A time-domain wavelet-based approach for fluorescence diffuse optical tomography.

Area of Science:

• Biomedical Optics
• Medical Imaging
• Optical Physics

Background:

• Time-resolved fluorescence diffuse optical tomography (FDOT) is an imaging technique that utilizes light propagation and detection over time.
• Extracting meaningful information from time-resolved signals is crucial for accurate FDOT reconstructions.
• Existing methods often rely on temporal moments, but their effectiveness can be limited.

Purpose of the Study:

• To introduce and evaluate a novel approach for feature extraction in FDOT using the wavelet transform.
• To compare the performance of wavelet-based feature extraction against traditional temporal moments.
• To investigate the impact of noise and photon count on reconstruction quality.

Main Methods:

• A comparative reconstruction study was performed using synthetic data from an inhomogeneous cubic medium.
• Measurements were simulated with Poisson noise statistics.
• The proposed wavelet transform approach was compared to the reference temporal moments method.
• A regularization parameter selection procedure was employed for fair comparison.

Main Results:

• Reconstruction quality is primarily determined by the number of features in noise-free conditions.
• In noisy scenarios, the wavelet approach significantly outperforms the moment approach.
• Optimal reconstruction quality, achievable with all time samples, was recovered using only eight wavelet functions, a feat not possible with moments.
• The utility of time-resolved information diminishes when the maximum detected photon count falls below 10^5.

Conclusions:

• The wavelet transform is a more effective method for exploiting time-resolved information in FDOT.
• This advantage is particularly pronounced when dealing with low photon counts.
• However, below a critical photon count threshold, the inherent utility of time-resolved data for reconstruction becomes limited.