[Preliminary study on internal information of the measured tissue based on distributed multi-position scattering...

Area of Science:

• Biomedical Optics
• Optical Phantoms
• Tissue Optics

Background:

• Accurate measurement of internal tissue properties is crucial for medical diagnostics.
• Superficial tissue layers can interfere with optical measurements of deeper tissues.
• Developing non-invasive methods to assess internal tissue characteristics is a significant challenge.

Purpose of the Study:

• To design and validate a phantom experiment for predicting internal tissue optical properties.
• To assess the effectiveness of scattering spectral analysis in overcoming superficial interference.
• To evaluate the impact of multi-position spectral data on prediction accuracy.

Main Methods:

• A double-layer phantom was created with milk solutions of varying concentrations (internal tissue) and pellicles (superficial tissue).
• 200 scattering spectral data points were collected from two positions.
• A backpropagation (BP) neural network was employed to predict milk concentration using spectral data.

Main Results:

• Both single-position and two-position scattering spectra achieved over 90% training fitting and prediction accuracy.
• Two-position scattering spectra demonstrated higher prediction accuracy, reaching 98.41%.
• The method effectively predicted milk concentration and mitigated superficial tissue interference.

Conclusions:

• Scattering spectra analysis, considering photon dissemination paths, can accurately predict internal tissue concentration.
• Incorporating multi-position spectral data significantly improves prediction accuracy.
• This non-invasive technique shows feasibility for assessing internal tissue information without compromising tissue integrity.