Quality evaluation of ultrasound images for fetal crown–rump length measurement at 11 to 14 weeks’ gestation: A fusion model can be interpreted using SHAP method has been removed from our system
Keywords:
Deep learning; Quality evaluation; Ultrasound standard planeAbstract
Objectives: To explore a fusion model designed for the quality evaluation of ultrasound images utilized in fetal crown–rump length (CRL) measurement, and to use SHapley Additive exPlanations (SHAP) method to elucidate the model's decision-making processes.
Methods: We retrospectively collected 1149 images of midsagittal planes of the entire fetus during early pregnancy from two hospitals. Two senior radiologists categorized the images into standard and non-standard planes. Seven image segmentation models were trained to select the best model for automatically segmenting the region of interest. The radiomics features and deep transfer learning (DTL) features were extracted and selected to establish radiomics models and DTL models. We also constructed fusion models to enhance the classification performance and the optimal one underwent comparison with radiologists. The SHAP method was employed to interpret and visualize the model.
Results: The DeepLabV3 ResNet101 segmentation model demonstrated the highest performance (DSC: 97.15%). The early fusion model exhibited superior classification performance in validation set (AUC: 0.947, 95% CI: 0.924-0.970, accuracy: 88.4%, sensitivity: 83.0%, specificity: 92.7%, PPV: 90.1%, NPV: 87.3%, precision: 90.1%). The model demonstrated performance commensurate with that of senior radiologists while surpassing junior radiologists. Notably, when leveraging the model's support, there was a substantial improvement in their overall performance.
Conclusions: The early fusion model demonstrated satisfactory performance in the intelligent quality evaluation of ultrasound images for CRL measurement. It has the potential to enhance the professional skills of junior radiologists.
References
[1] Salomon L J, Alfirevic Z, Bilardo C M, et al. ISUOG practice guidelines: performance of first-trimester fetal ultrasound scan[J]. Ultrasound Obstet Gynecol, 2013,41(1):102-113.
[2] Bilardo C M, Chaoui R, Hyett J A, et al. ISUOG Practice Guidelines (updated): performance of 11-14-week ultrasound scan[J]. Ultrasound Obstet Gynecol, 2023,61(1):127-143.
[3] Salomon L J, Alfirevic Z, Da S C F, et al. ISUOG Practice Guidelines: ultrasound assessment of fetal biometry and growth[J]. Ultrasound Obstet Gynecol, 2019,53(6):715-723.
[4] Papageorghiou A T, Kennedy S H, Salomon L J, et al. International standards for early fetal size and pregnancy dating based on ultrasound measurement of crown-rump length in the first trimester of pregnancy[J]. Ultrasound Obstet Gynecol, 2014,44(6):641-648.
[5] Wanyonyi S Z, Napolitano R, Ohuma E O, et al. Image-scoring system for crown-rump length measurement[J]. Ultrasound Obstet Gynecol, 2014,44(6):649-654.
[6] Ramirez Z R, Ghi T. Use of artificial intelligence and deep learning in fetal ultrasound imaging[J]. Ultrasound Obstet Gynecol, 2023,62(2):185-194.
[7] Fiorentino M C, Villani F P, Di Cosmo M, et al. A review on deep-learning algorithms for fetal ultrasound-image analysis[J]. Med Image Anal, 2023,83:102629.
[8] Shazly S A, Trabuco E C, Ngufor C G, et al. Introduction to Machine Learning in Obstetrics and Gynecology[J]. Obstet Gynecol, 2022,139(4):669-679.
[9] Wang Y, Lang J, Zuo J Z, et al. The radiomic-clinical model using the SHAP method for assessing the treatment response of whole-brain radiotherapy: a multicentric study[J]. Eur Radiol, 2022,32(12):8737-8747.
[10] Rodríguez-Pérez R, Bajorath J. Interpretation of Compound Activity Predictions from Complex Machine Learning Models Using Local Approximations and Shapley Values[J]. J Med Chem, 2020,63(16):8761-8777.
[11] Zhao S, Wang J, Jin C, et al. Stacking Ensemble Learning-Based [(18)F]FDG PET Radiomics for Outcome Prediction in Diffuse Large B-Cell Lymphoma[J]. J Nucl Med, 2023,64(10):1603-1609.
[12] Umans E, Dewilde K, Williams H, et al. Artificial Intelligence in Imaging in the First Trimester of Pregnancy: A Systematic Review[J]. Fetal Diagn Ther, 2024,51(4):343-356.
[13] Sciortino G, Tegolo D, Valenti C. Automatic detection and measurement of nuchal translucency[J]. Comput Biol Med, 2017,82:12-20.
[14] Zhen C, Wang H, Cheng J, et al. Locating Multiple Standard Planes in First-Trimester Ultrasound Videos via the Detection and Scoring of Key Anatomical Structures[J]. Ultrasound Med Biol, 2023,49(9):2006-2016.
[15] Tsai P Y, Hung C H, Chen C Y, et al. Automatic Fetal Middle Sagittal Plane Detection in Ultrasound Using Generative Adversarial Network[J]. Diagnostics (Basel), 2020,11(1).
[16] Nie S, Yu J, Chen P, et al. Automatic Detection of Standard Sagittal Plane in the First Trimester of Pregnancy Using 3-D Ultrasound Data[J]. Ultrasound Med Biol, 2017,43(1):286-300.
[17] Ji C, Liu K, Yang X, et al. A novel artificial intelligence model for fetal facial profile marker measurement during the first trimester[J]. BMC Pregnancy Childbirth, 2023,23(1):718.
[18] Sun Y, Zhang L, Dong D, et al. Application of an individualized nomogram in first-trimester screening for trisomy 21[J]. Ultrasound Obstet Gynecol, 2021,58(1):56-66.
[19] Zhang L, Dong D, Sun Y, et al. Development and Validation of a Deep Learning Model to Screen for Trisomy 21 During the First Trimester From Nuchal Ultrasonographic Images[J]. JAMA Netw Open, 2022,5(6):e2217854.
[20] Walker M C, Willner I, Miguel O X, et al. Using deep-learning in fetal ultrasound analysis for diagnosis of cystic hygroma in the first trimester[J]. PLoS One, 2022,17(6):e269323.
[21] Lin Q, Zhou Y, Shi S, et al. How much can AI see in early pregnancy: A multi-center study of fetus head characterization in week 10-14 in ultrasound using deep learning[J]. Comput Methods Programs Biomed, 2022,226:107170.
[22] Xue H, Yu W, Liu Z, et al. Early Pregnancy Fetal Facial Ultrasound Standard Plane-Assisted Recognition Algorithm[J]. J Ultrasound Med, 2023,42(8):1859-1880.
[23] Drukker L, Noble J A, Papageorghiou A T. Introduction to artificial intelligence in ultrasound imaging in obstetrics and gynecology[J]. Ultrasound Obstet Gynecol, 2020,56(4):498-505.
[24] Xu J, Chen T, Fang X, et al. Prediction model of pressure injury occurrence in diabetic patients during ICU hospitalization--XGBoost machine learning model can be interpreted based on SHAP[J]. Intensive Crit Care Nurs. 2024 Aug; 83:103715.
[25] Fu Q, Wu Y, Zhu M, et al. Identifying cardiovascular disease risk in the U.S. population using environmental volatile organic compounds exposure: A machine learning predictive model based on the SHAP methodology[J]. Ecotoxicol Environ Saf, 2024 Nov 1;286:117210.
[26] He F, Wang Y, Xiu Y, et al. Artificial Intelligence in Prenatal Ultrasound Diagnosis[J]. Front Med (Lausanne). 2021 Dec 16; 8:729978.
[27] Espinoza J, Good S, Russell E, et al. Does the use of automated fetal biometry improve clinical work flow efficiency[J]? J Ultrasound Med. 2013 May;32(5):847-50.
[28] Yazdi B, Zanker P, Wanger P, et al. Optimal caliper placement: manual vs automated methods[J]. Ultrasound Obstet Gynecol. 2014 Feb;43(2):170-5.
[29] Subbaswamy A, Saria S. From development to deployment: dataset shift, causality, and shift-stable models in health AI[J]. Biostatistics. 2020 Apr 1;21(2):345-352.
[30] Challen R, Denny J, Pitt M, et al. Artificial intelligence, bias and clinical safety[J]. BMJ Qual Saf. 2019 Mar;28(3):231-237.
Issue
Section
License
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Transfer of Copyright and Permission to Reproduce Parts of Published Papers.
Authors retain the copyright for their published work. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties.
How to Cite
