Learning Volumetric Shape Super-Resolution for Cranial Implant Design

Matthias Eder; Jianning Li; Jan Egger

doi:10.1007/978-3-030-64327-0_12

Learning Volumetric Shape Super-Resolution for Cranial Implant Design

Matthias Eder, Jianning Li, Jan Egger^*

^*Corresponding author for this work

Institute of Computer Graphics and Vision (7100)

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › peer-review

Abstract

Cranioplasty is the process of repairing cranial defects or deformations, which may be the result of injuries or necessary medical treatments such as brain tumor surgery. For this procedure, it is necessary to generate a high-quality cranial implant, which needs to be shaped individually for each skull and each defect. This tends to be a very time consuming task and requires also in-depth knowledge of various CAM/CAD programs. In this work, we present a novel automatic three-stage implant generation pipeline. First, skull completion is conducted in low resolution using a trained artificial neural network (ANN). Second, the completed low-resolution skull is sent to a super-resolution network, which up-samples the low-resolution skull to higher resolution while, at the same time, filling the skull surface with geometric details. Finally, by simple subtraction and blob filtering, the desired high-resolution implant is generated.

Original language	English
Title of host publication	Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings
Editors	Jianning Li, Jan Egger
Publisher	Springer, Cham
Pages	104 -113
Number of pages	10
Edition	1
ISBN (Electronic)	978-3-030-64327-0
ISBN (Print)	978-3-030-64326-3
DOIs	https://doi.org/10.1007/978-3-030-64327-0_12
Publication status	E-pub ahead of print - 4 Dec 2020
Event	1st Automatization of Cranial Implant Design in Cranioplasty Challenge: AutoImplant 2020 - Virtual, Lima, Peru Duration: 8 Oct 2020 → … https://autoimplant.grand-challenge.org/

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12439 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	1st Automatization of Cranial Implant Design in Cranioplasty Challenge
Country/Territory	Peru
City	Virtual, Lima
Period	8/10/20 → …
Internet address	https://autoimplant.grand-challenge.org/

Keywords

Cranioplasty
Deep learning
Shape completion
Super-resolution

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-030-64327-0_12

Cite this

Eder, M., Li, J., & Egger, J. (2020). Learning Volumetric Shape Super-Resolution for Cranial Implant Design. In J. Li, & J. Egger (Eds.), Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings (1 ed., pp. 104 -113). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12439 LNCS). Springer, Cham. https://doi.org/10.1007/978-3-030-64327-0_12

Learning Volumetric Shape Super-Resolution for Cranial Implant Design. / Eder, Matthias; Li, Jianning; Egger, Jan.

Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings. ed. / Jianning Li; Jan Egger. 1. ed. Springer, Cham, 2020. p. 104 -113 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12439 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › peer-review

Eder, M, Li, J & Egger, J 2020, Learning Volumetric Shape Super-Resolution for Cranial Implant Design. in J Li & J Egger (eds), Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings. 1 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12439 LNCS, Springer, Cham, pp. 104 -113, 1st Automatization of Cranial Implant Design in Cranioplasty Challenge, Virtual, Lima, Peru, 8/10/20. https://doi.org/10.1007/978-3-030-64327-0_12

Eder M, Li J, Egger J. Learning Volumetric Shape Super-Resolution for Cranial Implant Design. In Li J, Egger J, editors, Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings. 1 ed. Springer, Cham. 2020. p. 104 -113. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). Epub 2020 Dec 4. doi: 10.1007/978-3-030-64327-0_12

Eder, Matthias ; Li, Jianning ; Egger, Jan. / Learning Volumetric Shape Super-Resolution for Cranial Implant Design. Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings. editor / Jianning Li ; Jan Egger. 1. ed. Springer, Cham, 2020. pp. 104 -113 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{8cbc993b922e4884bb21c2a4846867e5,

title = "Learning Volumetric Shape Super-Resolution for Cranial Implant Design",

abstract = "Cranioplasty is the process of repairing cranial defects or deformations, which may be the result of injuries or necessary medical treatments such as brain tumor surgery. For this procedure, it is necessary to generate a high-quality cranial implant, which needs to be shaped individually for each skull and each defect. This tends to be a very time consuming task and requires also in-depth knowledge of various CAM/CAD programs. In this work, we present a novel automatic three-stage implant generation pipeline. First, skull completion is conducted in low resolution using a trained artificial neural network (ANN). Second, the completed low-resolution skull is sent to a super-resolution network, which up-samples the low-resolution skull to higher resolution while, at the same time, filling the skull surface with geometric details. Finally, by simple subtraction and blob filtering, the desired high-resolution implant is generated.",

keywords = "Cranioplasty, Deep learning, Shape completion, Super-resolution",

author = "Matthias Eder and Jianning Li and Jan Egger",

year = "2020",

month = dec,

day = "4",

doi = "10.1007/978-3-030-64327-0_12",

language = "English",

isbn = "978-3-030-64326-3",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer, Cham",

pages = "104 --113",

editor = "Jianning Li and Jan Egger",

booktitle = "Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings",

edition = "1",

note = "1st Automatization of Cranial Implant Design in Cranioplasty Challenge : AutoImplant 2020 ; Conference date: 08-10-2020",

url = "https://autoimplant.grand-challenge.org/",

}

TY - GEN

T1 - Learning Volumetric Shape Super-Resolution for Cranial Implant Design

AU - Eder, Matthias

AU - Li, Jianning

AU - Egger, Jan

PY - 2020/12/4

Y1 - 2020/12/4

N2 - Cranioplasty is the process of repairing cranial defects or deformations, which may be the result of injuries or necessary medical treatments such as brain tumor surgery. For this procedure, it is necessary to generate a high-quality cranial implant, which needs to be shaped individually for each skull and each defect. This tends to be a very time consuming task and requires also in-depth knowledge of various CAM/CAD programs. In this work, we present a novel automatic three-stage implant generation pipeline. First, skull completion is conducted in low resolution using a trained artificial neural network (ANN). Second, the completed low-resolution skull is sent to a super-resolution network, which up-samples the low-resolution skull to higher resolution while, at the same time, filling the skull surface with geometric details. Finally, by simple subtraction and blob filtering, the desired high-resolution implant is generated.

AB - Cranioplasty is the process of repairing cranial defects or deformations, which may be the result of injuries or necessary medical treatments such as brain tumor surgery. For this procedure, it is necessary to generate a high-quality cranial implant, which needs to be shaped individually for each skull and each defect. This tends to be a very time consuming task and requires also in-depth knowledge of various CAM/CAD programs. In this work, we present a novel automatic three-stage implant generation pipeline. First, skull completion is conducted in low resolution using a trained artificial neural network (ANN). Second, the completed low-resolution skull is sent to a super-resolution network, which up-samples the low-resolution skull to higher resolution while, at the same time, filling the skull surface with geometric details. Finally, by simple subtraction and blob filtering, the desired high-resolution implant is generated.

KW - Cranioplasty

KW - Deep learning

KW - Shape completion

KW - Super-resolution

UR - http://www.scopus.com/inward/record.url?scp=85097838539&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-64327-0_12

DO - 10.1007/978-3-030-64327-0_12

M3 - Conference paper

SN - 978-3-030-64326-3

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 104

EP - 113

BT - Towards the Automatization of Cranial Implant Design in Cranioplasty - First Challenge, AutoImplant 2020, Held in Conjunction with MICCAI 2020, Proceedings

A2 - Li, Jianning

A2 - Egger, Jan

PB - Springer, Cham

T2 - 1st Automatization of Cranial Implant Design in Cranioplasty Challenge

Y2 - 8 October 2020

ER -

Learning Volumetric Shape Super-Resolution for Cranial Implant Design

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this