arXiv:2511.04334v1 Announce Type: cross 
Abstract: The accurate delineation of tumours in radiological images like Computed Tomography is a very specialised and time-consuming task, and currently a bottleneck preventing quantitative analyses to be performed routinely in the clinical setting. For this reason, developing methods for the automated segmentation of tumours in medical imaging is of the utmost importance and has driven significant efforts in recent years. However, challenges regarding the impracticality of 3D scans, given the large amount of voxels to be analysed, usually requires the downsampling of such images or using patches thereof when applying traditional convolutional neural networks. To overcome this problem, in this paper we propose a new methodology that uses, divided into two stages, voxel sparsification and submanifold sparse convolutional networks. This method allows segmentations to be performed with high-resolution inputs and a native 3D model architecture, obtaining state-of-the-art accuracies while significantly reducing the computational resources needed in terms of GPU memory and time. We studied the deployment of this methodology in the context of Computed Tomography images of renal cancer patients from the KiTS23 challenge, and our method achieved results competitive with the challenge winners, with Dice similarity coefficients of 95.8% for kidneys + masses, 85.7% for tumours + cysts, and 80.3% for tumours alone. Crucially, our method also offers significant computational improvements, achieving up to a 60% reduction in inference time and up to a 75\% reduction in VRAM usage compared to an equivalent dense architecture, across both CPU and various GPU cards tested.

تقدم دراسة حديثة شبكات الالتفاف الفرعية النادرة بهدف أتمتة تقسيم الكلى والأورام الكلوية ثلاثية الأبعاد في الأشعة المقطعية. هذه الخطوة مهمة لأنها تعالج الطبيعة المتخصصة والمستهلكة للوقت لتحديد الأورام في التصوير الطبي، والتي كانت تمثل عنق زجاجة كبير في التحليلات السريرية. من خلال تحسين دقة وكفاءة تحديد الأورام، يمكن أن تعزز هذه التكنولوجيا الممارسات السريرية الروتينية وفي النهاية تؤدي إلى نتائج أفضل للمرضى.

Un estudio reciente presenta Redes de Convolución Sparse Submanifold destinadas a automatizar la segmentación 3D de riñones y tumores renales en tomografías computarizadas. Este avance es crucial ya que aborda la naturaleza especializada y que consume mucho tiempo de la delimitación de tumores en imágenes médicas, que ha sido un importante cuello de botella en los análisis clínicos. Al mejorar la precisión y la eficiencia en la identificación de tumores, esta tecnología podría mejorar las prácticas clínicas rutinarias y, en última instancia, llevar a mejores resultados para los pacientes.

Une étude récente présente des réseaux de convolution sparse sous-manifold visant à automatiser la segmentation 3D des reins et des tumeurs rénales dans les scans CT. Cette avancée est cruciale car elle répond à la nature chronophage et spécialisée de la délimitation des tumeurs en imagerie médicale, qui a constitué un goulot d'étranglement important dans les analyses cliniques. En améliorant la précision et l'efficacité de l'identification des tumeurs, cette technologie pourrait améliorer les pratiques cliniques de routine et, en fin de compte, conduire à de meilleurs résultats pour les patients.

A recent study introduces Submanifold Sparse Convolutional Networks aimed at automating the 3D segmentation of kidneys and kidney tumors in CT scans. This advancement is crucial as it addresses the time-consuming and specialized nature of tumor delineation in medical imaging, which has been a significant bottleneck in clinical analyses. By improving the accuracy and efficiency of tumor identification, this technology could enhance routine clinical practices and ultimately lead to better patient outcomes.

Submanifold Sparse Convolutional Networks for Automated 3D Segmentation of Kidneys and Kidney Tumours in Computed Tomography

arXiv:2601.08455v1 Announce Type: new 
Abstract: Objectives: High-grade serous ovarian carcinoma (HGSOC) is typically diagnosed at an advanced stage with extensive peritoneal metastases, making treatment challenging. Neoadjuvant chemotherapy (NACT) is often used to reduce tumor burden before surgery, but about 40% of patients show limited response. Radiomics, combined with machine learning (ML), offers a promising non-invasive method for predicting NACT response by analyzing computed tomography (CT) imaging data. This study aimed to improve response prediction in HGSOC patients undergoing NACT by integration different feature selection methods. Materials and methods: A framework for selecting robust radiomics features was introduced by employing an automated randomisation algorithm to mimic inter-observer variability, ensuring a balance between feature robustness and prediction accuracy. Four response metrics were used: chemotherapy response score (CRS), RECIST, volume reduction (VolR), and diameter reduction (DiaR). Lesions in different anatomical sites were studied. Pre- and post-NACT CT scans were used for feature extraction and model training on one cohort, and an independent cohort was used for external testing. Results: The best prediction performance was achieved using all lesions combined for VolR prediction, with an AUC of 0.83. Omental lesions provided the best results for CRS prediction (AUC 0.77), while pelvic lesions performed best for DiaR (AUC 0.76). Conclusion: The integration of robustness into the feature selection processes ensures the development of reliable models and thus facilitates the implementation of the radiomics models in clinical applications for HGSOC patients. Future work should explore further applications of radiomics in ovarian cancer, particularly in real-time clinical settings.

أظهرت دراسة حديثة تطوير نماذج راديو ميكولوجيا تنبؤية وموثوقة تهدف إلى تقييم استجابة العلاج الكيميائي لدى المرضى المصابين بسرطان المبيض الحاد عالي الدرجة (HGSOC)، وهو سرطان يتم تشخيصه عادة في مراحل متقدمة. تستخدم الدراسة تقنيات التعلم الآلي لتحليل بيانات التصوير المقطعي المحوسب، مما يعزز من دقة التنبؤ باستجابة العلاج الكيميائي المساعد.

Un estudio reciente ha desarrollado modelos de radiómica predictivos y robustos destinados a evaluar la respuesta a la quimioterapia en pacientes con carcinoma seroso de ovario de alto grado (HGSOC), un cáncer que generalmente se diagnostica en una etapa avanzada. La investigación utiliza técnicas de aprendizaje automático para analizar datos de imágenes por tomografía computarizada, mejorando así la predicción de la respuesta a la quimioterapia neoadyuvante.

Une étude récente a développé des modèles de radiomique prédictifs et robustes visant à évaluer la réponse à la chimiothérapie chez les patientes atteintes de carcinome ovarien séreux de haut grade (HGSOC), un cancer généralement diagnostiqué à un stade avancé. La recherche utilise des techniques d'apprentissage automatique pour analyser les données d'imagerie par tomodensitométrie, améliorant ainsi la prédiction de la réponse à la chimiothérapie néoadjuvante.

A recent study has developed predictive and robust radiomics models aimed at assessing chemotherapy response in patients with high-grade serous ovarian carcinoma (HGSOC), a cancer typically diagnosed at an advanced stage. The research utilizes machine learning techniques to analyze computed tomography imaging data, enhancing the prediction of neoadjuvant chemotherapy response.

Submanifold Sparse Convolutional Networks for Automated 3D Segmentation of Kidneys and Kidney Tumours in Computed Tomography

Was this article worth reading? Share it

LucidQuery AI

Airparser

Attentive AI

Supametas.AI

GPTHumanizer

Gonogo by Trampoline

Ready to build your own newsroom?