TGMS-UNet: A Dual-Branch Network for Endometrial Segmentation in Ultrasound Utilizing Sequence Guidance and Multi-Scale Feature Adjustment
By
Qiao Wei
Xiaowen Liang
Yanfen Zhang
Yan Lin
Zhili Guo
Qing Zhang
Kezhen Wang
Zhang Xiao
Jie Lan
Linyuan Jin
Nian Hu
Hong Yu
Yaocheng Xiao
Zhiyi Chen
July 10, 2026
Clinical Scorecard: TGMS-UNet: A Dual-Branch Network for Endometrial Segmentation in Ultrasound Utilizing Sequence Guidance and Multi-Scale Feature Adjustment
At a Glance
Category Detail
Condition Endometrial Segmentation Key Mechanisms Sequence-guided dual-branch segmentation network incorporating geometric contour encoding and feature correction. Target Population Women undergoing transvaginal ultrasound for reproductive health assessment. Care Setting Transvaginal ultrasound imaging in clinical settings.
Key Highlights
TGMS-UNet improves segmentation accuracy in ambiguous ultrasound images. Incorporates clinical reasoning through distance-angle-based contour features. Addresses multi-scale feature misalignment with adaptive fusion mechanisms. Guideline-Based Recommendations
Diagnosis
Utilize transvaginal ultrasound for assessing endometrial conditions. Management
Implement automated segmentation to enhance objectivity in endometrial receptivity assessment. Monitoring & Follow-up
Regularly evaluate endometrial thickness and morphology through ultrasound imaging. Risks
Consider inherent imaging limitations such as speckle noise and boundary variability. Patient & Prescribing Data
Women experiencing infertility or other endometrial-related conditions.
Accurate segmentation can guide optimal timing for assisted reproductive procedures.
Clinical Best Practices
Employ automated segmentation to reduce subjectivity in endometrial evaluation. Integrate clinical prior knowledge into imaging analysis for improved outcomes. Related Resources & Content
