Machine Learning Aided Photonic Diagnostic System for Minimally Invasive Optically Guided Surgery in the...

Area of Science:

• Oncology
• Medical Technology
• Biomedical Optics

Background:

• Hepatoduodenal area tumors have increasing mortality despite declining cancer death rates.
• Minimal access surgery (MAS) can improve patient survival and quality of life.
• Accurate, real-time tumor detection during surgery is crucial for effective treatment.

Purpose of the Study:

• To develop and characterize a tissue endogenous fluorescence (TEF) system for MAS.
• To assess machine learning (ML) for real-time hepatoduodenal tumor diagnosis during MAS.
• To evaluate blood perfusion oscillations as a supplementary diagnostic marker.

Main Methods:

• Utilized a fiber-optic probe to record TEF and blood perfusion parameters (laser Doppler flowmetry) during MAS.
• Employed ML algorithms (k-Nearest Neighbors, AdaBoost) for tissue classification.
• Analyzed cardiac and respiratory oscillations in blood perfusion for tissue vitality assessment.

Main Results:

• TEF combined with ML algorithms demonstrated high promise for differentiating cancerous from healthy tissue in situ.
• Blood perfusion oscillation amplitudes (cardiac, respiratory) were significantly higher in intact tissues than cancerous ones.
• Laser Doppler flowmetry served as a tissue vitality sensor, reducing TEF data variability.

Conclusions:

• A fiber-optic TEF probe with ML offers a promising approach for real-time, in situ differentiation of cancerous and healthy hepatoduodenal tissues.
• Parallel analysis of blood perfusion oscillations enhances diagnostic accuracy.
• This technology can improve surgical guidance and patient outcomes in hepatoduodenal cancer treatment.