Tracing in 2D to Reduce the Annotation Effort for 3D Deep Delineation

TL;DR

This paper introduces a method to train 3D deep learning models for biomedical image segmentation using only 2D annotations from maximum intensity projections, significantly reducing annotation effort while maintaining performance.

Contribution

The novel approach leverages 2D MIP annotations to train 3D models, inspired by space carving, reducing annotation time by half without sacrificing accuracy.

Findings

Achieved comparable 3D segmentation performance with half the annotation effort.

Validated on microscopy images of neurons and blood vessels, and on MRA brain scans.

Demonstrated effectiveness of 2D-to-3D training transfer in biomedical imaging.

Abstract

The difficulty of obtaining annotations to build training databases still slows down the adoption of recent deep learning approaches for biomedical image analysis. In this paper, we show that we can train a Deep Net to perform 3D volumetric delineation given only 2D annotations in Maximum Intensity Projections (MIP). As a consequence, we can decrease the amount of time spent annotating by a factor of two while maintaining similar performance. Our approach is inspired by space carving, a classical technique of reconstructing complex 3D shapes from arbitrarily-positioned cameras. We will demonstrate its effectiveness on 3D light microscopy images of neurons and retinal blood vessels and on Magnetic Resonance Angiography (MRA) brain scans.