Cramér-Rao Lower Bound for Point Based Image Registration With Heteroscedastic Error Model for Application in Single...

Area of Science:

• * Statistical modeling
• * Image processing
• * Computational biology

Background:

• * Feature-based image registration requires accurate estimation of transformation parameters.
• * Errors in control point localization, with varying covariance, complicate registration accuracy.
• * Fluorescence microscopy image registration presents specific challenges due to localization errors.

Purpose of the Study:

• * To derive the Cramér-Rao lower bound for affine transformation parameter estimation in a multivariate heteroscedastic errors-in-variables model.
• * To develop and analyze bounds for feature position estimation in registered images, particularly for fluorescence microscopy.
• * To simplify and validate these bounds under common assumptions in microscopy.

Main Methods:

• * Derivation of the Cramér-Rao lower bound for affine transformation parameters.
• * Application of the bound to feature position estimation in registered images.
• * Simplification of bounds for scalar covariance matrices and validation through simulations and experimental data.

Main Results:

• * The Cramér-Rao lower bound was successfully derived for the specified model.
• * Simplified expressions for the bound were obtained for scalar covariance matrices.
• * Theoretical results were validated using simulations and experimental data, showing agreement with prior estimators under certain assumptions.

Conclusions:

• * The derived Cramér-Rao lower bounds provide theoretical limits for parameter estimation accuracy in heteroscedastic errors-in-variables models.
• * These bounds are particularly relevant for feature-based image registration in applications like fluorescence microscopy.
• * The study confirms the utility of the derived bounds through empirical validation.