Unsupervised Adversarial Domain Adaptation for Uterine layer Segmentation: From Labeled Cine to Unlabeled Dynamic EPI MRI

2026-08-04

AI Summary

This study addresses the challenge of uterine layer segmentation in dynamic echo-planar imaging (EPI) MRI, which is hindered by susceptibility artifacts, low spatial resolution, and the absence of annotated data. To overcome these limitations, the authors propose an unsupervised adversarial domain adaptation framework that, for the first time, integrates temporal dynamics modeling with cross-modal knowledge transfer to enable segmentation transfer from labeled cine MRI to unlabeled dynamic EPI MRI. The method employs a U-Net-LSTM architecture coupled with multi-scale domain discriminators, leveraging temporal information through adversarial training. Experimental results demonstrate strong segmentation performance with Dice and Jaccard scores of 0.88 and 0.80, respectively. T2* relaxation times were measured as 108 ms, 76 ms, and 124 ms for the myometrium, junctional zone, and endometrium. Notably, in 14 out of 39 cases, the junctional zone area exhibited a significant negative correlation with T2*, suggesting a potential link between its contractile activity and oxygenation dynamics.

Abstract

Uterine peristalsis is a key physiological phenomenon responsible for various functions across the menstrual cycle, intimately linked to uterine wall microstructure. Alterations in uterine motion and tissue properties are implicated in the etiology of gynecological diseases, yet these processes have been studied in isolation. We introduce a dynamic multi-echo gradient echo EPI framework for simultaneous characterization and correlation of uterine peristaltic activity and time-resolved T2* changes at 0.55T. Inherent susceptibility artifacts, reduced resolution, and burden of manual uterine layer annotation are addressed by an unsupervised adversarial domain adaptation framework, transferring segmentation knowledge from labeled cine MRI to unlabeled dynamic EPI. We implemented Unet-LSTM with multi-scale domain discriminators that exploits temporal layer dynamics. A Dice score of 0.88 and Jaccard index of 0.80 was achieved. Mean T2* values were 108ms, 76ms, and 124ms for the myometrium, junctional zone, and endometrium. A negative correlation between junctional zone area and T2* was observed in 14/39 cases, providing first insights into oxygenation patterns associated with junctional zone contraction and motion, demonstrating feasibility of assessing the interplay between contractility and dynamic T2* changes.

Problem

• Uterine layer segmentation
• Dynamic EPI MRI
• Domain adaptation
• Manual annotation burden
• Susceptibility artifacts

Innovation

• unsupervised adversarial domain adaptation
• uterine layer segmentation
• dynamic EPI MRI
• U-Net-LSTM
• T2* mapping