Cell Instance Segmentation via Multi-Task Image-to-Image Schrödinger Bridge

Hayato Inoue; Shota Harada; Shumpei Takezaki; Ryoma Bise

cs.CVApr 14, 2026

Cell Instance Segmentation via Multi-Task Image-to-Image Schrödinger Bridge

Authors: Hayato Inoue, Shota Harada, Shumpei Takezaki, Ryoma Bise

Organizations: Dept. of Information Science and Technology, Kyushu University, Fukuoka, Japan

Abstract

Existing cell instance segmentation pipelines typically combine deterministic predictions with post-processing, which imposes limited explicit constraints on the global structure of instance masks. In this work, we propose a multi-task image-to-image Schrödinger Bridge framework that formulates instance segmentation as a distribution-based image-to-image generation problem. Boundary-aware supervision is integrated through a reverse distance map, and deterministic inference is employed to produce stable predictions. Experimental results on the PanNuke dataset demonstrate that the proposed method achieves competitive or superior performance without relying on SAM pre-training or additional post-processing. Additional results on the MoNuSeg dataset show robustness under limited training data. These findings indicate that Schrödinger Bridge-based image-to-image generation provides an effective framework for cell instance segmentation.

Figures & tables

Fig. 1: Conceptual illustration of distribution-based instance mask modeling. For a given input image, multiple candidate instance masks may exist, while only a subset corresponds to morphologically plausible cell instances.

Fig. 2: Overview of the proposed Multi-Task Image-to-Image SB framework.

Method	bPQ $\uparrow$	F1 $\uparrow$	Recall $\uparrow$	Precision $\uparrow$
U-Net	0.5799	0.7930	0.7989	0.7874
U-Net+R	0.5837	0.7942	0.7972	0.7908
CellViT	0.6149	0.8037	0.8328	0.7984
CellViT+proc	0.6221	0.7946	0.7847	0.8357
Ours	0.6362	0.8106	0.8011	0.8458

TABLE I: Quantitative comparison on the PanNuke dataset under the large-scale training setting.

Method	bPQ $\uparrow$	F1 $\uparrow$	Recall $\uparrow$	Precision $\uparrow$
Single-Task Mask	0.5044	0.6815	0.6790	0.7304
Single-Task RVdist	0.5903	0.7756	0.7460	0.8373
Multi-Task	0.6341	0.8100	0.8017	0.8448

TABLE II: Ablation study on the PanNuke dataset under the large-scale training setting.

Fig. 3: Qualitative cell instance segmentation results on the PanNuke dataset.

Fig. 4: Transformation process at time step.

Method	bPQ $\uparrow$	F1 $\uparrow$	Recall $\uparrow$	Precision $\uparrow$
U-Net	0.5066	0.8051	0.8426	0.7707
U-Net+R	0.5335	0.8098	0.8541	0.7698
CellViT	0.5842	0.8468	0.8891	0.8142
CellViT+proc	0.6066	0.8444	0.8506	0.8449
Ours	0.6017	0.8563	0.8689	0.8493

TABLE III: Quantitative comparison on the MoNuSeg dataset under the limited training data setting.

Fig. 5: Joint distributions of cell size and circularity for the ground truth (GT), the proposed method, and CellViT variants on the PanNuke dataset. The proposed method produces a distribution closer to the GT, suppressing morphologically atypical cell instances.

Instance Segmentation

Instance segmentation identifies each object and assigns it a separate pixel-level mask in images or video.

See all 159 papers

Boundary-Aware Instance Segmentation in Microscopy Imaging
Mar 22, 2026Thomas Mendelson, Joshua Francois, Galit Lahav +1Instance Segmentation Boundary-Aware Segmentation
The School of Electrical and Computer Engineering, Ben-Gurion University of the Negev · Department of Systems Biology, Harvard Medical School
Unsupervised Learning of Cell Instances with Generative Routing Pyramids
Aug 17, 2026Ziwen Liu, Martin WeigertInstance Segmentation Unsupervised Image Segmentation
Center for Scalable Data Analytics and Artificial Intelligence (ScaDS.AI), Technische Universität Dresden, Germany
One Click per Cell Type Suffices: Training-free Group Interaction for Cell Instance Segmentation
May 28, 2026Sanghyun Jo, Seo Jin Lee, Seohyung Hong +4Interactive Medical Image Segmentation Cross-Modality Medical Image Segmentation
1OGQ, Korea · 2Seoul National University, Korea · 3LG CNS, Korea

Mar 22, 2026J/K move · Enter open · S save

Boundary-Aware Instance Segmentation in Microscopy Imaging

Thomas Mendelson, Joshua Francois, Galit Lahav, Tammy Riklin-Raviv

The School of Electrical and Computer Engineering, Ben-Gurion University of the Negev · Department of Systems Biology, Harvard Medical School

Instance Segmentation Boundary-Aware Segmentation Microscopy Image Analysis

Open paper AskPDF arXiv

Accurate delineation of individual cells in microscopy videos is essential for studying cellular dynamics, yet separating touching or overlapping instances remains a persistent challenge. Although foundation-model for segmentation such as SAM have broadened the accessibility of image segmentation, they still struggle to separate nearby cell instances in dense microscopy scenes without extensive prompting. We propose a prompt-free, boundary-aware instance segmentation framework that predicts signed distance functions (SDFs) instead of binary masks, enabling smooth and geometry-consistent modeling of cell contours. A learned sigmoid mapping converts SDFs into probability maps, yielding sharp boundary localization and robust separation of adjacent instances. Training is guided by a unified Modified Hausdorff Distance (MHD) loss that integrates region- and boundary-based terms. Evaluations on both public and private high-throughput microscopy datasets demonstrate improved boundary accuracy and instance-level performance compared to recent SAM-based and foundation-model approaches. Source code is available at: https://github.com/ThomasMendelson/BAISeg.git

Abstract

Instance Segmentation

Explore similar work

Boundary-Aware Instance Segmentation in Microscopy Imaging

Unsupervised Learning of Cell Instances with Generative Routing Pyramids

One Click per Cell Type Suffices: Training-free Group Interaction for Cell Instance Segmentation