Fitting Skeletal Models via Graph-based Learning

TL;DR

This paper introduces a graph convolutional network-based method for efficient skeletonization of objects from segmentation masks, improving speed and reducing manual tuning compared to traditional approaches.

Contribution

The paper presents a novel graph-based learning approach for skeletonization that automates and accelerates the process of fitting skeletal models from segmentation data.

Findings

Effective on synthetic and real hippocampus data

Achieves promising accuracy and fast inference

Reduces manual parameter tuning

Abstract

Skeletonization is a popular shape analysis technique that models an object's interior as opposed to just its boundary. Fitting template-based skeletal models is a time-consuming process requiring much manual parameter tuning. Recently, machine learning-based methods have shown promise for generating s-reps from object boundaries. In this work, we propose a new skeletonization method which leverages graph convolutional networks to produce skeletal representations (s-reps) from dense segmentation masks. The method is evaluated on both synthetic data and real hippocampus segmentations, achieving promising results and fast inference.