arXiv:2512.02188v1 Announce Type: new 
Abstract: While recent advances in deep learning for surgical scene segmentation have demonstrated promising results on single-centre and single-imaging modality data, these methods usually do not generalise to unseen distribution (i.e., from other centres) and unseen modalities. Current literature for tackling generalisation on out-of-distribution data and domain gaps due to modality changes has been widely researched but mostly for natural scene data. However, these methods cannot be directly applied to the surgical scenes due to limited visual cues and often extremely diverse scenarios compared to the natural scene data. Inspired by these works in natural scenes to push generalisability on OOD data, we hypothesise that exploiting the style and content information in the surgical scenes could minimise the appearances, making it less variable to sudden changes such as blood or imaging artefacts. This can be achieved by performing instance normalisation and feature covariance mapping techniques for robust and generalisable feature representations. Further, to eliminate the risk of removing salient feature representation associated with the objects of interest, we introduce a restitution module within the feature learning ResNet backbone that can enable the retention of useful task-relevant features. To tackle the lack of multiclass and multicentre data for surgical scene segmentation, we also provide a newly curated dataset that can be vital for addressing generalisability in this domain. Our proposed RobustSurg obtained nearly 23% improvement on the baseline DeepLabv3+ and from 10-32% improvement on the SOTA in terms of mean IoU score on an unseen centre HeiCholSeg dataset when trained on CholecSeg8K. Similarly, RobustSurg also obtained nearly 22% improvement over the baseline and nearly 11% improvement on a recent SOTA method for the target set of the EndoUDA polyp dataset.

دراسة جديدة بعنوان 'RobustSurg' تتناول تحديات تعميم المجال في تقسيم مشاهد الجراحة، مشددة على قيود الأساليب الحالية في التعلم العميق التي تواجه صعوبات مع التوزيعات والأنماط غير المرئية. تقترح الدراسة أن الاستفادة من معلومات الأسلوب والمحتوى في مشاهد الجراحة يمكن أن تقلل من التباين الناتج عن عوامل مثل الدم أو عيوب التصوير.

Un nuevo estudio titulado 'RobustSurg' aborda los desafíos de la generalización de dominio en la segmentación de escenas quirúrgicas, destacando las limitaciones de los métodos de aprendizaje profundo actuales que luchan con distribuciones y modalidades no vistas. La investigación sugiere que aprovechar la información de estilo y contenido en las escenas quirúrgicas puede reducir la variabilidad causada por factores como la sangre o los artefactos de imagen.

Une nouvelle étude intitulée 'RobustSurg' aborde les défis de la généralisation de domaine dans la segmentation de scènes chirurgicales, soulignant les limites des méthodes d'apprentissage profond actuelles qui peinent avec des distributions et des modalités non vues. La recherche suggère que l'exploitation des informations de style et de contenu dans les scènes chirurgicales peut réduire la variabilité causée par des facteurs tels que le sang ou les artefacts d'imagerie.

A new study titled 'RobustSurg' addresses the challenges of domain generalisation in surgical scene segmentation, highlighting the limitations of current deep learning methods that struggle with unseen distributions and modalities. The research suggests that leveraging style and content information in surgical scenes can reduce variability caused by factors like blood or imaging artefacts.

RobustSurg: Tackling domain generalisation for out-of-distribution surgical scene segmentation

arXiv:2512.03014v1 Announce Type: new 
Abstract: When applied sequentially to video, frame-based networks often exhibit temporal inconsistency - for example, outputs that flicker between frames. This problem is amplified when the network inputs contain time-varying corruptions. In this work, we introduce a general approach for adapting frame-based models for stable and robust inference on video. We describe a class of stability adapters that can be inserted into virtually any architecture and a resource-efficient training process that can be performed with a frozen base network. We introduce a unified conceptual framework for describing temporal stability and corruption robustness, centered on a proposed accuracy-stability-robustness loss. By analyzing the theoretical properties of this loss, we identify the conditions where it produces well-behaved stabilizer training. Our experiments validate our approach on several vision tasks including denoising (NAFNet), image enhancement (HDRNet), monocular depth (Depth Anything v2), and semantic segmentation (DeepLabv3+). Our method improves temporal stability and robustness against a range of image corruptions (including compression artifacts, noise, and adverse weather), while preserving or improving the quality of predictions.

تم تقديم نهج جديد لتثبيت الشبكات القائمة على الإطارات في الفيديو، حيث يعالج هذا النهج التناقضات الزمنية التي تظهر غالبًا في مخرجات الفيديو. تتضمن هذه الطريقة استخدام محولات الاستقرار التي يمكن دمجها في الهياكل الحالية، مما يسمح باستنتاج موثوق حتى في وجود تدهورات متغيرة زمنياً.

Se ha introducido un nuevo enfoque para estabilizar redes de video basadas en fotogramas, abordando las inconsistencias temporales que a menudo surgen en las salidas de video. Este método implica el uso de adaptadores de estabilidad que se pueden integrar en arquitecturas existentes, permitiendo inferencias robustas incluso en presencia de corrupciones variables en el tiempo.

Une nouvelle approche a été introduite pour stabiliser les réseaux vidéo basés sur des images, abordant les incohérences temporelles qui surviennent souvent dans les sorties vidéo. Cette méthode implique l'utilisation d'adaptateurs de stabilité pouvant être intégrés dans des architectures existantes, permettant une inférence robuste même en présence de corruptions variables dans le temps.

A new approach has been introduced for stabilizing frame-based video networks, addressing the temporal inconsistencies that often arise in video outputs. This method involves the use of stability adapters that can be integrated into existing architectures, allowing for robust inference even in the presence of time-varying corruptions.

RobustSurg: Tackling domain generalisation for out-of-distribution surgical scene segmentation

Was this article worth reading? Share it

Dubsmart LLC

Augmeta

Gestell