Catheter detection in 3D ultrasound by CNN

Hongxu Yang, Caifeng Shan, Alexander F. Kolen, Peter de With

Research output: Contribution to conferenceOtherAcademic


In this paper, we propose a catheter detection method based on convolutional neural networks (CNNs) in 3D US. Voxels in US images are classified as catheter (or not) using triplanarbased CNNs. Our proposed CNN employs two-stage training with weighted loss function, which can cope with highly imbalanced training data and improves classification accuracy. Based on classified volumes, the catheters are localized with an average position error of smaller than 3 voxels in the examined datasets, indicating that catheters are always detected
in noisy and low-resolution images.
Original languageEnglish
Number of pages2
Publication statusPublished - 2018
EventThe Netherlands Conference on Computer Vision 2018 -
Duration: 26 Sep 201827 Sep 2018


ConferenceThe Netherlands Conference on Computer Vision 2018
Abbreviated titleNCCV18

Fingerprint Dive into the research topics of 'Catheter detection in 3D ultrasound by CNN'. Together they form a unique fingerprint.

Cite this