Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jan 16, 2026

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

ADVERSARIAL SYNTHESIS LEARNING ENABLES SEGMENTATION WITHOUT TARGET MODALITY GROUND TRUTH.

Yuankai Huo¹, Zhoubing Xu¹, Shunxing Bao²

¹Electrical Engineering, Vanderbilt University, Nashville, TN, USA 37235.

Proceedings. IEEE International Symposium on Biomedical Imaging

|September 29, 2025

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

A model to estimate spleen volume from palpated spleen length in patients with myelofibrosis.

Annals of hematology·2026

Same author

Add-on parsaclisib for patients with myelofibrosis and suboptimal response to ruxolitinib: a randomized phase 3 study.

The oncologist·2026

Same author

Early adipose tissue wasting in a novel preclinical model of human lung cancer cachexia.

bioRxiv : the preprint server for biology·2024

Same author

MICCAI-CDMRI 2023 QuantConn Challenge Findings on Achieving Robust Quantitative Connectivity through Harmonized Preprocessing of Diffusion MRI.

ArXiv·2024

Same author

Primary Visual Pathway Changes in Individuals With Chronic Mild Traumatic Brain Injury.

JAMA ophthalmology·2024

Same author

Super-resolution multi-contrast unbiased eye atlases with deep probabilistic refinement.

Journal of medical imaging (Bellingham, Wash.)·2024

This study introduces an end-to-end network for medical image segmentation, enabling the transfer of labels between MRI and CT scans without manual annotation. The novel approach significantly improves segmentation accuracy for CT images, outperforming existing methods.

Area of Science:

Medical Imaging
Deep Learning
Computer Vision

Background:

Deep learning segmentation models often lack generalizability across different imaging modalities (e.g., MRI, CT) or patient cohorts.
Manual labeling of new datasets for different modalities or diseases is time-consuming and resource-intensive.
Existing methods using CycleGAN for cross-modality synthesis and segmentation are often performed in two independent stages, failing to leverage complementary information.

Purpose of the Study:

To develop a novel end-to-end synthesis and segmentation network (EssNet) for unpaired MRI to CT image synthesis and CT splenomegaly segmentation.
To eliminate the need for manual labels on the target CT images during the segmentation process.
To improve the generalizability and accuracy of deep learning-based medical image segmentation.

Related Experiment Videos

Last Updated: Jan 16, 2026

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

Main Methods:

Proposed a novel end-to-end synthesis and segmentation network (EssNet).
Achieved simultaneous unpaired MRI to CT image synthesis and CT splenomegaly segmentation.
Trained the network without requiring manual labels on the target CT images.

Main Results:

The end-to-end EssNet achieved a median Dice similarity coefficient (DSC) of 0.9188 for CT splenomegaly segmentation.
EssNet significantly outperformed a two-stage synthesis-then-segmentation strategy (DSC: 0.8801).
EssNet also surpassed canonical multi-atlas segmentation (DSC: 0.9125) and a ResNet method (DSC: 0.9107) that utilized CT manual labels.

Conclusions:

The proposed EssNet effectively addresses the generalizability limitation in deep learning segmentation by enabling cross-modality label transfer.
Simultaneous synthesis and segmentation in an end-to-end framework capture complementary information, leading to superior performance.
EssNet demonstrates the potential to reduce manual annotation efforts and improve segmentation accuracy in medical imaging, even without target-domain labels.