Understanding Task Aggregation for Generalizable Ultrasound Foundation Models

• URL: http://arxiv.org/abs/2603.18123v2
• Date: Fri, 20 Mar 2026 09:24:34 GMT
• Title: Understanding Task Aggregation for Generalizable Ultrasound Foundation Models
• Abstract summary: We introduce M2DINO, a multi-organ, multi-task framework built on DINOv3 with task-conditioned Mixture-of-Experts blocks for adaptive capacity allocation.<n>We evaluate 27 ultrasound tasks spanning segmentation, classification, detection, and regression under three paradigms: task-specific, clinically-grouped, and all-task unified training.<n>Our results show that aggregation effectiveness depends strongly on training data scale.
• Score: 4.4775919772138515
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Foundation models promise to unify multiple clinical tasks within a single framework, but recent ultrasound studies report that unified models can underperform task-specific baselines. We hypothesize that this degradation arises not from model capacity limitations, but from task aggregation strategies that ignore interactions between task heterogeneity and available training data scale. In this work, we systematically analyze when heterogeneous ultrasound tasks can be jointly learned without performance loss, establishing practical criteria for task aggregation in unified clinical imaging models. We introduce M2DINO, a multi-organ, multi-task framework built on DINOv3 with task-conditioned Mixture-of-Experts blocks for adaptive capacity allocation. We systematically evaluate 27 ultrasound tasks spanning segmentation, classification, detection, and regression under three paradigms: task-specific, clinically-grouped, and all-task unified training. Our results show that aggregation effectiveness depends strongly on training data scale. While clinically-grouped training can improve performance in data-rich settings, it may induce substantial negative transfer in low-data settings. In contrast, all-task unified training exhibits more consistent performance across clinical groups. We further observe that task sensitivity varies by task type in our experiments: segmentation shows the largest performance drops compared with regression and classification. These findings provide practical guidance for ultrasound foundation models, emphasizing that aggregation strategies should jointly consider training data availability and task characteristics rather than relying on clinical taxonomy alone.

Related papers

• Learning Multi-Indicator Weights for Data Selection: A Joint Task-Model Adaptation Framework with Efficient Proxies [50.39041754816285]
  We propose a framework for learning multi-validation weights that jointly adapts data selection to both the downstream task and the specific model.<n>Our approach achieves performance comparable to or exceeding full-dataset tuning while using only 30% of the training samples on GSM8K.
  arXiv  Detail & Related papers  (2026-05-10T17:30:16Z)
• Towards Better Ultrasound Video Segmentation Foundation Model: An Empirical study on SAM2 Finetuning from Data Perspective [0.7629717457706325]
  We present a data-centric investigation of SAM2 adaptation for ultrasound video segmentation.<n>We analyze how training-set size, video duration, and augmentation schemes affect adaptation performance.
  arXiv  Detail & Related papers  (2025-11-07T21:45:18Z)
• MEJO: MLLM-Engaged Surgical Triplet Recognition via Inter- and Intra-Task Joint Optimization [52.149337961205624]
  We propose a framework that empowers both inter- and intra-task optimization for surgical triplet recognition.<n>For inter-task optimization, we introduce the Shared-Specific-Disentangled (S$2$D) learning scheme that decomposes representations into task-shared and task-specific components.<n>For intra-task optimization conflicts, we develop a Coordinated Gradient Learning (CGL) strategy, which dissects and rebalances the positive-negative ambiguities.
  arXiv  Detail & Related papers  (2025-09-16T09:48:52Z)
• Repurposing Foundation Model for Generalizable Medical Time Series Classification [16.21546283978257]
  FORMED is a framework for repurposing a backbone foundation model to enable highly generalizable MedTS classification on unseen datasets.<n>We evaluate FORMED on 5 diverse MedTS datasets, benchmarking against 11 Task-Specific Models (TSM) and 4 Task-Specific Adaptation (TSA) methods.<n>Our results demonstrate FORMED's dominant performance, achieving up to 35% absolute improvement in F1-score (on ADFTD dataset) over specialized baselines.
  arXiv  Detail & Related papers  (2024-10-03T23:50:04Z)
• Hybrid Classification-Regression Adaptive Loss for Dense Object Detection [19.180514552400883]
  We propose a Hybrid Classification-Regression Adaptive Loss, termed as HCRAL. We introduce the Residual of Classification and IoU (RCI) module for cross-task supervision, addressing task inconsistencies, and the Conditioning Factor (CF) to focus on difficult-to-train samples within each task. We also introduce a new strategy named Expanded Adaptive Training Sample Selection (EATSS) to provide additional samples that exhibit classification and regression inconsistencies.
  arXiv  Detail & Related papers  (2024-08-30T10:31:39Z)
• Boosting Few-Shot Learning with Disentangled Self-Supervised Learning and Meta-Learning for Medical Image Classification [8.975676404678374]
  We present a strategy for improving the performance and generalization capabilities of models trained in low-data regimes. The proposed method starts with a pre-training phase, where features learned in a self-supervised learning setting are disentangled to improve the robustness of the representations for downstream tasks. We then introduce a meta-fine-tuning step, leveraging related classes between meta-training and meta-testing phases but varying the level.
  arXiv  Detail & Related papers  (2024-03-26T09:36:20Z)
• Improving GANs with A Dynamic Discriminator [106.54552336711997]
  We argue that a discriminator with an on-the-fly adjustment on its capacity can better accommodate such a time-varying task. A comprehensive empirical study confirms that the proposed training strategy, termed as DynamicD, improves the synthesis performance without incurring any additional cost or training objectives.
  arXiv  Detail & Related papers  (2022-09-20T17:57:33Z)
• Meta-learning One-class Classifiers with Eigenvalue Solvers for Supervised Anomaly Detection [55.888835686183995]
  We propose a neural network-based meta-learning method for supervised anomaly detection. We experimentally demonstrate that the proposed method achieves better performance than existing anomaly detection and few-shot learning methods.
  arXiv  Detail & Related papers  (2021-03-01T01:43:04Z)
• Self-supervised driven consistency training for annotation efficient histopathology image analysis [13.005873872821066]
  Training a neural network with a large labeled dataset is still a dominant paradigm in computational histopathology. We propose a self-supervised pretext task that harnesses the underlying multi-resolution contextual cues in histology whole-slide images to learn a powerful supervisory signal for unsupervised representation learning. We also propose a new teacher-student semi-supervised consistency paradigm that learns to effectively transfer the pretrained representations to downstream tasks based on prediction consistency with the task-specific un-labeled data.
  arXiv  Detail & Related papers  (2021-02-07T19:46:21Z)
• Select-ProtoNet: Learning to Select for Few-Shot Disease Subtype Prediction [55.94378672172967]
  We focus on few-shot disease subtype prediction problem, identifying subgroups of similar patients. We introduce meta learning techniques to develop a new model, which can extract the common experience or knowledge from interrelated clinical tasks. Our new model is built upon a carefully designed meta-learner, called Prototypical Network, that is a simple yet effective meta learning machine for few-shot image classification.
  arXiv  Detail & Related papers  (2020-09-02T02:50:30Z)
• Task-Feature Collaborative Learning with Application to Personalized Attribute Prediction [166.87111665908333]
  We propose a novel multi-task learning method called Task-Feature Collaborative Learning (TFCL) Specifically, we first propose a base model with a heterogeneous block-diagonal structure regularizer to leverage the collaborative grouping of features and tasks. As a practical extension, we extend the base model by allowing overlapping features and differentiating the hard tasks.
  arXiv  Detail & Related papers  (2020-04-29T02:32:04Z)
• Learning What Makes a Difference from Counterfactual Examples and Gradient Supervision [57.14468881854616]
  We propose an auxiliary training objective that improves the generalization capabilities of neural networks. We use pairs of minimally-different examples with different labels, a.k.a counterfactual or contrasting examples, which provide a signal indicative of the underlying causal structure of the task. Models trained with this technique demonstrate improved performance on out-of-distribution test sets.
  arXiv  Detail & Related papers  (2020-04-20T02:47:49Z)

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.