Abstract:Traditionally, segmentation algorithms require dense annotations for training, demanding significant annotation efforts, particularly within the 3D medical imaging field. Scribble-supervised learning emerges as a possible solution to this challenge, promising a reduction in annotation efforts when creating large-scale datasets. Recently, a plethora of methods for optimized learning from scribbles have been proposed, but have so far failed to position scribble annotation as a beneficial alternative. We relate this shortcoming to two major issues: 1) the complex nature of many methods which deeply ties them to the underlying segmentation model, thus preventing a migration to more powerful state-of-the-art models as the field progresses and 2) the lack of a systematic evaluation to validate consistent performance across the broader medical domain, resulting in a lack of trust when applying these methods to new segmentation problems. To address these issues, we propose a comprehensive scribble supervision benchmark consisting of seven datasets covering a diverse set of anatomies and pathologies imaged with varying modalities. We furthermore propose the systematic use of partial losses, i.e. losses that are only computed on annotated voxels. Contrary to most existing methods, these losses can be seamlessly integrated into state-of-the-art segmentation methods, enabling them to learn from scribble annotations while preserving their original loss formulations. Our evaluation using nnU-Net reveals that while most existing methods suffer from a lack of generalization, the proposed approach consistently delivers state-of-the-art performance. Thanks to its simplicity, our approach presents an embarrassingly simple yet effective solution to the challenges of scribble supervision. Source code as well as our extensive scribble benchmarking suite will be made publicly available upon publication.

Scribble-Based 3D Shape Segmentation via Weakly-Supervised Learning

Scribble-Supervised Segmentation of Aerial Building Footprints Using Adversarial Learning

3D Graph-S<SUP>2</SUP>Net: Shape-Aware Self-ensembling Network for Semi-supervised Segmentation with Bilateral Graph Convolution

Semi-Supervised 3D Shape Segmentation via Self Refining

Unsupervised 3D Shape Segmentation and Co-Segmentation Via Deep Learning.

3D Shape Segmentation Using Soft Density Peak Clustering and Semi-Supervised Learning

ZScribbleSeg: Zen and the Art of Scribble Supervised Medical Image Segmentation

Weakly Supervised 3D Instance Segmentation without Instance-level Annotations

Embarrassingly Simple Scribble Supervision for 3D Medical Segmentation

Learning Fine-Grained Segmentation of 3D Shapes without Part Labels

Weakly-Supervised Learning of Category-Specific 3D Object Shapes.

Semi-supervised 3D shape segmentation with multilevel consistency and part substitution

Weakly Supervised Medical Image Segmentation Via Superpixel-Guided Scribble Walking and Class-Wise Contrastive Regularization.

Scribble Hides Class: Promoting Scribble-Based Weakly-Supervised Semantic Segmentation with Its Class Label

Weakly‐supervised structural component segmentation via scribble annotations

Scribble-Based Medical Image Segmentation Via Shape Perturbation Consistency and Boundary Enhancement Constraint

Shape-aware Semi-supervised 3D Semantic Segmentation for Medical Images

Segmentation only uses sparse annotations: Unified weakly and semi-supervised learning in medical images

Size Aware Cross-shape Scribble Supervision for Medical Image Segmentation

One-Shot Weakly-Supervised Segmentation in 3D Medical Images.

Scribble-Supervised Medical Image Segmentation Via Dual-Branch Network and Dynamically Mixed Pseudo Labels Supervision