We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

eess.IV

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Electrical Engineering and Systems Science > Image and Video Processing

Title: SkullEngine: A Multi-stage CNN Framework for Collaborative CBCT Image Segmentation and Landmark Detection

Abstract: We propose a multi-stage coarse-to-fine CNN-based framework, called SkullEngine, for high-resolution segmentation and large-scale landmark detection through a collaborative, integrated, and scalable JSD model and three segmentation and landmark detection refinement models. We evaluated our framework on a clinical dataset consisting of 170 CBCT/CT images for the task of segmenting 2 bones (midface and mandible) and detecting 175 clinically common landmarks on bones, teeth, and soft tissues.
Comments: 10 pages, 5 figures, accepted by MLMI 2021
Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2110.03828 [eess.IV]
  (or arXiv:2110.03828v2 [eess.IV] for this version)

Submission history

From: Qin Liu [view email]
[v1] Thu, 7 Oct 2021 23:25:33 GMT (7540kb,D)
[v2] Tue, 21 Dec 2021 16:45:49 GMT (7540kb,D)

Link back to: arXiv, form interface, contact.