Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A label-free and data-free training strategy for vasculature segmentation in serial sectioning OCT data

Etienne Chollet, Yael Balbastre, Caroline Magnain, Bruce Fischl, Hui Wang

TL;DR

The study addresses automated segmentation of brain vasculature in serial-section OCT (sOCT), a task hindered by noise and scarce labeled data. It introduces a label-free training strategy using synthetic data, including domain-randomized OCT textures, spline-based vascular geometry, and comparisons with realistic CCO-generated labels, trained on a U-Net with residual blocks and Dice loss. Dice scores for complex-label and CCO-label training are broadly comparable (~0.61–0.63) but with different error profiles, suggesting that exact physical realism is not strictly necessary for effective training. This synthetic-label, data-free approach enables scalable, accurate analysis of neurovasculature in postmortem sOCT and can generalize to other imaging modalities with limited annotations.

Abstract

Serial sectioning Optical Coherence Tomography (sOCT) is a high-throughput, label free microscopic imaging technique that is becoming increasingly popular to study post-mortem neurovasculature. Quantitative analysis of the vasculature requires highly accurate segmentation; however, sOCT has low signal-to-noise-ratio and displays a wide range of contrasts and artifacts that depend on acquisition parameters. Furthermore, labeled data is scarce and extremely time consuming to generate. Here, we leverage synthetic datasets of vessels to train a deep learning segmentation model. We construct the vessels with semi-realistic splines that simulate the vascular geometry and compare our model with realistic vascular labels generated by constrained constructive optimization. Both approaches yield similar Dice scores, although with very different false positive and false negative rates. This method addresses the complexity inherent in OCT images and paves the way for more accurate and efficient analysis of neurovascular structures.

A label-free and data-free training strategy for vasculature segmentation in serial sectioning OCT data

TL;DR

The study addresses automated segmentation of brain vasculature in serial-section OCT (sOCT), a task hindered by noise and scarce labeled data. It introduces a label-free training strategy using synthetic data, including domain-randomized OCT textures, spline-based vascular geometry, and comparisons with realistic CCO-generated labels, trained on a U-Net with residual blocks and Dice loss. Dice scores for complex-label and CCO-label training are broadly comparable (~0.61–0.63) but with different error profiles, suggesting that exact physical realism is not strictly necessary for effective training. This synthetic-label, data-free approach enables scalable, accurate analysis of neurovasculature in postmortem sOCT and can generalize to other imaging modalities with limited annotations.

Abstract

Serial sectioning Optical Coherence Tomography (sOCT) is a high-throughput, label free microscopic imaging technique that is becoming increasingly popular to study post-mortem neurovasculature. Quantitative analysis of the vasculature requires highly accurate segmentation; however, sOCT has low signal-to-noise-ratio and displays a wide range of contrasts and artifacts that depend on acquisition parameters. Furthermore, labeled data is scarce and extremely time consuming to generate. Here, we leverage synthetic datasets of vessels to train a deep learning segmentation model. We construct the vessels with semi-realistic splines that simulate the vascular geometry and compare our model with realistic vascular labels generated by constrained constructive optimization. Both approaches yield similar Dice scores, although with very different false positive and false negative rates. This method addresses the complexity inherent in OCT images and paves the way for more accurate and efficient analysis of neurovascular structures.
Paper Structure (3 sections, 2 figures)

This paper contains 3 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Methods
  3. Results & Discussion

Figures (2)

  • Figure 1: 3D renderings of CCO (A) and complex (B) labels, and (C) data synthesis pipeline for unsupervised training of UNet in vasculature segmentation task.
  • Figure 2: (A): Qualitative comparison of model performance. (B): Dice score, false positive rate (FPR), and false negative rate (FNR) for models (n=4) trained on complex, CCO, and simple datasets. ANOVA analysis was used to delineate statistical significance (****: $P<0.0001$, ***: $P<0.001$)