🔬 applied physics

A Reciprocity-Based Signal Compensation Framework for Ultrasonic Backscatter Measurements in Heterogeneous Scattering Media

This paper presents a reciprocity-based signal compensation framework that utilizes opposing inspection surfaces to estimate and remove distance-dependent propagation biases in ultrasonic backscatter measurements, thereby significantly improving the consistency of microstructural characterisation in heterogeneous anisotropic materials like Ti-6Al-4V compared to conventional attenuation-based methods.

Original authors: Wei Yi Yeoh, Bo Lan, Michael J. S. Lowe

Published 2026-05-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Wei Yi Yeoh, Bo Lan, Michael J. S. Lowe

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: Listening to a Noisy Room

Imagine you are trying to listen to the sound of rain hitting a specific spot on a roof. You want to know if the roof tiles in that spot are rough or smooth. However, the sound has to travel through the air to get to your ear. As the sound travels, it gets quieter and distorted because of the distance, the wind, and the air itself.

In this paper, the "roof" is a block of metal (specifically a titanium alloy used in airplanes), and the "sound" is an ultrasound wave. Scientists use ultrasound to look inside the metal to see its internal structure (like tiny crystals called "macrozones"). The problem is that as the ultrasound travels deeper into the metal, it loses energy and gets messy. This makes it hard to tell if a change in the signal is because the metal is actually different, or just because the sound traveled further.

The Problem: The "One-Way Street" Trap

Usually, when scientists try to fix this, they measure how much energy the sound loses as it travels all the way through the metal (like measuring how much a flashlight beam dims after passing through a foggy window). They use this "dimming" number to try to correct the signal.

The authors found a flaw in this method. Imagine you are walking through a forest with uneven terrain.

Going North: You walk through a muddy patch first, then a flat path. You get tired quickly at the start.
Going South: You walk through the flat path first, then the muddy patch. You get tired slowly at the start, then suddenly struggle.

Even though the total effort to cross the forest is the same in both directions, the experience of the journey is different at every step. Similarly, in these metal blocks, the ultrasound encounters different "muddy patches" (macrozones) depending on which side you scan from. The standard "dimming" correction assumes the journey is the same both ways, but it isn't. It fails to fix the signal properly when the metal is messy and uneven.

The Solution: The "Two-Way Mirror" Trick

The authors came up with a clever new method called Reciprocity-Based Compensation.

Think of it like this: You have a very long, foggy hallway. You stand at one end and shout, recording how your voice echoes back. Then, you stand at the other end and shout, recording that echo too.

You know for a fact that the total fog in the hallway is the same for both of you.
However, the fog might be thicker near your end in the first case, and thicker near the other end in the second case.

The authors' method looks at both recordings at the same time. Instead of guessing how the fog behaves, they find the shared pattern of the fog that exists in both recordings. They assume that the "slow, steady fade" of the sound is a shared trait of the hallway, even if the specific bumps and dips in the floor are different.

They calculate this "shared fade" mathematically and subtract it from both recordings. This leaves behind only the "bumps and dips" (the actual metal structure) without the "fog" (the distance-related signal loss).

What They Found

They tested this on two blocks of titanium metal that had different patterns of internal "macrozones" (clusters of crystals).

The Old Way (Attenuation): When they used the old method, the signals from the two sides still looked very different. It was like trying to match two photos of the same room taken from opposite corners, but one photo was still blurry and the other was sharp.
The New Way: When they used their new "Two-Way Mirror" method, the signals from both sides lined up perfectly. The "fog" was removed.
- They measured the "mismatch" between the two sides. The old method reduced the mismatch by about half.
- The new method reduced the mismatch by 96%. It made the two sides look almost identical, revealing the true structure underneath.

The Result: A Clearer Map

By removing the "distance bias," the scientists could create a much clearer 3D map of where the strong "echoes" (hotspots) were inside the metal.

Before: The hotspots looked like they were in different places depending on which side you looked from.
After: The hotspots from both sides overlapped perfectly, showing exactly where the interesting metal structures were.

Summary

The paper doesn't claim to fix the metal or predict how long a plane will last. Instead, it offers a better signal processing tool. It's like giving a photographer a new filter that removes the glare and distortion caused by the distance, allowing them to see the true texture of the object they are photographing, regardless of which side they are standing on. This makes it much easier to analyze complex, uneven materials like the titanium used in aerospace.

Technical Summary: Reciprocity-Based Signal Compensation for Ultrasonic Backscatter in Heterogeneous Media

Problem Statement
Ultrasonic backscatter measurements are a standard tool for microstructural characterisation, particularly in polycrystalline materials like Ti–6Al–4V. However, in materials exhibiting strong anisotropy and spatial heterogeneity (e.g., macrozones), interpreting backscatter signals is challenging. Distance-dependent propagation effects—such as beam divergence, cumulative scattering, and anisotropic wavefield distortion—introduce systematic biases that obscure genuine microstructural variations across depth.

Conventional compensation methods rely on attenuation measurements derived from front- and back-wall echoes. While attenuation profiles often satisfy reciprocity (i.e., total transmission loss is identical regardless of propagation direction), this paper demonstrates that reciprocal agreement in attenuation does not guarantee cross-directional consistency in depth-resolved backscatter profiles. In heterogeneous media, the ordering of scattering regions along the propagation path changes when the inspection direction is reversed, leading to different backscatter decay trends and amplitude offsets that simple attenuation correction cannot reconcile.

Methodology
The authors propose a cross-directional compensation framework that exploits the reciprocal constraint on through-thickness propagation without assuming that local backscatter responses are identical in opposing directions. The method operates as follows:

Data Acquisition: Ultrasonic backscatter RMS (Root-Mean-Square) amplitude profiles are acquired from opposing inspection faces of the same component volume.
Log-Domain Decomposition: The profiles are transformed into the logarithmic amplitude domain, where multiplicative propagation effects become additive. The signal is modeled as the sum of a shared, smooth, depth-dependent propagation baseline ( $\mathcal{l}(z)$ ) and direction-specific scattering residuals ( $s_i(z)$ ).
Baseline Estimation: A shared linear baseline is estimated in the log-domain using an anchor-based fitting approach. Two reference windows (near-field and far-field) are selected within the interior scattering region. The average logarithmic RMS amplitudes across both inspection directions are computed for these windows to determine a shared slope ( $\alpha$ ) and offset ( $\beta$ ). This approach ensures the estimated trend is consistent with reciprocal transmission while minimizing sensitivity to direction-specific fluctuations.
Compensation: A depth-dependent compensation factor is derived from the estimated baseline and applied uniformly to both opposing profiles. This removes the shared monotonic propagation bias while preserving residual local and direction-dependent scattering variations.

Key Contributions

Empirical Demonstration of Limitation: The study experimentally demonstrates that reciprocal agreement in attenuation does not imply cross-directional consistency in depth-resolved backscatter profiles, particularly in macrozone-containing Ti–6Al–4V samples.
Reciprocity-Based Framework: Introduction of a log-domain framework that estimates a shared depth-dependent propagation baseline directly from opposing-face backscatter measurements, avoiding the need for detailed microstructural models or assumptions of homogeneous material behavior.
Quantitative Validation: Quantification of the improvement in bidirectional consistency using monotonic bias metrics (standard deviation and absolute fitted gradient) and volumetric hotspot overlap metrics (Dice and Jaccard indices).

Results
The method was tested on two Ti–6Al–4V samples (small- and big-MTR) containing macrozones, analyzing six opposing-face signal pairs.

Reduction of Mismatch: Compared to the original uncompensated profiles, the proposed method reduced the mean standard deviation of the directional mismatch profile from 0.367 to 0.120 (a ~67% reduction) and the mean absolute fitted gradient from 0.171 to 0.0067 (a ~96% reduction).
Comparison to Attenuation Compensation: The proposed method outperformed conventional attenuation-based compensation, which only reduced the mean standard deviation to 0.199 and the gradient to 0.069. The reciprocity-based approach provided an additional ~39% reduction in variability and ~90% reduction in gradient relative to attenuation compensation.
Volumetric Consistency: In volumetric hotspot reconstruction (using the big-MTR face 3–6 pair), the proposed method significantly improved spatial alignment between opposing inspections. The Dice coefficient for hotspot overlap increased from 0.039 (original) to 0.073 (attenuation compensated) and 0.194 (cross-compensated), representing a ~2.7-fold improvement over attenuation compensation.
Robustness: Sensitivity analyses confirmed that the method is robust to variations in backscatter window length and anchor-window placement, with the ordering of profile agreement cases remaining consistent across tested parameters.

Significance and Claims
The paper positions this work as a signal normalisation step rather than a direct microstructure inversion procedure. The primary significance lies in providing a practical, low-assumption framework to reduce propagation-related bias in heterogeneous anisotropic materials without requiring detailed knowledge of the microstructure or additional experimental measurements.

By isolating the shared propagation trend, the method produces more consistent and interpretable depth-resolved backscatter profiles. This improves the reliability of ultrasonic signals as inputs for subsequent microstructure characterisation or inversion frameworks. The authors explicitly state that the compensated signals are not interpreted as direct indicators of macrozone size or morphology; rather, the method serves to remove depth-dependent bias, allowing the underlying directional scattering variations to be analyzed with greater fidelity. The approach is presented as a practical solution for backscatter analysis in materials where conventional attenuation correction fails to reconcile opposing-face measurements.