Automatic localization of locally similar structures based on the scale-widening random regression forest

Darko Stern, Thomas Ebner, Martin Urschler

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

Selection of set of training pixels and feature range show to be critical scale-related parameters with high impact on results in localization methods based on random regression forests (RRF). Trained on pixels randomly selected from images with long range features, RRF captures the variation in landmark location but often without reaching satisfying accuracy. Conversely, training an RRF with short range features in a landmark's close surroundings enables accurate localization, but at the cost of ambiguous localization results in the presence of locally similar structures. We present a scale-widening RRF method that effectively handles such ambiguities. On a challenging hand radiography image data set, we achieve median and 90th percentile localization errors of 0.81 and 2.64mm, respectively, outperforming related state-of-the-art methods.
LanguageEnglish
Title of host publication2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI)
PublisherInstitute of Electrical and Electronics Engineers
Pages1422-1425
Number of pages4
DOIs
StatusPublished - May 2016
Event2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI) - Prague, Czech Republic
Duration: 13 Apr 201616 Apr 2016

Conference

Conference2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI)
CountryCzech Republic
CityPrague
Period13/04/1616/04/16

Fingerprint

Pixels
Radiography

Fields of Expertise

  • Information, Communication & Computing

Cite this

Stern, D., Ebner, T., & Urschler, M. (2016). Automatic localization of locally similar structures based on the scale-widening random regression forest. In 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI) (pp. 1422-1425). Institute of Electrical and Electronics Engineers. DOI: 10.1109/ISBI.2016.7493534

Automatic localization of locally similar structures based on the scale-widening random regression forest. / Stern, Darko; Ebner, Thomas; Urschler, Martin.

2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers, 2016. p. 1422-1425.

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Stern, D, Ebner, T & Urschler, M 2016, Automatic localization of locally similar structures based on the scale-widening random regression forest. in 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers, pp. 1422-1425, 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI), Prague, Czech Republic, 13/04/16. DOI: 10.1109/ISBI.2016.7493534
Stern D, Ebner T, Urschler M. Automatic localization of locally similar structures based on the scale-widening random regression forest. In 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers. 2016. p. 1422-1425. Available from, DOI: 10.1109/ISBI.2016.7493534
Stern, Darko ; Ebner, Thomas ; Urschler, Martin. / Automatic localization of locally similar structures based on the scale-widening random regression forest. 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers, 2016. pp. 1422-1425
@inproceedings{0809769041bf4ca79486e34a518878e6,
title = "Automatic localization of locally similar structures based on the scale-widening random regression forest",
abstract = "Selection of set of training pixels and feature range show to be critical scale-related parameters with high impact on results in localization methods based on random regression forests (RRF). Trained on pixels randomly selected from images with long range features, RRF captures the variation in landmark location but often without reaching satisfying accuracy. Conversely, training an RRF with short range features in a landmark's close surroundings enables accurate localization, but at the cost of ambiguous localization results in the presence of locally similar structures. We present a scale-widening RRF method that effectively handles such ambiguities. On a challenging hand radiography image data set, we achieve median and 90th percentile localization errors of 0.81 and 2.64mm, respectively, outperforming related state-of-the-art methods.",
author = "Darko Stern and Thomas Ebner and Martin Urschler",
year = "2016",
month = "5",
doi = "10.1109/ISBI.2016.7493534",
language = "English",
pages = "1422--1425",
booktitle = "2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI)",
publisher = "Institute of Electrical and Electronics Engineers",
address = "United States",

}

TY - GEN

T1 - Automatic localization of locally similar structures based on the scale-widening random regression forest

AU - Stern,Darko

AU - Ebner,Thomas

AU - Urschler,Martin

PY - 2016/5

Y1 - 2016/5

N2 - Selection of set of training pixels and feature range show to be critical scale-related parameters with high impact on results in localization methods based on random regression forests (RRF). Trained on pixels randomly selected from images with long range features, RRF captures the variation in landmark location but often without reaching satisfying accuracy. Conversely, training an RRF with short range features in a landmark's close surroundings enables accurate localization, but at the cost of ambiguous localization results in the presence of locally similar structures. We present a scale-widening RRF method that effectively handles such ambiguities. On a challenging hand radiography image data set, we achieve median and 90th percentile localization errors of 0.81 and 2.64mm, respectively, outperforming related state-of-the-art methods.

AB - Selection of set of training pixels and feature range show to be critical scale-related parameters with high impact on results in localization methods based on random regression forests (RRF). Trained on pixels randomly selected from images with long range features, RRF captures the variation in landmark location but often without reaching satisfying accuracy. Conversely, training an RRF with short range features in a landmark's close surroundings enables accurate localization, but at the cost of ambiguous localization results in the presence of locally similar structures. We present a scale-widening RRF method that effectively handles such ambiguities. On a challenging hand radiography image data set, we achieve median and 90th percentile localization errors of 0.81 and 2.64mm, respectively, outperforming related state-of-the-art methods.

U2 - 10.1109/ISBI.2016.7493534

DO - 10.1109/ISBI.2016.7493534

M3 - Conference contribution

SP - 1422

EP - 1425

BT - 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI)

PB - Institute of Electrical and Electronics Engineers

ER -