Nonlinear Shape Regression For Filtering Segmentation Results From Calcium Imaging

TL;DR

This paper introduces a shape filtering method for calcium imaging neuron segmentation that improves shape smoothness and accuracy by using a shape manifold and data-driven weighting, outperforming unweighted models.

Contribution

It presents a novel shape filter using a shape manifold and data-driven weights to enhance segmentation results in calcium imaging.

Findings

The shape filter effectively smooths neuron shapes.

Weighted methods outperform unweighted models.

Quantitative results show improved segmentation accuracy.

Abstract

A shape filter is presented to repair segmentation results obtained in calcium imaging of neurons in vivo. This post-segmentation algorithm can automatically smooth the shapes obtained from a preliminary segmentation, while precluding the cases where two neurons are counted as one combined component. The shape filter is realized using a square-root velocity to project the shapes on a shape manifold in which distances between shapes are based on elastic changes. Two data-driven weighting methods are proposed to achieve a trade-off between shape smoothness and consistency with the data. Intuitive comparisons of proposed methods via projection onto Cartesian maps demonstrate the smoothing ability of the shape filter. Quantitative measures also prove the superiority of our methods over models that do not employ any weighting criterion.