🔬 applied physics

A Novel Differential Pathlength Factor Model for Near-Infrared Diffuse Optical Imaging

This paper proposes and validates two novel, distance- and property-dependent Differential Pathlength Factor (DPF) models derived from Monte Carlo simulations and phantom experiments, which significantly reduce quantitative errors in continuous-wave near-infrared diffuse optical imaging compared to conventional formulations, particularly at small source-detector separations.

Original authors: Kaiser Niknam, Mannu Bardhan Paul, Mini Das

Published 2026-02-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Kaiser Niknam, Mannu Bardhan Paul, Mini Das

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Measuring What's Inside Without Cutting Open

Imagine you want to know how much "red" (oxygen) and "blue" (deoxygenated blood) is inside a person's muscle or brain. You can't cut them open, so you shine a flashlight (near-infrared light) through their skin. The light bounces around inside the tissue like a pinball in a machine before coming out the other side.

The problem is that the light doesn't travel in a straight line. It zig-zags wildly. If you just measure how much light comes out, you can't tell if the tissue is dark because it's thick, or because the light took a very long, winding path. To fix this, scientists use a "correction number" called the Differential Pathlength Factor (DPF). Think of the DPF as a magnifying glass for distance: it tells you, "Even though the light only traveled 2 inches across your skin, it actually traveled 10 inches inside because it was bouncing around so much."

The Problem: The Old Maps Were Wrong

For a long time, scientists used a few different ways to calculate this "magnifying glass" number.

The "Straight Line" Guess: Some assumed the light traveled a fixed extra distance no matter how far apart the light source and detector were.
The "Average" Guess: Others tried to calculate the average distance the light traveled.

The authors of this paper found that these old methods are like using a flat map to navigate a mountain range. They work okay if you are far away, but if you are close to the mountain (or in this case, if the light source and detector are close together), the map is completely wrong. In fact, when the detectors are close, the old methods can be off by more than 100%. That's like guessing a 10-mile trip is actually 20 miles or 0 miles. This makes it impossible to get accurate measurements of blood oxygen levels, especially in areas where you need to detect signals from shallow depths.

The Solution: Two New, Better Maps

The researchers used a super-powerful computer simulation (called Monte Carlo) to track millions of virtual photons (light particles) as they bounced through a virtual block of tissue. They watched exactly where every single photon went.

Based on this, they created two new ways to calculate the DPF:

The "Gold Standard" Map (True DPF):
This is the perfect map. It calculates the exact distance every single photon traveled, weighted by how much light was lost along the way. It is 100% accurate.
- The Catch: It's too complicated to use in real life. You'd need a supercomputer to calculate it for every single measurement a doctor takes.
The "Smart Shortcut" Map (Inverse-Distance DPF):
This is the star of the show. The researchers noticed that the relationship between the distance of the light and the "winding path" it takes follows a specific, predictable pattern. It's like realizing that while a mountain road isn't a straight line, you can predict its length using a simple formula involving the straight-line distance and a bit of logarithmic math.
- The Benefit: This new formula is simple enough to use on a regular computer or even a handheld device, but it is almost as accurate as the "Gold Standard."

The Proof: Testing in the Lab

To make sure their new "Smart Shortcut" wasn't just a computer fantasy, they built a fake body part (a phantom).

They made a block of clear gel (like Jell-O) mixed with white powder to mimic how human skin scatters light.
They shined a real LED light into it and measured how much came out at different distances.
The Result: The new "Smart Shortcut" model predicted the results almost perfectly. The old methods, however, were wildly inaccurate, especially when the light source and detector were close together.

The Takeaway

The paper concludes that if you want to measure blood oxygen or tissue health using light, you need to stop using the old, simple "constant" numbers. They are too inaccurate, especially for close-range measurements.

Instead, the authors propose using their new Inverse-Distance Model. It's like upgrading from a rough sketch to a GPS navigation system: it's still easy to use, but it knows exactly how the terrain (the tissue) bends the path of the light, giving you a much truer picture of what's happening inside the body.

Technical Summary: A Novel Differential Pathlength Factor Model for Near-Infrared Diffuse Optical Imaging

Problem Statement
Continuous-wave near-infrared (CW-NIR) diffuse optical imaging is a widely used, cost-effective modality for monitoring tissue oxygenation and hemodynamics. However, its quantitative accuracy relies heavily on the Modified Beer–Lambert Law (MBLL), which requires an accurate Differential Pathlength Factor (DPF) to convert measured optical attenuation into chromophore concentrations. Existing DPF definitions suffer from formulation-dependent limitations that introduce significant errors, particularly in reflectance mode at small source–detector separations.

Current approaches include:

Mean-pathlength DPF: Defined as the ratio of mean photon pathlength to source–detector separation. It neglects absorption-weighted photon statistics, leading to overestimation in highly absorbing tissues.
Slope-based DPF: Defined as the slope of optical density changes with respect to the absorption coefficient. While theoretically grounded, it still focuses on average trajectories rather than the specific goal of concentration estimation.
Constant/Semi-infinite DPF: Often approximated as a constant value for separations greater than 2.5 cm based on asymptotic diffusion theory. This assumption ignores the distance dependence of the DPF and the variability of tissue optical properties, frequently resulting in estimation errors exceeding 100% in practical scenarios.

These inaccuracies limit the applicability of CW-NIR imaging in situations requiring large-area detection or where signal depth varies, as errors are most pronounced at shorter separations where signal strength is highest.

Methodology
The authors employed a dual approach combining computational simulations and physical experiments to evaluate existing models and propose new ones.

Computational Framework: Monte Carlo photon-transport simulations were performed using MCmatlab on a homogeneous tissue-slab geometry (29 × 29 × 5 cm³). The study spanned a broad range of optical properties: absorption coefficients ( $\mu_a$ ) from 0 to 0.5 cm⁻¹ and scattering coefficients ( $\mu_s$ ) from 0 to 500 cm⁻¹. For each condition, the "true" optical density was calculated, and six different DPF formulations were applied to estimate the absorption coefficient. The relative error between the estimated and ground-truth absorption coefficients was used as the primary metric for performance.
Experimental Validation: Controlled phantom experiments were conducted using a CW-NIR setup with a 750 nm LED and a silicon photodiode. Two homogeneous tissue-mimicking phantoms were fabricated using a super-absorbent polymer hydrogel matrix doped with Titanium Dioxide (TiO₂) at concentrations of 0.25 mg/mL and 1.00 mg/mL to create distinct scattering regimes. Measurements were taken in a diffuse reflectance geometry with source–detector separations ranging from 1 cm to 8.5 cm.

Key Contributions: Novel DPF Models
The paper introduces two new DPF models derived from the simulation data:

True DPF ( $DPF_{true}$ ): Defined directly from the MBLL by rearranging the equation to solve for DPF using the known ground-truth absorption coefficient and the simulated optical density. This model serves as the theoretical gold standard, capturing the full statistical nature of photon transport, including exponential weighting by absorption.
Inverse-Distance DPF ( $DPF_{inv}$ ): A practical, analytical model proposed to approximate the True DPF. Based on the observation that optical density can be accurately modeled as a linear combination of source–detector separation ( $d$ ) and $\log_{10}(d)$ with a non-zero intercept, the authors derived the following formulation:
$DPF_{inv}(d) = A + \frac{B}{d} + \frac{C\log_{10}(d)}{d}$
where $A$ , $B$ , and $C$ are regression coefficients dependent on the tissue's absorption and scattering properties. The authors further provided empirical relationships for these coefficients as functions of $\mu_a$ and $\mu_s$ , allowing for a fully empirical implementation without requiring new simulations for every measurement.

Results

Simulation Performance: The True DPF yielded negligible error across all tested conditions. The proposed Inverse-Distance DPF achieved relative errors below 10% across the majority of the optical property space and separation range. In contrast, conventional models (Constant, Mean-pathlength, Slope-based, and Semi-infinite) consistently produced errors exceeding 100%, particularly at source–detector separations below 8 cm.
Experimental Validation: Phantom experiments confirmed the simulation trends. The optical density measurements matched Monte Carlo predictions, validating the setup. When estimating absorption coefficients from experimental data, the True and Inverse-Distance models maintained relative errors below 20% across the measured separations. Conversely, conventional models exhibited errors frequently exceeding 100%, especially at shorter separations.
Parameter Behavior: The coefficients ( $A, B, C$ ) of the Inverse-Distance model were found to vary smoothly and predictably with optical properties, allowing for robust empirical calibration.

Significance and Claims
The paper claims that the proposed Inverse-Distance DPF model offers an optimal balance between physical rigor and computational efficiency. By moving away from the assumption of a constant DPF or simple mean-pathlength ratios, the new model significantly enhances the quantitative accuracy of CW-NIR imaging, particularly in the short-separation regimes where conventional models fail.

The authors emphasize that while the True DPF represents the theoretical limit of accuracy, the Inverse-Distance DPF is a practical alternative suitable for routine clinical and research applications. The study demonstrates that accurate DPF modeling is critical for reliable chromophore concentration estimation, especially in reflectance mode. The findings suggest that adopting these refined models can bridge the gap between the simplicity of CW-NIR systems and the quantitative rigor required for robust functional and clinical applications. The paper concludes modestly, noting that future work will need to address heterogeneous tissue structures and extend these modeling strategies to time-domain and frequency-domain modalities.