Fit Pixels, Get Labels: Meta-learned Implicit Networks for Image Segmentation

2025-10-05

AI Summary

Implicit Neural Representations (INRs) are inherently ill-suited for medical image segmentation due to their difficulty in capturing semantic structures. To address this, we propose MetaSeg—the first lightweight segmentation framework that deeply integrates meta-learning with INRs. MetaSeg employs a unified implicit network that jointly regresses pixel intensities and semantic labels, enabling end-to-end segmentation decoding via fine-tuning in just a few optimization steps. Its core innovation lies in meta-optimization to learn a task-agnostic initialization, allowing rapid adaptation to new images with minimal supervision. Evaluated on 2D and 3D brain MRI datasets, MetaSeg achieves Dice scores comparable to U-Net while reducing model parameters by 90% (retaining only 10% of U-Net’s parameter count), significantly improving computational efficiency and scalability. This work establishes a novel, efficient paradigm for medical image segmentation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Implicit neural representations (INRs) have achieved remarkable successes in learning expressive yet compact signal representations. However, they are not naturally amenable to predictive tasks such as segmentation, where they must learn semantic structures over a distribution of signals. In this study, we introduce MetaSeg, a meta-learning framework to train INRs for medical image segmentation. MetaSeg uses an underlying INR that simultaneously predicts per pixel intensity values and class labels. It then uses a meta-learning procedure to find optimal initial parameters for this INR over a training dataset of images and segmentation maps, such that the INR can simply be fine-tuned to fit pixels of an unseen test image, and automatically decode its class labels. We evaluated MetaSeg on 2D and 3D brain MRI segmentation tasks and report Dice scores comparable to commonly used U-Net models, but with $90%$ fewer parameters. MetaSeg offers a fresh, scalable alternative to traditional resource-heavy architectures such as U-Nets and vision transformers for medical image segmentation. Our project is available at https://kushalvyas.github.io/metaseg.html .

Problem

• Meta-learned implicit networks for medical image segmentation
• Training INRs to predict pixel intensities and class labels
• Finding optimal initial parameters for fast fine-tuning on unseen images

Innovation

• Meta-learned implicit networks for segmentation
• Simultaneously predicts pixel intensity and labels
• Fine-tunes with 90% fewer parameters than U-Net