Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) were employed for spectral data analysis.
• A classification model was built to differentiate between normal, malignant, and staged cervical cancer tissues.

Main Results:

• Raman spectroscopy identified significant alterations in lipids, proteins, and glycogen in malignant tissues.
• The classification model achieved 100% accuracy in distinguishing malignant from normal cervical tissues.
• The model demonstrated 70.8% accuracy in differentiating early-stage (I/II) from advanced-stage (III/IV) cervical cancer.

Conclusions:

• Raman spectroscopy provides a molecular fingerprint for cervical cancer detection.
• Integrating Raman spectroscopy with machine learning offers a rapid, accurate, and objective diagnostic approach.
• This technique has the potential to enhance cervical cancer diagnosis and staging through molecular profiling.