Clipping spline: interactive, dynamic 4D volume clipping and analysis based on thin plate spline.

Abstract

Methods for seeing inside volumetric images are increasingly important with the rapid advancements in 3D and 4D (3D + time) biomedical imaging techniques. Here, we report a novel volume clipping method and its open-source implementation which enables unprecedented 4D visualization and analysis of embryonic mouse heart development with data from optical coherence tomography (OCT). Clipping a volume to extract information inside has long been a vital approach in biomedical image analysis; however, it is challenging to make a dynamic non-planar cutaway view that is simultaneously smooth, adjustable, efficient to compute, easy to control, and interactive in real time. We addressed this challenge by applying the thin plate spline (TPS) to create a new way of volume clipping, called the clipping spline. Specifically, the clipping spline produces a cutaway view by generating a binary mask based on the unique TPS surface that intersects with a set of 3D control points while having minimal curvature. We implemented this method in an open-source platform where the clipping spline can be interactively controlled for real-time, adjustable, and dynamic cutaway views into a volume. We also developed an algorithm that automatically connects and interpolates different sets of control points over time, providing 4D volume clipping. In addition to characterizing the clipping spline, we demonstrate its application by revealing a series of never-before-seen dynamics and processes of embryonic mouse heart development based on OCT data. We also show a TPS-based method for tracking the embryonic myocardium with control points over two timescales (heartbeat and development). Our results indicate that the clipping spline promises to be broadly used in volumetric biomedical image visualization and analysis, especially by the OCT community.