Registration of Multiview Echocardiography Sequences Using a Subspace Error Metric

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Processing

Background:

• 3D+t echocardiography (3DtE) is crucial for assessing left ventricular anatomy and function.
• Poor image quality and limited field of view can affect 3DtE diagnostic information.
• Multiview 3DtE registration aims to improve image quality and coverage by combining data from different acoustic windows.

Purpose of the Study:

• To introduce a novel subspace error metric for automatic and robust registration of multiview 3DtE sequences.
• To enhance the accuracy, robustness, and efficiency of multiview 3DtE image registration.

Main Methods:

• The proposed metric utilizes linear dimensionality reduction to leverage temporal variations in multiview 3DtE sequences.
• A low-dimensional subspace minimizes the impact of noise and artifacts on registration optimization.
• The method enables fast and robust registration without requiring an initial estimate.

Main Results:

• Validation on 48 pairwise registrations demonstrated superior performance compared to a state-of-the-art phase-based metric.
• The subspace error metric achieved a 21%/31% improvement in median/75th percentile target registration error.
• Mean execution time was reduced by 45% compared to the phase-based metric.

Conclusions:

• The subspace error metric significantly outperforms sum-of-squared differences and phase-based metrics for multiview 3DtE registration.
• It offers improvements in accuracy, robustness, and execution time.
• This metric holds potential to supersede standard methods for automatic and robust multiview 3DtE registration.