arXiv:2510.22004v1 Announce Type: new 
Abstract: In recent years, diffusion models have demonstrated remarkable success in high-fidelity image synthesis. However, fine-tuning these models for specialized domains, such as medical imaging, remains challenging due to limited domain-specific data and the high computational cost of full model adaptation. In this paper, we introduce Lite-Diff (Lightweight Diffusion Model Adaptation), a novel finetuning approach that integrates lightweight adaptation layers into a frozen diffusion U-Net while enhancing training with a latent morphological autoencoder (for domain-specific latent consistency) and a pixel level discriminator(for adversarial alignment). By freezing weights of the base model and optimizing only small residual adapter modules, LiteDiff significantly reduces the computational overhead and mitigates overfitting, even in minimal-data settings. Additionally, we conduct ablation studies to analyze the effects of selectively integrating adaptation layers in different U-Net blocks, revealing an optimal balance between efficiency and performance. Experiments on three chest X-ray datasets - (1) Kaggle Chest X-Ray Pneumonia, (2) NIH Chest X-ray14 and (3) VinBigData Chest X_ray demonstrate that LiteDiff achieves superior adaptation efficiency compared to naive full fine-tuning. Our framework provides a promising direction for transfer learning in diffusion models, facilitating their deployment in diverse low data domains.

تقديم Lite-Diff يمثل تقدمًا كبيرًا في مجال توليد الصور، خاصة في المجالات المتخصصة مثل التصوير الطبي. تعالج هذه الطريقة الجديدة للتكيف الخفيف التحديات المتعلقة بضبط نماذج الانتشار، التي كانت محدودة بسبب الحاجة إلى بيانات محددة للمجال وتكاليف حسابية مرتفعة. من خلال تسهيل تكيف هذه النماذج، يمكن أن يحسن Lite-Diff جودة وكفاءة عمليات التصوير الطبي، مما يعود بالنفع في النهاية على المتخصصين في الرعاية الصحية والمرضى.

La introducción de Lite-Diff marca un avance significativo en el campo de la síntesis de imágenes, especialmente en áreas especializadas como la imagen médica. Este nuevo método de adaptación ligera aborda los desafíos de ajustar modelos de difusión, que han estado limitados por la necesidad de datos específicos del dominio y altos costos computacionales. Al facilitar la adaptación de estos modelos, Lite-Diff podría mejorar la calidad y eficiencia de los procesos de imagen médica, beneficiando en última instancia a los profesionales de la salud y a los pacientes.

L'introduction de Lite-Diff représente une avancée significative dans le domaine de la synthèse d'images, en particulier pour des domaines spécialisés comme l'imagerie médicale. Cette nouvelle méthode d'adaptation légère répond aux défis de l'ajustement des modèles de diffusion, qui ont été limités par le besoin de données spécifiques au domaine et des coûts computationnels élevés. En facilitant l'adaptation de ces modèles, Lite-Diff pourrait améliorer la qualité et l'efficacité des processus d'imagerie médicale, bénéficiant ainsi aux professionnels de la santé et aux patients.

The introduction of Lite-Diff marks a significant advancement in the field of image synthesis, particularly for specialized areas like medical imaging. This new lightweight adaptation method addresses the challenges of fine-tuning diffusion models, which have been limited by the need for extensive domain-specific data and high computational costs. By making it easier to adapt these models, Lite-Diff could enhance the quality and efficiency of medical imaging processes, ultimately benefiting healthcare professionals and patients alike.

LiteDiff

arXiv:2503.06717v2 Announce Type: replace 
Abstract: Interactive segmentation uses real-time user inputs, such as mouse clicks, to iteratively refine model predictions. Although not originally designed to address distribution shifts, this paradigm naturally lends itself to such challenges. In medical imaging, where distribution shifts are common, interactive methods can use user inputs to guide models towards improved predictions. Moreover, once a model is deployed, user corrections can be used to adapt the network parameters to the new data distribution, mitigating distribution shift. Based on these insights, we aim to develop a practical, effective method for improving the adaptive capabilities of interactive segmentation models to new data distributions in medical imaging. Firstly, we found that strengthening the model's responsiveness to clicks is important for the initial training process. Moreover, we show that by treating the post-interaction user-refined model output as pseudo-ground-truth, we can design a lean, practical online adaptation method that enables a model to learn effectively across sequential test images. The framework includes two components: (i) a Post-Interaction adaptation process, updating the model after the user has completed interactive refinement of an image, and (ii) a Mid-Interaction adaptation process, updating incrementally after each click. Both processes include a Click-Centered Gaussian loss that strengthens the model's reaction to clicks and enhances focus on user-guided, clinically relevant regions. Experiments on 5 fundus and 4 brain-MRI databases show that our approach consistently outperforms existing methods under diverse distribution shifts, including unseen imaging modalities and pathologies. Code and pretrained models will be released upon publication.

قدمت دراسة جديدة طريقة للتكيف عبر الإنترنت لنماذج التقسيم التفاعلي في التصوير الطبي، تركز على التعامل مع تغييرات التوزيع من خلال مدخلات المستخدم في الوقت الحقيقي. يعزز هذا النهج توقعات النموذج من خلال السماح بتصحيحات المستخدمين لتوجيه النموذج، مما يزيد من قابليته للتكيف مع توزيعات البيانات الجديدة.

Un nuevo estudio ha presentado un método de adaptación en línea para modelos de segmentación interactiva en imágenes médicas, centrado en manejar cambios de distribución mediante entradas de usuario en tiempo real. Este enfoque mejora las predicciones del modelo al permitir que las correcciones de los usuarios guíen al modelo, aumentando así su adaptabilidad a nuevas distribuciones de datos.

Une nouvelle étude a introduit une méthode d'adaptation en ligne pour les modèles de segmentation interactive en imagerie médicale, axée sur la gestion des changements de distribution grâce à des entrées utilisateur en temps réel. Cette approche améliore les prédictions du modèle en permettant aux corrections des utilisateurs de guider le modèle, améliorant ainsi son adaptabilité aux nouvelles distributions de données.

A new study has introduced an online adaptation method for interactive segmentation models in medical imaging, focusing on handling distribution shifts through real-time user inputs. This approach enhances model predictions by allowing user corrections to guide the model, thereby improving its adaptability to new data distributions.

You Point, I Learn: Online Adaptation of Interactive Segmentation Models for Handling Distribution Shifts in Medical Imaging

arXiv:2512.03477v1 Announce Type: cross 
Abstract: Vision-language models achieve expert-level performance on medical imaging tasks but exhibit significant diagnostic accuracy disparities across demographic groups. We introduce fairness-aware Low-Rank Adaptation for medical VLMs, combining parameter efficiency with explicit fairness optimization. Our key algorithmic contribution is a differentiable MaxAccGap loss that enables end-to-end optimization of accuracy parity across demographic groups. We propose three methods: FR-LoRA integrates MaxAccGap regularization into the training objective, GR-LoRA applies inverse frequency weighting to balance gradient contributions, and Hybrid-LoRA combines both mechanisms.Evaluated on 10,000 glaucoma fundus images, GR-LoRA reduces diagnostic accuracy disparities by 69% while maintaining 53.15% overall accuracy. Ablation studies reveal that strong regularization strength achieves optimal fairness with minimal accuracy trade-off, and race-specific optimization yields 60% disparity reduction. Our approach requires only 0.24% trainable parameters, enabling practical deployment of fair medical AI in resource-constrained healthcare settings.

قدمت دراسة حديثة تقنيات تكييف منخفضة الرتبة واعية بالعدالة لنماذج الرؤية-اللغة (VLM) تهدف إلى تحسين دقة التشخيص في التصوير الطبي، وخاصة لتشخيص الزرق. تركز الطرق المقترحة، بما في ذلك FR-LoRA و GR-LoRA، على تقليل الفجوات في دقة التشخيص بين المجموعات السكانية مع الحفاظ على الأداء العام. أظهرت التقييمات على 10,000 صورة قاع زرقية انخفاضًا كبيرًا في الفجوات التشخيصية بنسبة 69% باستخدام GR-LoRA.

Un estudio reciente ha introducido técnicas de adaptación de bajo rango conscientes de la equidad para modelos de visión-lenguaje (VLM) con el objetivo de mejorar la precisión diagnóstica en imágenes médicas, específicamente para el diagnóstico de glaucoma. Los métodos propuestos, incluidos FR-LoRA y GR-LoRA, se centran en reducir las disparidades de precisión entre grupos demográficos mientras mantienen el rendimiento general. Las evaluaciones en 10,000 imágenes de fondo de glaucoma demostraron una reducción significativa de las disparidades diagnósticas del 69% con GR-LoRA.

Une étude récente a introduit des techniques d'adaptation à faible rang conscientes de l'équité pour les modèles de vision-langage (VLM) visant à améliorer la précision diagnostique en imagerie médicale, spécifiquement pour le diagnostic du glaucome. Les méthodes proposées, y compris FR-LoRA et GR-LoRA, se concentrent sur la réduction des disparités de précision entre les groupes démographiques tout en maintenant la performance globale. Les évaluations sur 10 000 images de fondus de glaucome ont montré une réduction significative des disparités diagnostiques de 69 % avec GR-LoRA.

A recent study has introduced fairness-aware Low-Rank Adaptation techniques for vision-language models (VLMs) aimed at improving diagnostic accuracy in medical imaging, specifically for glaucoma diagnosis. The proposed methods, including FR-LoRA and GR-LoRA, focus on reducing accuracy disparities across demographic groups while maintaining overall performance. Evaluations on 10,000 glaucoma fundus images demonstrated a significant reduction in diagnostic disparities by 69% with GR-LoRA.

LiteDiff

Was this article worth reading? Share it

LucidQuery AI

Blunge

AIPortalX