arXiv:2507.15487v3 Announce Type: replace-cross 
Abstract: Magnetic Resonance Imaging (MRI) sequences provide rich spatial and frequency domain information, which is crucial for accurate lesion classification in medical imaging. However, effectively integrating multi-sequence MRI data for robust 3D lesion classification remains a challenge. In this paper, we propose DeSamba (Decoupled Spectral Adaptive Network and Mamba-Based Model), a novel framework designed to extract decoupled representations and adaptively fuse spatial and spectral features for lesion classification. DeSamba introduces a Decoupled Representation Learning Module (DRLM) that decouples features from different MRI sequences through self-reconstruction and cross-reconstruction, and a Spectral Adaptive Modulation Block (SAMB) within the proposed SAMNet, enabling dynamic fusion of spectral and spatial information based on lesion characteristics. We evaluate DeSamba on two clinically relevant 3D datasets. On a six-class spinal metastasis dataset (n=1,448), DeSamba achieves 62.10% Top-1 accuracy, 63.62% F1-score, 87.71% AUC, and 93.55% Top-3 accuracy on an external validation set (n=372), outperforming all state-of-the-art (SOTA) baselines. On a spondylitis dataset (n=251) involving a challenging binary classification task, DeSamba achieves 70.00%/64.52% accuracy and 74.75/73.88 AUC on internal and external validation sets, respectively. Ablation studies demonstrate that both DRLM and SAMB significantly contribute to overall performance, with over 10% relative improvement compared to the baseline. Our results highlight the potential of DeSamba as a generalizable and effective solution for 3D lesion classification in multi-sequence medical imaging.

تقدم الورقة البحثية DeSamba، وهو إطار مبتكر لتصنيف الآفات في التصوير بالرنين المغناطيسي ثلاثي الأبعاد متعدد التسلسلات. يتناول التحدي المتمثل في دمج بيانات الرنين المغناطيسي المتنوعة من خلال تقديم وحدة تعلم تمثيل مفصول (DRLM) تفصل الميزات من تسلسلات مختلفة من خلال إعادة البناء الذاتي وإعادة البناء المتبادل. بالإضافة إلى ذلك، يتضمن الإطار كتلة تعديل طيفي تكيفي (SAMB) ضمن SAMNet، مما يسمح بالدمج الديناميكي للمعلومات المكانية والطيفية بناءً على خصائص الآفات. يتم تقييم DeSamba على مجموعتين من البيانات ثلاثية الأبعاد ذات الصلة السريرية.

El artículo presenta DeSamba, un marco novedoso para la clasificación de lesiones en IRM 3D de múltiples secuencias. Aborda el desafío de integrar datos diversos de IRM mediante la introducción de un Módulo de Aprendizaje de Representación Desacoplada (DRLM) que separa características de diferentes secuencias a través de auto-reconstrucción y reconstrucción cruzada. Además, el marco incluye un Bloque de Modulación Adaptativa Espectral (SAMB) dentro del SAMNet, lo que permite la fusión dinámica de información espacial y espectral adaptada a las características de las lesiones. La evaluación de …

Cet article présente DeSamba, un cadre novateur pour la classification des lésions par IRM 3D multi-séquence. Il aborde le défi d'intégrer des données IRM diverses en introduisant un module d'apprentissage de représentation découplée (DRLM) qui sépare les caractéristiques de différentes séquences par auto-reconstruction et reconstruction croisée. De plus, le cadre comprend un bloc de modulation adaptative spectrale (SAMB) au sein du SAMNet, permettant une fusion dynamique des informations spatiales et spectrales adaptées aux caractéristiques des lésions. L'évaluation de DeSamba est réalisée su…

The paper presents DeSamba, a novel framework for 3D multi-sequence MRI lesion classification. It addresses the challenge of integrating diverse MRI data by introducing a Decoupled Representation Learning Module (DRLM) that separates features from various sequences through self-reconstruction and cross-reconstruction. Additionally, the framework includes a Spectral Adaptive Modulation Block (SAMB) within the SAMNet, allowing for dynamic fusion of spatial and spectral information tailored to lesion characteristics. The evaluation of DeSamba is conducted on two clinically relevant 3D datasets.

DeSamba: Decoupled Spectral Adaptive Framework for 3D Multi-Sequence MRI Lesion Classification

arXiv:2512.23130v2 Announce Type: replace 
Abstract: We present PathoSyn, a unified generative framework for Magnetic Resonance Imaging (MRI) image synthesis that reformulates imaging-pathology as a disentangled additive deviation on a stable anatomical manifold. Current generative models typically operate in the global pixel domain or rely on binary masks, these paradigms often suffer from feature entanglement, leading to corrupted anatomical substrates or structural discontinuities. PathoSyn addresses these limitations by decomposing the synthesis task into deterministic anatomical reconstruction and stochastic deviation modeling. Central to our framework is a Deviation-Space Diffusion Model designed to learn the conditional distribution of pathological residuals, thereby capturing localized intensity variations while preserving global structural integrity by construction. To ensure spatial coherence, the diffusion process is coupled with a seam-aware fusion strategy and an inference-time stabilization module, which collectively suppress boundary artifacts and produce high-fidelity internal lesion heterogeneity. PathoSyn provides a mathematically principled pipeline for generating high-fidelity patient-specific synthetic datasets, facilitating the development of robust diagnostic algorithms in low-data regimes. By allowing interpretable counterfactual disease progression modeling, the framework supports precision intervention planning and provides a controlled environment for benchmarking clinical decision-support systems. Quantitative and qualitative evaluations on tumor imaging benchmarks demonstrate that PathoSyn significantly outperforms holistic diffusion and mask-conditioned baselines in both perceptual realism and anatomical fidelity. The source code of this work will be made publicly available.

تم تقديم PathoSyn كإطار توليدي جديد لتوليد صور الرنين المغناطيسي (MRI)، مع التركيز على الانحرافات الإضافية المنفصلة على مانيفولد تشريحي مستقر لتحسين جودة الصور ودقة التشريح. تتناول هذه الطريقة القيود المفروضة على النماذج التوليدية الحالية التي غالبًا ما تؤدي إلى تشابك الميزات وانقطاع الهياكل.

PathoSyn se ha presentado como un nuevo marco generativo para la síntesis de imágenes de resonancia magnética (IRM), centrado en desviaciones aditivas disentrelazadas sobre un manifold anatómico estable para mejorar la calidad de las imágenes y la precisión anatómica. Este enfoque aborda las limitaciones de los modelos generativos existentes que a menudo conducen a un entrelazado de características y discontinuidades estructurales.

PathoSyn a été introduit comme un nouveau cadre génératif pour la synthèse d'images par résonance magnétique (IRM), se concentrant sur des déviations additives désentrelacées sur un manifold anatomique stable pour améliorer la qualité des images et l'exactitude anatomique. Cette approche répond aux limitations des modèles génératifs existants qui conduisent souvent à un entrelacement des caractéristiques et à des discontinuités structurelles.

PathoSyn has been introduced as a novel generative framework for Magnetic Resonance Imaging (MRI) synthesis, focusing on disentangled additive deviations on a stable anatomical manifold to improve image quality and anatomical accuracy. This approach addresses the limitations of existing generative models that often lead to feature entanglement and structural discontinuities.

DeSamba: Decoupled Spectral Adaptive Framework for 3D Multi-Sequence MRI Lesion Classification

Was this article worth reading? Share it

LucidQuery AI

Supametas.AI

Attentive AI

Deptho.ai

DeployMCP

FastML

Ready to build your own newsroom?