arXiv:2511.15067v1 Announce Type: new 
Abstract: Precise prognostic stratification of colorectal cancer (CRC) remains a major clinical challenge due to its high heterogeneity. The conventional TNM staging system is inadequate for personalized medicine. We aimed to develop and validate a novel multiple instance learning model TDAM-CRC using histopathological whole-slide images for accurate prognostic prediction and to uncover its underlying molecular mechanisms. We trained the model on the TCGA discovery cohort (n=581), validated it in an independent external cohort (n=1031), and further we integrated multi-omics data to improve model interpretability and identify novel prognostic biomarkers. The results demonstrated that the TDAM-CRC achieved robust risk stratification in both cohorts. Its predictive performance significantly outperformed the conventional clinical staging system and multiple state-of-the-art models. The TDAM-CRC risk score was confirmed as an independent prognostic factor in multivariable analysis. Multi-omics analysis revealed that the high-risk subtype is closely associated with metabolic reprogramming and an immunosuppressive tumor microenvironment. Through interaction network analysis, we identified and validated Mitochondrial Ribosomal Protein L37 (MRPL37) as a key hub gene linking deep pathomic features to clinical prognosis. We found that high expression of MRPL37, driven by promoter hypomethylation, serves as an independent biomarker of favorable prognosis. Finally, we constructed a nomogram incorporating the TDAM-CRC risk score and clinical factors to provide a precise and interpretable clinical decision-making tool for CRC patients. Our AI-driven pathological model TDAM-CRC provides a robust tool for improved CRC risk stratification, reveals new molecular targets, and facilitates personalized clinical decision-making.

تم تطوير نموذج جديد للتعلم عبر الحالات المتعددة، TDAM-CRC، لتحسين التنبؤات التشخيصية في سرطان القولون والمستقيم (CRC) باستخدام صور الأنسجة الكاملة. تم تدريب هذا النموذج على مجموعة من 581 مريضًا وتم التحقق من صحته في مجموعة مستقلة تضم 1,031 مريضًا، مما أظهر أداءً تنبؤيًا متفوقًا مقارنةً بأنظمة التصنيف السريرية التقليدية. كما دمجت الدراسة بيانات متعددة الأوميات لتحسين قابلية تفسير النموذج وتحديد مؤشرات حيوية تنبؤية جديدة.

Se ha desarrollado un nuevo modelo de aprendizaje por instancias múltiples, TDAM-CRC, para mejorar la predicción pronóstica en el cáncer colorrectal (CRC) utilizando imágenes histopatológicas de diapositivas completas. Este modelo fue entrenado en una cohorte de 581 pacientes y validado en una cohorte externa independiente de 1,031 pacientes, demostrando un rendimiento predictivo superior en comparación con los sistemas de estadificación clínica convencionales. El estudio también integró datos multi-ómicos para mejorar la interpretabilidad del modelo e identificar nuevos biomarcadores pronósti…

Un nouveau modèle d'apprentissage par instances multiples, TDAM-CRC, a été développé pour améliorer la prédiction pronostique dans le cancer colorectal (CRC) en utilisant des images histopathologiques de diapositives entières. Ce modèle a été formé sur une cohorte de 581 patients et validé avec une cohorte indépendante de 1 031 patients, démontrant une performance prédictive supérieure par rapport aux systèmes de stadification clinique conventionnels. L'étude a également intégré des données multi-omiques pour améliorer l'interprétabilité du modèle et identifier de nouveaux biomarqueurs pronost…

A novel multiple instance learning model, TDAM-CRC, has been developed to enhance prognostic prediction in colorectal cancer (CRC) by utilizing histopathological whole-slide images. This model was trained on a cohort of 581 patients and validated with an independent cohort of 1,031 patients, demonstrating superior predictive performance compared to conventional clinical staging systems. The study also integrated multi-omics data to improve model interpretability and identify new prognostic biomarkers.

Deep Pathomic Learning Defines Prognostic Subtypes and Molecular Drivers in Colorectal Cancer

arXiv:2308.00929v3 Announce Type: replace 
Abstract: Colonoscopic Polyp Re-Identification aims to match the same polyp from a large gallery with images from different views taken using different cameras and plays an important role in the prevention and treatment of colorectal cancer. However, traditional methods for object ReID directly adopting CNN models trained on the ImageNet dataset usually produce unsatisfactory retrieval performance on colonoscopic datasets due to the large domain gap. Additionally, these methods neglect to explore the potential of self-discrepancy among intra-class or inter-class relations in the colonoscopic polyp dataset, which remains an open research problem in the medical community. To solve this dilemma, we propose a simple but effective training method named Colo-ReID, which can help our model learn more general and discriminative knowledge based on the meta-learning strategy in scenarios with fewer samples. Based on this, a dynamic Meta-Learning Regulation mechanism called MLR is introduced to further boost the performance of polyp re-identification. Our experimental results show that Colo-ReID consistently outperforms second-best method in terms of mAP performance by +2.3% on polyp re-identification task. Our source code is also publicly available at https://github.com/JeremyXSC/Colo-ReID.

تم اقتراح طريقة جديدة تُدعى Colo-ReID لإعادة تحديد الهوية للزوائد القولونية، والتي تهدف إلى تحسين مطابقة الزوائد من وجهات نظر مختلفة باستخدام كاميرات متعددة، مما يعالج تحديًا كبيرًا في الوقاية والعلاج من سرطان القولون. لقد واجهت نماذج CNN التقليدية صعوبات بسبب الفجوات في المجال وعدم استكشاف العلاقات داخل الفئة وبين الفئات في مجموعات بيانات الزوائد.

Se ha propuesto un nuevo método llamado Colo-ReID para la re-identificación de pólipos colonoscópicos, que busca mejorar la coincidencia de pólipos desde diversas vistas de cámaras, abordando un desafío significativo en la prevención y tratamiento del cáncer colorrectal. Los modelos CNN tradicionales han tenido dificultades debido a las brechas de dominio y la falta de exploración de las relaciones intra-clase e inter-clase en los conjuntos de datos de pólipos.

Une nouvelle méthode appelée Colo-ReID a été proposée pour la ré-identification des polypes coloscopiques, visant à améliorer l'appariement des polypes à partir de diverses vues de caméras, répondant à un défi majeur dans la prévention et le traitement du cancer colorectal. Les modèles CNN traditionnels ont rencontré des difficultés en raison des écarts de domaine et du manque d'exploration des relations intra-classe et inter-classe dans les ensembles de données de polypes.

A new method called Colo-ReID has been proposed for Colonoscopic Polyp Re-Identification, which aims to enhance the matching of polyps from various camera views, addressing a significant challenge in colorectal cancer prevention and treatment. Traditional CNN models have struggled with this task due to domain gaps and the lack of exploration of intra-class and inter-class relations in polyp datasets.

Deep Pathomic Learning Defines Prognostic Subtypes and Molecular Drivers in Colorectal Cancer

Was this article worth reading? Share it

Attentive AI

IntelliQ

Research AI