EUVIMEDEUVIMED entdecken
Crossref

Recognition of Normal Fetal Echocardiograms Based on an Explainable Denoising Deep Learning Model

Shuhao Song iD, Yushan Liu, Ganqiong Xu iD, Yang Yang, Minzhi Ouyang, Shi Zeng iD

Journal of Clinical Ultrasound · 2026

DOI: 10.1002/jcu.70183 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Clinical Ultrasound, 2026. Verfasst von Shuhao Song und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Ultrasound Diagnosis, The Second Xiangya Hospital, Central South University, Changsha, Hunan, China. Hauptthemen laut MEDLINE: Deep Learning, Ultrasonography, Prenatal, Echocardiography, Fetal Heart. Die Arbeit nennt 22 Literaturangaben. Gefördert von: Natural Science Foundation of Hunan Province, Natural Science Foundation of Changsha City.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Purpose

To evaluate the proposed explainable denoising deep learning model, Grouped Shared Convolutional Attention Vision Transformer (GSCAViT), for classifying normal fetal echocardiogram.

Methods

A retrospective study was conducted on 358 fetal cardiac ultrasound exams with a total of 2501 images. GSCAViT utilized seven echocardiograms and was compared against baseline and enhanced models using metrics including accuracy, precision, recall, F1 score. The SHAP method clarified key image features, and the denoising guided GSCA module was assessed through visual comparisons and image quality metrics.

Results

GSCAViT achieved an accuracy of 97.1%, 99.4%, 81.3%, 72.9% on the validation and three test sets. In addition, GSCAViT achieved low error rates of 2.9%, 0.6%, 18.7%, and 27.1%. To improve upon existing Vision Transformer based models and denoising modules, we propose the GSCAViT, which integrates a novel denoising‐guided GSCA module for enhanced image quality and interpretability. SHAP visualizations confirmed the model's ability to identify critical cardiac structures, while the denoising module enhanced image quality, yielding the highest contrast‐to‐noise ratio and peak signal‐to‐noise ratio values.

Conclusion

GSCAViT outperformed baseline and several enhanced models in classifying seven types of normal fetal echocardiograms, SHAP visualization enhanced the interpretability of classification, comparisons of visual effectiveness and image parameters confirmed the efficacy of the GSCA module.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningUltrasonography, PrenatalEchocardiographyFetal HeartImage Interpretation, Computer-AssistedHumansFemaleRetrospective StudiesPregnancySignal-To-Noise Ratio

Stichwörter der Autor:innen

congenital heart diseasedeep learningfetal echocardiographyimage denoising

Autor:innen und Institutionen

  • Department of Ultrasound Diagnosis, The Second Xiangya Hospital, Central South University, Changsha, Hunan, China.
  • Hunan Provincial Ultrasound Diagnosis and Therapy Clinical Medical Research Center, Changsha, Hunan, China.
  • Institute of Ultrasound Imaging, the Second Xiangya Hospital, Central South University, Changsha, Hunan, China.

ORCID-Profile: Shuhao Song, Ganqiong Xu, Shi Zeng

Förderung

  • Natural Science Foundation of Hunan Province
  • Natural Science Foundation of Changsha City

Zitieren

APA

Song, S., Liu, Y., Xu, G., Yang, Y., Ouyang, M., & Zeng, S. (2026). Recognition of Normal Fetal Echocardiograms Based on an Explainable Denoising Deep Learning Model. Journal of Clinical Ultrasound. https://doi.org/10.1002/jcu.70183

Vancouver

Song S, Liu Y, Xu G, Yang Y, Ouyang M, Zeng S. Recognition of Normal Fetal Echocardiograms Based on an Explainable Denoising Deep Learning Model. Journal of Clinical Ultrasound. 2026. doi:10.1002/jcu.70183

BibTeX
@article{Song2026,
  author = {Song, Shuhao and Liu, Yushan and Xu, Ganqiong and Yang, Yang and Ouyang, Minzhi and Zeng, Shi},
  title = {Recognition of Normal Fetal Echocardiograms Based on an Explainable Denoising Deep Learning Model},
  journal = {Journal of Clinical Ultrasound},
  year = {2026},
  doi = {10.1002/jcu.70183}
}
Kostenlos starten →