ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering

Hanna Leena Halme; Antti Korvenoja; Eero Salli

doi:10.1007/978-3-319-30858-6_18

ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering

Hanna Leena Halme^*, Antti Korvenoja, Eero Salli

^*Corresponding author for this work

Department of Neuroscience and Biomedical Engineering

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

8 Citations (Scopus)

Abstract

Automated methods for segmentation of ischemic stroke lesions could significantly reduce the workload of radiologists and speed up the beginning of patient treatment. In this paper, we present a method for subacute ischemic stroke lesion segmentation from multispectral magnetic resonance images (MRI). The method involves classification of voxels with a Random Forest algorithm and subsequent classification refinement with contextual clustering. In addition, we utilize the training data to build statistical group-specific templates and use them for calculation of individual voxel-wise differences from the global mean. Our method achieved a Dice coefficient of 0.61 for the leave-one-out cross-validated training data and 0.47 for the testing data of the ISLES challenge 2015.

Original language	English
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Publisher	Springer Verlag
Pages	211-221
Number of pages	11
Volume	9556
ISBN (Print)	9783319308579
DOIs	https://doi.org/10.1007/978-3-319-30858-6_18
Publication status	Published - 2016
MoE publication type	A4 Article in a conference publication
Event	International Workshop on Brainlesion - Munich, Germany Duration: 5 Oct 2015 → 5 Oct 2015 Conference number: 1

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9556
ISSN (Print)	03029743
ISSN (Electronic)	16113349

Workshop

Workshop	International Workshop on Brainlesion
Abbreviated title	Brainles
Country	Germany
City	Munich
Period	05/10/2015 → 05/10/2015

Access to Document

10.1007/978-3-319-30858-6_18

Link to publication in Scopus

Fingerprint Dive into the research topics of 'ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering'. Together they form a unique fingerprint.

Cite this

Halme, H. L., Korvenoja, A., & Salli, E. (2016). ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9556, pp. 211-221). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9556). Springer Verlag. https://doi.org/10.1007/978-3-319-30858-6_18

Halme, Hanna Leena ; Korvenoja, Antti ; Salli, Eero. / ISLES (SISS) challenge 2015 : Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9556 Springer Verlag, 2016. pp. 211-221 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{97d8c5e31f274373aab7d9f5be6addc7,

title = "ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering",

abstract = "Automated methods for segmentation of ischemic stroke lesions could significantly reduce the workload of radiologists and speed up the beginning of patient treatment. In this paper, we present a method for subacute ischemic stroke lesion segmentation from multispectral magnetic resonance images (MRI). The method involves classification of voxels with a Random Forest algorithm and subsequent classification refinement with contextual clustering. In addition, we utilize the training data to build statistical group-specific templates and use them for calculation of individual voxel-wise differences from the global mean. Our method achieved a Dice coefficient of 0.61 for the leave-one-out cross-validated training data and 0.47 for the testing data of the ISLES challenge 2015.",

author = "Halme, {Hanna Leena} and Antti Korvenoja and Eero Salli",

year = "2016",

doi = "10.1007/978-3-319-30858-6_18",

language = "English",

isbn = "9783319308579",

volume = "9556",

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

publisher = "Springer Verlag",

pages = "211--221",

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

address = "Germany",

note = "International Workshop on Brainlesion , Brainles ; Conference date: 05-10-2015 Through 05-10-2015",

}

Halme, HL, Korvenoja, A & Salli, E 2016, ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9556, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9556, Springer Verlag, pp. 211-221, International Workshop on Brainlesion , Munich, Germany, 05/10/2015. https://doi.org/10.1007/978-3-319-30858-6_18

ISLES (SISS) challenge 2015 : Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering. / Halme, Hanna Leena; Korvenoja, Antti; Salli, Eero.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9556 Springer Verlag, 2016. p. 211-221 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9556).

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

TY - GEN

T1 - ISLES (SISS) challenge 2015

T2 - International Workshop on Brainlesion

AU - Halme, Hanna Leena

AU - Korvenoja, Antti

AU - Salli, Eero

N1 - Conference code: 1

PY - 2016

Y1 - 2016

N2 - Automated methods for segmentation of ischemic stroke lesions could significantly reduce the workload of radiologists and speed up the beginning of patient treatment. In this paper, we present a method for subacute ischemic stroke lesion segmentation from multispectral magnetic resonance images (MRI). The method involves classification of voxels with a Random Forest algorithm and subsequent classification refinement with contextual clustering. In addition, we utilize the training data to build statistical group-specific templates and use them for calculation of individual voxel-wise differences from the global mean. Our method achieved a Dice coefficient of 0.61 for the leave-one-out cross-validated training data and 0.47 for the testing data of the ISLES challenge 2015.

AB - Automated methods for segmentation of ischemic stroke lesions could significantly reduce the workload of radiologists and speed up the beginning of patient treatment. In this paper, we present a method for subacute ischemic stroke lesion segmentation from multispectral magnetic resonance images (MRI). The method involves classification of voxels with a Random Forest algorithm and subsequent classification refinement with contextual clustering. In addition, we utilize the training data to build statistical group-specific templates and use them for calculation of individual voxel-wise differences from the global mean. Our method achieved a Dice coefficient of 0.61 for the leave-one-out cross-validated training data and 0.47 for the testing data of the ISLES challenge 2015.

UR - http://www.scopus.com/inward/record.url?scp=84961584027&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-30858-6_18

DO - 10.1007/978-3-319-30858-6_18

M3 - Conference contribution

AN - SCOPUS:84961584027

SN - 9783319308579

VL - 9556

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

SP - 211

EP - 221

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

PB - Springer Verlag

Y2 - 5 October 2015 through 5 October 2015

ER -

Halme HL, Korvenoja A, Salli E. ISLES (SISS) challenge 2015: Segmentation of stroke lesions using spatial normalization, random forest classification and contextual clustering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9556. Springer Verlag. 2016. p. 211-221. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-30858-6_18