Automated scan quality evaluation for DDH using transfer learning - Development of a novel ensemble system

Aug 25, 2026

Automated scan quality assessment for Developmental Dysplasia of the Hip (DDH) is developed using transfer learning models that evaluate pelvic ultrasound images by scoring five anatomical landmarks. The approach targets variability in ultrasound landmark placement that affects diagnostic accuracy with the Graf method. Models are verified with gradient-weighted class activation mapping to confirm learned image features. An ensemble system is trained on 1,891 subjects from two Korean hospitals and uses an alternative sequence method to improve real-time scan quality assessment.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Abstract

Background

Methods

Results

Introduction

OPEN ACCESS Citation: Ko Y-K, Lee S-B, Lee S-W (2025) Automated scan quality evaluation for DDH using transfer learning: Development of a novel ensemble system. PLoS ONE 20(3): e0317251 . [https://doi.org/10.1371/journal.pone.0317251](https://doi.org/10.1371/journal.pone.0317251) Editor: Fahad Farhan Almutairi, King Abdulaziz University, SAUDI ARABIA Received: July 21, 2024 Accepted: December 23, 2024 Published: March 27, 2025 Copyright: © 2025 Ko et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data availability statement: The data used in this study were collected as part of a national project led by Industry-Academic Cooperation Foundation, Keimyung University, and we do not have the authority to share the data directly. The data are publicly available atthe link provided below, and researchers can request access through this platform. If any researchers require access to the data, they can apply via this link to obtain the necessary dataset. Data are available from AI hub (contact RESEARCH ARTICLE Automated scan quality evaluation for DDH using transfer learning: Development of a novel ensemble system Yeon-Kyoung Ko1,2, Seung-Bo Lee2, Si-Wook Lee3* 1 Department of Brain and Cognitive Engineering, Korea University, Seoul, South Korea, 2 Department of Medical Informatics, Keimyung University School of Medicine, Daegu, South Korea, 3 Department of Orthopedic Surgery, Dongsan Medical Center, School of Medicine, Keimyung University, Daegu, South Korea * [shuk@dsmc.or.kr](shuk@dsmc.or.kr) Abstract Background Developmental Dysplasia of the Hip (DDH) is a relatively common hip joint disorders in infants, affecting one to three per a thousand births. If found early, it can be treated preemptively by simple non-invasive methods. But if not, then several surgical procedures may be required that can cause high economic burden. The accuracy of diagnosis using ultrasound (US) images heavily relies on locating anatomical landmarks on the image. However, there is an intra-observer/inter-observer variability in determining the exact location of the landmarks. In this study, an automated scan quality assessment system of pelvic US image by evaluating quality of five landmarks using transfer learning models was proposed. Methods US images from 1,891 subjects were obtained at two hospitals in the Republic of Korea (henceforth Korea). Also, an ensemble system was developed using transfer learning models to automatically evaluate the scan quality by scoring five anatomical landmarks. Gradient-weighted class activation mapping was used for verifying whether models that reflect the geographical features of the images had been properly trained. Considering the applicability in the real-time environment, this study proposes an alternative sequence method (ASM) that has been discovered to have improved the lapse of scan quality assessment. Results All the selected models achieved kappa values of 0.6 or higher, indicating substantial agreement, and the AUC score for classifying standard images based on the total score was 0.89. The activation map of the trained models properly reflected the structural features of the image. The time lapse for standard image classification was 0.35 second per via [https://www.aihub.or.kr/aihubdata/data/](https://www.aihub.or.kr/aihubdata/data/)[ ](https://www.aihub.or.kr/aihubdata/data/)[view.do?currMenu=1](view.do?currMenu=1)15&topMenu=100&aihubDataSe=data&dataSetSn=583) for researcher who request and meet the criteria for access. Funding: This work was supported by the Korea Medical Device Development Fund grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Trade, Industry and Energy, the Ministry of Health & Welfare, the Ministry of Food and Drug

FAQs

Why is scan quality assessment important for DDH ultrasound diagnosis?

DDH ultrasound accuracy depends on correctly locating anatomical landmarks. Landmark placement varies between observers, reducing diagnostic consistency and effectiveness.

How does the proposed system evaluate pelvic ultrasound scan quality?

It uses transfer learning-based models assembled into an ensemble system to score five anatomical landmarks on pelvic ultrasound images.

How were the models validated for proper learning and real-time applicability?

Gradient-weighted class activation mapping was used to verify that activation maps reflect structural image features. Runtime performance was assessed using standard and alternative sequence methods for scan quality classification.

Aug 25, 2026

Upload by Seraphina