Benchmarking Ultrasound Foundation Models for Fetal Plane Classification

• URL: http://arxiv.org/abs/2605.27796v1
• Date: Wed, 27 May 2026 00:32:40 GMT
• Title: Benchmarking Ultrasound Foundation Models for Fetal Plane Classification
• Abstract summary: This work presents a benchmark of ultrasound-specific foundation models (FMs) for fetal plane classification.<n>FMs use a large number of ultrasound images to learn transferable representations that can generalize with limited labeled data.<n>We trained all models under two complementary settings: full fine-tuning and linear probing with a frozen encoder.
• Score: 3.35969908228128
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Ultrasound is widely used in obstetric care due to its safety, accessibility, and real-time imaging. However, interpretation remains operator-dependent and susceptible to noise and artifacts. Deep learning models have shown strong performance to solve these problem, but they typically require large annotated datasets that are difficult to obtain in clinical ultrasound. Foundation models (FMs) offer an alternative, using a large number of ultrasound images to learn transferable representations that can generalize with limited labeled data. This work presents a comprehensive benchmark of ultrasound-specific FMs for fetal plane classification. We evaluated four ultrasound FMs (USFM, MOFO, UltraSAM, FetalCLIP) against two CNN baselines (ResNet50, EfficientNet-V2) and a ViT (DINOv3) pretrained on natural images. We trained all models under two complementary settings: full fine-tuning and linear probing with a frozen encoder. All models were trained using 5-fold patient-level cross-validation on a Spanish fetal ultrasound dataset and tested on both in-domain data and an external African cohort to assess cross-population generalization. We found that FetalCLIP achieved the best results in the linear probing setting (F1 = 0.9261 for in-domain, F1 = 0.9731 for out-of-domain), while USFM performed best in the full fine-tuning setting (F1 = 0.9476 for in-domain, F1 = 0.9515 for out-of-domain). MOFO and UltraSAM degraded most in both settings, underperforming natural image pretrained models in some cases. These findings highlight how the choice of pretrained model strongly affects fetal plane classification performance, since different pretraining objectives lead to different levels of transferability.

Related papers

• A Foundational EDM2-Based Generative Model for High-Resolution Synthetic Fetal Ultrasound Imaging from Open Datasets [13.027245104199912]
  We propose a high-resolution fetal ultrasound synthesis framework based on the EDM2 diffusion architecture.<n>Our method achieved improved image quality with lower FID scores and enhanced downstream fetal plane classification, reaching 93.36% ensemble accuracy after fine-tuning.
  arXiv  Detail & Related papers  (2026-08-05T23:39:48Z)
• SonoCLIP: Mask-Guided Region-Aware Vision-Language Pretraining for Fetal Ultrasound Analysis [51.81845420301345]
  SonoCLIP is a vision-language foundation model for fetal ultrasound analysis.<n>It integrates segmentation masks as mask-channel visual prompts within the vision encoder.<n>It achieves superior zero-shot transfer performance under both global and mask-guided inference.
  arXiv  Detail & Related papers  (2026-06-28T20:04:49Z)
• Challenging DINOv3 Foundation Model under Low Inter-Class Variability: A Case Study on Fetal Brain Ultrasound [4.07447364754644]
  This study provides the first comprehensive evaluation of foundation models in fetal ultrasound (US) imaging under low interclass variability conditions.<n>We focus on fetal brain standard planes--transthalamic (TT), transventricular (TV), and transcerebellar (TC)--which exhibit highly overlapping anatomical features.<n>Models pretrained on fetal ultrasound data consistently outperformed those on natural images, with weighted F1-score improvements of up to 20 percent.
  arXiv  Detail & Related papers  (2025-11-01T13:37:22Z)
• USF-MAE: Ultrasound Self-Supervised Foundation Model with Masked Autoencoding [0.205246094017924]
  We introduce the Ultrasound Self-Supervised Foundation Model with Masked Autoencoding (USF-MAE)<n>USF-MAE is the first large-scale self-supervised MAE framework pretrained exclusively on ultrasound data.<n>The model was pre-trained on 370,000 2D and 3D ultrasound images curated from 46 open-source datasets.
  arXiv  Detail & Related papers  (2025-10-27T04:16:43Z)
• Epistemic-aware Vision-Language Foundation Model for Fetal Ultrasound Interpretation [83.02147613524032]
  We introduce FetalMind, a medical AI system tailored to fetal ultrasound for both report generation and diagnosis.<n>We propose Salient Epistemic Disentanglement (SED), which injects an expert-curated bipartite graph into the model to decouple view-disease associations.<n>FetalMind outperforms open- and closed-source baselines across all gestational stages, achieving +14% average gains and +61.2% higher accuracy on critical conditions.
  arXiv  Detail & Related papers  (2025-10-14T19:57:03Z)
• Advancing Fetal Ultrasound Image Quality Assessment in Low-Resource Settings [3.982826074217475]
  We leverage FetalCLIP, a vision-caption model pretrained on a curated dataset of over 210,000 fetal ultrasound image-language pairs.<n>We introduce an IQA model adapted from FetalCLIP using Low-Rank Adaptation (LoRA), and evaluate it on the ACOUS-AI dataset.<n>We show that an adapted segmentation model, when repurposed for classification, further improves performance, achieving an F1 score of 0.771.
  arXiv  Detail & Related papers  (2025-07-30T16:09:29Z)
• The Efficacy of Semantics-Preserving Transformations in Self-Supervised Learning for Medical Ultrasound [60.80780313225093]
  This study systematically investigated the impact of data augmentation and preprocessing strategies in self-supervised learning for lung ultrasound.<n>Three data augmentation pipelines were assessed: a baseline pipeline commonly used across imaging domains, a novel semantic-preserving pipeline designed for ultrasound, and a distilled set of the most effective transformations from both pipelines.
  arXiv  Detail & Related papers  (2025-04-10T16:26:47Z)
• FetalFlex: Anatomy-Guided Diffusion Model for Flexible Control on Fetal Ultrasound Image Synthesis [16.640351512021017]
  We introduce a Flexible Fetal US image generation framework (FetalFlex) to address these challenges.<n>FetalFlex incorporates a pre-alignment module to enhance controllability and introduces a repaint strategy to ensure consistent texture and appearance.<n> Experiments on multi-center datasets demonstrate that FetalFlex achieved state-of-the-art performance across multiple image quality metrics.
  arXiv  Detail & Related papers  (2025-03-19T05:16:19Z)
• Domain Transfer Through Image-to-Image Translation for Uncertainty-Aware Prostate Cancer Classification [42.75911994044675]
  We present a novel approach for unpaired image-to-image translation of prostate MRIs and an uncertainty-aware training approach for classifying clinically significant PCa. Our approach involves a novel pipeline for translating unpaired 3.0T multi-parametric prostate MRIs to 1.5T, thereby augmenting the available training data. Our experiments demonstrate that the proposed method significantly improves the Area Under ROC Curve (AUC) by over 20% compared to the previous work.
  arXiv  Detail & Related papers  (2023-07-02T05:26:54Z)
• Performance of GAN-based augmentation for deep learning COVID-19 image classification [57.1795052451257]
  The biggest challenge in the application of deep learning to the medical domain is the availability of training data. Data augmentation is a typical methodology used in machine learning when confronted with a limited data set. In this work, a StyleGAN2-ADA model of Generative Adversarial Networks is trained on the limited COVID-19 chest X-ray image set.
  arXiv  Detail & Related papers  (2023-04-18T15:39:58Z)
• Significantly improving zero-shot X-ray pathology classification via fine-tuning pre-trained image-text encoders [50.689585476660554]
  We propose a new fine-tuning strategy that includes positive-pair loss relaxation and random sentence sampling. Our approach consistently improves overall zero-shot pathology classification across four chest X-ray datasets and three pre-trained models.
  arXiv  Detail & Related papers  (2022-12-14T06:04:18Z)
• Data-Efficient Vision Transformers for Multi-Label Disease Classification on Chest Radiographs [55.78588835407174]
  Vision Transformers (ViTs) have not been applied to this task despite their high classification performance on generic images. ViTs do not rely on convolutions but on patch-based self-attention and in contrast to CNNs, no prior knowledge of local connectivity is present. Our results show that while the performance between ViTs and CNNs is on par with a small benefit for ViTs, DeiTs outperform the former if a reasonably large data set is available for training.
  arXiv  Detail & Related papers  (2022-08-17T09:07:45Z)
• Vision Transformers for femur fracture classification [59.99241204074268]
  The Vision Transformer (ViT) was able to correctly predict 83% of the test images. Good results were obtained in sub-fractures with the largest and richest dataset ever.
  arXiv  Detail & Related papers  (2021-08-07T10:12:42Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.