Precise Rayleigh-scale estimation from real ultrasound data

Determine whether precise scanner-normalized Rayleigh-scale estimates can be obtained from real ultrasound data despite the unknown-compression penalty and the large number of pooled windows required to approach the known-compression limit.

Background

The paper’s theoretical and synthetic analyses show that estimating the Rayleigh scale from B-mode data with an unknown compression contrast incurs a substantial variance penalty. Although pooling homogeneous windows sharing the same compression settings reduces the excess nuisance-parameter contribution at a rate proportional to 1/M, the experiments indicate that many windows may be required to approach the known-compression Rayleigh limit. The authors therefore leave unresolved whether precise scale estimates are practically attainable for real ultrasound data, where model misspecification, spatial dependence, unknown display processing, and limited availability of suitable homogeneous windows may further impede estimation. This question motivates the subsequent real-data validation but is not fully resolved by it.

References

The smaller values of $\sigma$ require more than $103$ windows to approach this level of relative error, suggesting that it may not be feasible to overcome the unknown-compression penalty in some practical scenarios. Together, the single-window and multi-window synthetic experiments show both sides of the theoretical result: unknown compression produces the predicted single-window penalty, while pooling reduces the excess nuisance contribution at the predicted $1/M$ rate towards the known-compression Rayleigh floor. However, the large number of windows required to approach the Rayleigh floor opens the question of whether it is possible to obtain precise scale estimates for real ultrasound data.

— Compression Hurts, Pooling Helps: Information Loss in Rayleigh-Scale Estimation from B-Mode Ultrasound  (2609.19525 - Smith et al., 17 Sep 2026) in Section 4, Experiments, subsection 4.1 Synthetic validation, paragraph “Multi-window”