Self super resolution for hepatic vessel CT segmentation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Self super resolution for hepatic vessel CT segmentation

Résumé

Computed tomography (CT) images are often reconstructed into anisotropic voxelized volumes. This leads to coarser and often inconsistent delineations across slices, which are nevertheless used as 3D ground truth for deep segmentation model training and performance reports. Recently, deep selfsupervised super resolution (SSR) was proposed to improve through-plane resolution in brain MR images without the need of calibrated training data (i.e pairs of low resolution / high resolution images). In this work, we study whether SSR can be useful to improve segmentation accuracy of hepatic vessels from abdominal CT scans using deep learning. Results on the public IRCADb dataset suggest that SSR can improve segmentation performances not only in terms of volumetric overlaps (Dice similarity), but also using more relevant topology preserving evaluation metrics (clDice).
Fichier principal
Vignette du fichier
Self_Super_Resolution_for_Hepatic_Vessel_CT_segmentation.pdf (519.14 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04470666 , version 1 (21-02-2024)

Identifiants

Citer

Vincent Jaouen, Ziqiao Wang, Pierre-Henri Conze, Dimitris Visvikis. Self super resolution for hepatic vessel CT segmentation. 2023 IEEE Nuclear Science Symposium, Medical Imaging Conference and International Symposium on Room-Temperature Semiconductor Detectors (NSS MIC RTSD), Nov 2023, Vancouver, France. pp.1-1, ⟨10.1109/NSSMICRTSD49126.2023.10337892⟩. ⟨hal-04470666⟩
7 Consultations
11 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More