aiqbench analysis
This research addresses catastrophic forgetting in continual learning by analyzing how parameter updates across encoder-decoder depths affect performance in gynecological image segmentation. The study identifies that restricting training to bottleneck-adjacent regions significantly mitigates knowledge loss compared to updating shallower network layers, offering a structural approach to stable model adaptation under distribution shifts.
What to Preserve, Where to Adapt: A Depth-Wise Analysis of Forgetting in Continual Gynecological Image Segmentation
arXiv:2608.13660v1 Announce Type: cross Abstract: Medical image segmentation models are typically trained under the assumption that all data are available simultaneously. However, in clinical practice, datasets often arrive sequentially, requiring models to adapt continuously to evolving data distributions. We study this problem in gynecological image segmentation, where substantial heterogeneity across imaging modalities, anatomical structures, and annotation protocols creates a particularly challenging continual learning setting. Under these large distribution shifts, existing continual learning methods struggle to preserve previously learned knowledge, leading to catastrophic forgetting. To better understand forgetting in this setting, we investigate how different encoder--decoder regions influence segmentation performance and forgetting during continual gynecological segmentation. Through block-wise ablation analysis, we observe that ablating early encoder and late decoder regions results in the largest performance degradation, indicating that segmentation performance depends unevenly across the network hierarchy. Using controlled adaptation experiments, we further show that forgetting remains limited when updates are restricted to bottleneck-adjacent regions, but increases sharply once shallower encoders and decoders become trainable, even when only a small subset of parameters is updated. These findings suggest that forgetting in the encoder-decoder architecture is strongly influenced by where updates occur across network depth during continual learning. Full code and analysis pipelines will be made publicly available upon acceptance.
Abstract:Medical image segmentation models are typically trained under the assumption that all data are available simultaneously. However, in clinical practice, datasets often arrive sequentially, requiring models to adapt continuously to evolving data distributions. We study this problem in gynecological image segmentation, where substantial heterogeneity across imaging modalities, anatomical structures, and annotation protocols creates a particularly challenging continual learning setting. Under these large distribution shifts, existing continual learning methods struggle to preserve previously learned knowledge, leading to catastrophic forgetting. To better understand forgetting in this setting, we investigate how different encoder--decoder regions influence segmentation performance and forgetting during continual gynecological segmentation. Through block-wise ablation analysis, we observe that ablating early encoder and late decoder regions results in the largest performance degradation, indicating that segmentation performance depends unevenly across the network hierarchy. Using controlled adaptation experiments, we further show that forgetting remains limited when updates are restricted to bottleneck-adjacent regions, but increases sharply once shallower encoders and decoders become trainable, even when only a small subset of parameters is updated. These findings suggest that forgetting in the encoder-decoder architecture is strongly influenced by where updates occur across network depth during continual learning. Full code and analysis pipelines will be made publicly available upon acceptance.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2608.13660 [cs.CV]
(or arXiv:2608.13660v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.13660 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 302 words
Related coverage
Same Methods, Different Rankings: Trainable Depth as an Evaluation Variable in Continual Learning
arXiv · Oct 6, 2026
Reasoning-Preserving Fine-Tuning of Post-RL LLMs with Null-Basis LoRA
arXiv · Sep 24, 2026
Gated Adaptation for Continual Learning in Human Activity Recognition
arXiv · Jul 31, 2026
Unveiling Language Routing Isolation in Multilingual MoE Models for Interpretable Subnetwork Adaptation
arXiv · Sep 1, 2026