JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

FTIR-based machine learning for prediction of malignant transformation in oral epithelial dysplasia

  • 0Department of Oral and Craniofacial Sciences, School of Dentistry, University of Missouri Kansas City, Kansas City, MO, USA. wangrong@umkc.edu.
The Analyst +

|

May 23, 2025

|

May 23, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Fourier-transform infrared (FTIR) spectroscopy predicts malignant transformation risk in oral epithelial dysplasia (OED). This advanced technique offers a more objective assessment than traditional methods for identifying high-risk precancerous oral lesions.

Area Of Science

  • Biomedical Spectroscopy
  • Cancer Research
  • Machine Learning in Pathology

Background

  • Oral squamous cell carcinoma (OSCC) is aggressive with poor prognosis.
  • Oral epithelial dysplasia (OED) is a precancerous lesion with malignant transformation (MT) risk.
  • Current histopathology for OED is subjective and lacks MT risk prediction accuracy.

Purpose Of The Study

  • To evaluate an FTIR-based OSCC-Benign classifier for predicting MT risk in OED.
  • To assess the potential of FTIR biomolecular fingerprinting as an objective diagnostic tool for OED.

Main Methods

  • Thirty OED patient biopsies with known MT outcomes were analyzed.
  • FTIR imaging was used to acquire biochemical profiles (biomolecular fingerprints) from tissue sections.
  • An established FTIR-based OSCC-Benign classifier was applied to predict MT risk in OED samples.

Main Results

  • The FTIR classifier achieved 81.7% accuracy (F1 score 0.77) at the ROI level and 83.3% accuracy (F1 score 0.8) at the biopsy level in predicting MT in OED.
  • OEDs with FTIR fingerprints similar to OSCC indicated a higher MT risk.
  • OEDs with fingerprints resembling benign tissue showed a lower MT risk.

Conclusions

  • FTIR biomolecular fingerprinting can effectively predict malignant transformation risk in OED.
  • The FTIR-based machine learning approach surpasses traditional histopathology in assessing OED MT risk.
  • This method offers a quantitative, objective tool for improved clinical diagnosis and management of OED.