Journal Article

PUBDB-2021-02773

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Scaling the U-net: Segmentation of biodegradable bone implants in high-resolution synchrotron microtomograms

Baltruschat, I.-M. (Corresponding author)DESY* ; Cwieka, H.Extern*HZG* ; Krueger, D.Extern* ; Zeller-Plumhoff, B.Extern* ; Schlünzen, F.XFEL.EU*DESY* ; Willumeit, R.Extern* ; Moosmann, J. (Corresponding author)Extern* ; Heuser, P.DESY*

2021
Macmillan Publishers Limited, part of Springer Nature [London]

This record in other databases:        

Please use a persistent id in citations: doi:10.1038/s41598-021-03542-y  doi:10.3204/PUBDB-2021-02773

Abstract: Highly accurate segmentation of large 3D volumes is a demanding task. Challenging applications like the segmentation of synchrotron microtomograms (SRuCT) at high-resolution, which suffer from low contrast, high spatial variability and measurement artifacts, readily exceed the capacities of conventional segmentation methods, including the manual segmentation by human experts. The quantitative characterization of the osseointegration and spatio-temporal biodegradation process of bone implants requires reliable, and very precise segmentation. 2D U-net based segmentation has been a natural choice. To obtain optimal results, we investigated the scaling of 2D U-net for high resolution grayscale volumes, the impact of most crucial model hyper-parameters (i.e., the model width, depth, and input size) and multi-axes prediction fusing. To leverage the 3D information of high-resolution SRuCT, common three axes prediction fusing is extended, investigating the effect of adding more than three axes prediction. In a systematic evaluation we compare the performance of scaling the U-net by intersection over union and quantitative measurements of osseointegration and degradation parameters. Overall, we observe that a compound scaling of the U-net and multi-axes prediction fusing with soft voting yields the best results. The quantitative analysis shows that the trained model is better and significantly faster than the current semi-manual segmentation.

Note: - SynchroLoad project (BMBF project number 05K16CGA) - MgBone project (BMBF project number 05K16CGB) which are funded by the Röntgen-Ångström Cluster (RÅC), a bilateral research collaboration of the Swedish government and the German Federal Ministry of Education and Research (BMBF). - HC acknowledges funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant, agreement No 811226. - We acknowledge provision of beamtime, related to the proposals II-20170009, I-20160104 I-20170074 at beamline P05 at PETRA III at DESY, a member of the Helmholtz Association (HGF). - Other parts were supported by the Helmholtz Association Initiative and Networking Fund under project number ZT-I-0003 (Helmholtz Analytics FrameworkHAF), - the Helmholtz Imaging Platform HIP (a platform of the Helmholtz Incubator on Information and Data Science), - and the MDLMA project (BMBF project number 031L0202C). - We thank Diamond Light Source for granting beamtime MG22346-1 at beamline I13-2

---

Contributing Institute(s):

1. DOOR-User (DOOR ; HAS-User)
2. Helmholtz-Zentrum Hereon (Hereon)
3. Informationstechnologie (IT)

Research Program(s):

1. 6G3 - PETRA III (DESY) (POF4-6G3) (POF4-6G3)
2. HAF - Helmholtz Analytics Framework (ZT-I-0003) (ZT-I-0003)
3. FS-Proposal: I-20160104 (I-20160104) (I-20160104)
4. FS-Proposal: II-20170009 (II-20170009) (II-20170009)
5. FS-Proposal: I-20170074 (I-20170074) (I-20170074)

Experiment(s):

1. PETRA Beamline P05 (PETRA III)

---

Database coverage:
; ; ; ; Article Processing Charges ; BIOSIS Previews ; Biological Abstracts ; Clarivate Analytics Master Journal List ; Current Contents - Physical, Chemical and Earth Sciences ; DOAJ Seal ; Ebsco Academic Search ; Essential Science Indicators ; Fees ; IF < 5 ; JCR ; PubMed Central ; SCOPUS ; Science Citation Index Expanded ; Web of Science Core Collection ; Zoological Record

---