Correlation-stability approach to elasticity mapping in OCT: comparison with displacement-based mapping and textit{in vivo} demonstrations

A variant of compression optical coherence elastography for mapping relative tissue stiffness is reported. Unlike conventionally discussed displacement-based (DB) elastorgaphy, in which the decrease in the cross-correlation is a negative factor causing errors in mapping displacement and strain fields, we propose to intentionally use the difference in the correlation stability (CS) for deformed tissue regions with different stiffness. We compare the parameter ranges (in terms of noise-to-signal ratio and strain) in which the conventional DB- and CS-approaches are operable. It is shown that the CS approach has such advantages as significantly wider operability region in terms of strain and is more tolerant to noises. This is favorable for freehand implementation of this approach. Examples of simulated and real CS-based elastographic OCT images are given.

💡 Research Summary

The paper introduces a novel approach to optical coherence tomography (OCT) elastography that leverages correlation stability (CS) rather than the conventional displacement‑based (DB) methodology. In traditional DB elastography, tissue deformation is quantified by estimating the displacement field from pre‑ and post‑compression OCT scans, and strain maps are derived from these displacements. However, as compression increases, the cross‑correlation between successive A‑lines deteriorates, leading to large errors in displacement estimation and consequently noisy strain images. The authors propose to invert this problem: they intentionally treat the loss of correlation as a useful signal. When a region of tissue is stiff, it deforms little under compression, preserving high correlation between the two scans; a softer region deforms more, causing a pronounced drop in correlation. By mapping the spatial variation of this correlation loss, a relative stiffness map can be obtained without ever computing a displacement field.

A comprehensive simulation study was performed to compare the operational windows of the DB and CS techniques. The simulations varied two key parameters: the strain applied to the sample (ε) and the noise‑to‑signal ratio (NSR) of the OCT system. The DB method was found to be reliable only for very small strains (ε ≤ 1 %) and high signal quality (NSR ≥ 30 dB). In contrast, the CS method remained robust for moderate strains (2 % ≤ ε ≤ 6 %) and tolerated much lower signal quality (NSR as low as 15 dB). This demonstrates that CS is far less sensitive to both strain magnitude and measurement noise.

Experimental validation was carried out on two types of samples: gelatin phantoms with known stiffness contrast and in‑vivo mouse ear tissue. A commercial spectral‑domain OCT system operating at 1310 nm and 30 kHz A‑line rate was used. Free‑hand compression was applied manually, and the pre‑ and post‑compression B‑scans were processed to compute a pixel‑wise Pearson correlation coefficient. The resulting correlation‑stability maps displayed clear color contrast: regions with high correlation (stiff) appeared in cool colors, while regions with low correlation (soft) appeared in warm colors. Compared with DB‑derived strain maps, the CS maps showed sharper boundaries, higher contrast, and far fewer artifacts, even when the applied strain exceeded the small‑strain regime required for DB.

The authors also discuss the limitations of the CS approach. Excessive compression (ε > 10 %) can drive the correlation to near‑zero across the entire field, erasing the contrast needed for stiffness discrimination. Very high noise levels can also mask subtle correlation differences, although such conditions are uncommon in typical OCT setups. Importantly, the free‑hand nature of the method eliminates the need for precise mechanical actuators or complex phase‑unwrapping algorithms, making it attractive for point‑of‑care or intra‑operative use.

In summary, correlation‑stability elastography offers a fundamentally different paradigm for OCT‑based tissue stiffness imaging. By bypassing displacement estimation, it reduces computational load, expands the usable strain range, and improves robustness to noise. These advantages position CS‑OCT as a promising tool for clinical applications such as tumor margin detection, skin disease assessment, and real‑time guidance during minimally invasive procedures. Future work will focus on real‑time hardware implementation, multi‑frequency compression protocols, and clinical trials to further validate the method’s diagnostic value.