# Which Pretext Task Transfers? Self-Supervised Pretraining Objectives for Lung Ultrasound

arXiv: https://arxiv.org/abs/2609.16551 · HTML: https://mlsanity.com/paper/2609.16551/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 25 references that are in the map.
- First posted on arXiv 15 Sep 2026.

## Builds on (most cited first, 19 of 19)

- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [An Empirical Study of Training Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.02057/) (2021)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://mlsanity.com/paper/2203.12602/) (2022)
- [Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture](https://mlsanity.com/paper/2301.08243/) (2023)
- [V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning](https://mlsanity.com/paper/2506.09985/) (2025)
- [VideoMoCo: Contrastive Video Representation Learning with Temporally Adversarial Examples](https://mlsanity.com/paper/2103.05905/) (2021)
- [USFM: A Universal Ultrasound Foundation Model Generalized to Tasks and Organs towards Label Efficient Image Analysis](https://mlsanity.com/paper/2401.00153/) (2023)
- [Learning by Reconstruction Produces Uninformative Features For Perception](https://mlsanity.com/paper/2402.11337/) (2024)
- [Joint Embedding vs Reconstruction: Provable Benefits of Latent Space Prediction for Self Supervised Learning](https://mlsanity.com/paper/2505.12477/) (2025)
- [USCL: Pretraining Deep Ultrasound Image Diagnosis Model through Video Contrastive Representation Learning](https://mlsanity.com/paper/2011.13066/) (2020)
- [How JEPA Avoids Noisy Features: The Implicit Bias of Deep Linear Self Distillation Networks](https://mlsanity.com/paper/2407.03475/) (2024)
- [USF-MAE: Ultrasound Self-Supervised Foundation Model with Masked Autoencoding](https://mlsanity.com/paper/2510.22990/) (2025)
- [COVID-BLUeS -- A Prospective Study on the Value of AI in Lung Ultrasound Analysis](https://mlsanity.com/paper/2509.10556/) (2025)
- [Tasting the cake: evaluating self-supervised generalization on out-of-distribution multimodal MRI data](https://mlsanity.com/paper/2103.15914/) (2021)
- [Self-supervised Learning of Echocardiographic Video Representations via Online Cluster Distillation](https://mlsanity.com/paper/2506.11777/) (2025)
- [Self-Supervised Ultrasound-Video Segmentation with Feature Prediction and 3D Localised Loss](https://mlsanity.com/paper/2507.18424/) (2025)
- [Pretext Matters: An Empirical Study of SSL Methods in Medical Imaging](https://mlsanity.com/paper/2603.22649/) (2026)

## Abstract (arXiv)

Self-supervised learning (SSL) can reduce the need for labelled medical images, but the choice of pretext objective remains unclear for lung ultrasound (LUS). Contrastive learning, masked reconstruction, and joint-embedding predictive architectures (JEPA) differ in the space in which their targets are defined, yet existing ultrasound studies compare them under different corpora, backbones, and evaluation protocols. We compare these three objective families using the same encoder backbone, pretraining corpus, optimisation schedule, and frozen-evaluation protocol. Encoders are pretrained on COVID-BLUeS LUS videos and evaluated with linear, kNN, and attentive probes at 5%, 10%, 50%, and 100% label budgets. Evaluation is performed on POCUS using patient-level five-fold cross-validation and on the independently acquired Mendeley-Uganda dataset, which is excluded from both pretraining and probe fitting. At the full label budget under linear probing, VideoMAE and V-JEPA achieve 66.5 ± 13.1 and 65.4 ± 11.7 balanced accuracy on POCUS, while MoCo achieves 42.1 ± 1.2. On Mendeley-Uganda, the ranking reverses: MoCo performs best at 62.7 ± 1.0, followed by VideoMAE at 53.8 ± 2.8, while V-JEPA falls near chance at 35.1 ± 4.9. These results show that POCUS probe accuracy alone does not identify the objective that transfers best across datasets. We also outline planned representation-level analyses to examine this reversal. Code is publicly available at https://github.com/moeinheidari7829/LUSVideoSSL.