Saved in:
Bibliographic Details
Main Authors: Exler, David, Friederich, Nils, Krüger, Martin, Jbeily, John, Vitacolonna, Mario, Rudolf, Rüdiger, Mikut, Ralf, Reischl, Markus
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.14720
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911597745668096
author Exler, David
Friederich, Nils
Krüger, Martin
Jbeily, John
Vitacolonna, Mario
Rudolf, Rüdiger
Mikut, Ralf
Reischl, Markus
author_facet Exler, David
Friederich, Nils
Krüger, Martin
Jbeily, John
Vitacolonna, Mario
Rudolf, Rüdiger
Mikut, Ralf
Reischl, Markus
contents Myotubes are multinucleated muscle fibers serving as key model systems for studying muscle physiology, disease mechanisms, and drug responses. Mechanistic studies and drug screening thereby rely on quantitative morphological readouts such as diameter, length, and branching degree, which in turn require precise three-dimensional instance segmentation. Yet established pretrained biomedical segmentation models fail to generalize to this domain due to the absence of large annotated myotube datasets. We introduce a geometry-driven synthesis pipeline that models individual myotubes via polynomial centerlines, locally varying radii, branching structures, and ellipsoidal end caps derived from real microscopy observations. Synthetic volumes are rendered with realistic noise, optical artifacts, and CycleGAN-based Domain Adaptation (DA). A compact 3D U-Net with self-supervised encoder pretraining, trained exclusively on synthetic data, achieves a mean IPQ of 0.22 on real data, significantly outperforming three established zero-shot segmentation models, demonstrating that biophysics-driven synthesis enables effective instance segmentation in annotation-scarce biomedical domains.
format Preprint
id arxiv_https___arxiv_org_abs_2604_14720
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Data Synthesis Improves 3D Myotube Instance Segmentation
Exler, David
Friederich, Nils
Krüger, Martin
Jbeily, John
Vitacolonna, Mario
Rudolf, Rüdiger
Mikut, Ralf
Reischl, Markus
Computer Vision and Pattern Recognition
Myotubes are multinucleated muscle fibers serving as key model systems for studying muscle physiology, disease mechanisms, and drug responses. Mechanistic studies and drug screening thereby rely on quantitative morphological readouts such as diameter, length, and branching degree, which in turn require precise three-dimensional instance segmentation. Yet established pretrained biomedical segmentation models fail to generalize to this domain due to the absence of large annotated myotube datasets. We introduce a geometry-driven synthesis pipeline that models individual myotubes via polynomial centerlines, locally varying radii, branching structures, and ellipsoidal end caps derived from real microscopy observations. Synthetic volumes are rendered with realistic noise, optical artifacts, and CycleGAN-based Domain Adaptation (DA). A compact 3D U-Net with self-supervised encoder pretraining, trained exclusively on synthetic data, achieves a mean IPQ of 0.22 on real data, significantly outperforming three established zero-shot segmentation models, demonstrating that biophysics-driven synthesis enables effective instance segmentation in annotation-scarce biomedical domains.
title Data Synthesis Improves 3D Myotube Instance Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.14720