Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography

Aleksei Tiulpin; Mikko Finnilä; Petri Lehenkari; Heikki J. Nieminen; Simo Saarakkala

doi:10.1007/978-3-030-40605-9_12

Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography

Aleksei Tiulpin^*
, Mikko Finnilä
, Petri Lehenkari
, Heikki J. Nieminen
, Simo Saarakkala

^*Corresponding author for this work

Department of Neuroscience and Biomedical Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

8 Citations (Scopus)

Abstract

Three-dimensional (3D) semi-quantitative grading of pathological features in articular cartilage (AC) offers significant improvements in basic research of osteoarthritis (OA). We have earlier developed the 3D protocol for imaging of AC and its structures which includes staining of the sample with a contrast agent (phosphotungstic acid, PTA) and a consequent scanning with micro-computed tomography. Such a protocol was designed to provide X-ray attenuation contrast to visualize AC structure. However, at the same time, this protocol has one major disadvantage: the loss of contrast at the tidemark (calcified cartilage interface, CCI). An accurate segmentation of CCI can be very important for understanding the etiology of OA and ex-vivo evaluation of tidemark condition at early OA stages. In this paper, we present the first application of Deep Learning to PTA-stained osteochondral samples that allows to perform tidemark segmentation in a fully-automatic manner. Our method is based on U-Net trained using a combination of binary cross-entropy and soft-Jaccard loss. On cross-validation, this approach yielded intersection over the union of 0.59, 0.70, 0.79, 0.83 and 0.86 within 15 μm, 30 μm, 45 μm, 60 μm, and 75 μm padded zones around the tidemark, respectively. Our codes and the dataset that consisted of 35 PTA-stained human AC samples are made publicly available together with the segmentation masks to facilitate the development of biomedical image segmentation methods.

Original language	English
Title of host publication	Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings
Editors	Jacques Blanc-Talon, Patrice Delmas, Wilfried Philips, Dan Popescu, Paul Scheunders
Publisher	Springer
Pages	131-138
Number of pages	8
ISBN (Print)	9783030406042
DOIs	https://doi.org/10.1007/978-3-030-40605-9_12
Publication status	Published - 1 Jan 2020
MoE publication type	A4 Conference publication
Event	International Conference on Advanced Concepts for Intelligent Vision Systems - Auckland, New Zealand Duration: 10 Feb 2020 → 14 Feb 2020 Conference number: 20

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Publisher	Springer
Volume	12002 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Conference on Advanced Concepts for Intelligent Vision Systems
Abbreviated title	ACIVS
Country/Territory	New Zealand
City	Auckland
Period	10/02/2020 → 14/02/2020

Funding

This work was supported by Academy of Finland (grants 268378, 303786, 311586 and 314286), European Research Council under the European Union?s Seventh Framework Programme (FP/2007-2013)/ERC Grant Agreement no. 336267, the strategic funding of the University of Oulu and KAUTE foundation. We would also like to acknowledge CSC IT Center for Science, Finland, for generous computational resources. Tuomas Frondelius is acknowledged for the initial experiments with the data and Santeri Rytky is acknowledged for the useful comments and proofreading of the paper.

Keywords

3D histology
Deep Learning
Osteoarthritis

Access to Document

10.1007/978-3-030-40605-9_12

Cite this

Tiulpin, A., Finnilä, M., Lehenkari, P., Nieminen, H. J., & Saarakkala, S. (2020). Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography. In J. Blanc-Talon, P. Delmas, W. Philips, D. Popescu, & P. Scheunders (Eds.), Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings (pp. 131-138). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12002 LNCS). Springer. https://doi.org/10.1007/978-3-030-40605-9_12

Tiulpin, Aleksei ; Finnilä, Mikko ; Lehenkari, Petri et al. / Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography. Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings. editor / Jacques Blanc-Talon ; Patrice Delmas ; Wilfried Philips ; Dan Popescu ; Paul Scheunders. Springer, 2020. pp. 131-138 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{56ed22bf56ae462a9e9870ce43681dc3,

title = "Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography",

abstract = "Three-dimensional (3D) semi-quantitative grading of pathological features in articular cartilage (AC) offers significant improvements in basic research of osteoarthritis (OA). We have earlier developed the 3D protocol for imaging of AC and its structures which includes staining of the sample with a contrast agent (phosphotungstic acid, PTA) and a consequent scanning with micro-computed tomography. Such a protocol was designed to provide X-ray attenuation contrast to visualize AC structure. However, at the same time, this protocol has one major disadvantage: the loss of contrast at the tidemark (calcified cartilage interface, CCI). An accurate segmentation of CCI can be very important for understanding the etiology of OA and ex-vivo evaluation of tidemark condition at early OA stages. In this paper, we present the first application of Deep Learning to PTA-stained osteochondral samples that allows to perform tidemark segmentation in a fully-automatic manner. Our method is based on U-Net trained using a combination of binary cross-entropy and soft-Jaccard loss. On cross-validation, this approach yielded intersection over the union of 0.59, 0.70, 0.79, 0.83 and 0.86 within 15 μm, 30 μm, 45 μm, 60 μm, and 75 μm padded zones around the tidemark, respectively. Our codes and the dataset that consisted of 35 PTA-stained human AC samples are made publicly available together with the segmentation masks to facilitate the development of biomedical image segmentation methods.",

keywords = "3D histology, Deep Learning, Osteoarthritis",

author = "Aleksei Tiulpin and Mikko Finnil{\"a} and Petri Lehenkari and Nieminen, \{Heikki J.\} and Simo Saarakkala",

year = "2020",

month = jan,

day = "1",

doi = "10.1007/978-3-030-40605-9\_12",

language = "English",

isbn = "9783030406042",

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

publisher = "Springer",

pages = "131--138",

editor = "Jacques Blanc-Talon and Patrice Delmas and Wilfried Philips and Dan Popescu and Paul Scheunders",

booktitle = "Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings",

address = "Germany",

note = "International Conference on Advanced Concepts for Intelligent Vision Systems, ACIVS ; Conference date: 10-02-2020 Through 14-02-2020",

}

Tiulpin, A, Finnilä, M, Lehenkari, P, Nieminen, HJ & Saarakkala, S 2020, Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography. in J Blanc-Talon, P Delmas, W Philips, D Popescu & P Scheunders (eds), Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12002 LNCS, Springer, pp. 131-138, International Conference on Advanced Concepts for Intelligent Vision Systems, Auckland, New Zealand, 10/02/2020. https://doi.org/10.1007/978-3-030-40605-9_12

Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography. / Tiulpin, Aleksei; Finnilä, Mikko; Lehenkari, Petri et al.
Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings. ed. / Jacques Blanc-Talon; Patrice Delmas; Wilfried Philips; Dan Popescu; Paul Scheunders. Springer, 2020. p. 131-138 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12002 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography

AU - Tiulpin, Aleksei

AU - Finnilä, Mikko

AU - Lehenkari, Petri

AU - Nieminen, Heikki J.

AU - Saarakkala, Simo

N1 - Conference code: 20

PY - 2020/1/1

Y1 - 2020/1/1

N2 - Three-dimensional (3D) semi-quantitative grading of pathological features in articular cartilage (AC) offers significant improvements in basic research of osteoarthritis (OA). We have earlier developed the 3D protocol for imaging of AC and its structures which includes staining of the sample with a contrast agent (phosphotungstic acid, PTA) and a consequent scanning with micro-computed tomography. Such a protocol was designed to provide X-ray attenuation contrast to visualize AC structure. However, at the same time, this protocol has one major disadvantage: the loss of contrast at the tidemark (calcified cartilage interface, CCI). An accurate segmentation of CCI can be very important for understanding the etiology of OA and ex-vivo evaluation of tidemark condition at early OA stages. In this paper, we present the first application of Deep Learning to PTA-stained osteochondral samples that allows to perform tidemark segmentation in a fully-automatic manner. Our method is based on U-Net trained using a combination of binary cross-entropy and soft-Jaccard loss. On cross-validation, this approach yielded intersection over the union of 0.59, 0.70, 0.79, 0.83 and 0.86 within 15 μm, 30 μm, 45 μm, 60 μm, and 75 μm padded zones around the tidemark, respectively. Our codes and the dataset that consisted of 35 PTA-stained human AC samples are made publicly available together with the segmentation masks to facilitate the development of biomedical image segmentation methods.

AB - Three-dimensional (3D) semi-quantitative grading of pathological features in articular cartilage (AC) offers significant improvements in basic research of osteoarthritis (OA). We have earlier developed the 3D protocol for imaging of AC and its structures which includes staining of the sample with a contrast agent (phosphotungstic acid, PTA) and a consequent scanning with micro-computed tomography. Such a protocol was designed to provide X-ray attenuation contrast to visualize AC structure. However, at the same time, this protocol has one major disadvantage: the loss of contrast at the tidemark (calcified cartilage interface, CCI). An accurate segmentation of CCI can be very important for understanding the etiology of OA and ex-vivo evaluation of tidemark condition at early OA stages. In this paper, we present the first application of Deep Learning to PTA-stained osteochondral samples that allows to perform tidemark segmentation in a fully-automatic manner. Our method is based on U-Net trained using a combination of binary cross-entropy and soft-Jaccard loss. On cross-validation, this approach yielded intersection over the union of 0.59, 0.70, 0.79, 0.83 and 0.86 within 15 μm, 30 μm, 45 μm, 60 μm, and 75 μm padded zones around the tidemark, respectively. Our codes and the dataset that consisted of 35 PTA-stained human AC samples are made publicly available together with the segmentation masks to facilitate the development of biomedical image segmentation methods.

KW - 3D histology

KW - Deep Learning

KW - Osteoarthritis

UR - https://www.scopus.com/pages/publications/85080888133

U2 - 10.1007/978-3-030-40605-9_12

DO - 10.1007/978-3-030-40605-9_12

M3 - Conference article in proceedings

AN - SCOPUS:85080888133

SN - 9783030406042

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

SP - 131

EP - 138

BT - Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings

A2 - Blanc-Talon, Jacques

A2 - Delmas, Patrice

A2 - Philips, Wilfried

A2 - Popescu, Dan

A2 - Scheunders, Paul

PB - Springer

T2 - International Conference on Advanced Concepts for Intelligent Vision Systems

Y2 - 10 February 2020 through 14 February 2020

ER -

Tiulpin A, Finnilä M, Lehenkari P, Nieminen HJ, Saarakkala S. Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography. In Blanc-Talon J, Delmas P, Philips W, Popescu D, Scheunders P, editors, Advanced Concepts for Intelligent Vision Systems - 20th International Conference, ACIVS 2020, Proceedings. Springer. 2020. p. 131-138. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-40605-9_12

Deep-Learning for Tidemark Segmentation in Human Osteochondral Tissues Imaged with Micro-computed Tomography

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this