Linear feature selection in texture analysis - A PLS based method

Joselene Marques; Christian Igel; Martin Lillholm; Erik Dam

doi:10.1007/s00138-012-0461-1

Linear feature selection in texture analysis - A PLS based method

Joselene Marques, Christian Igel, Martin Lillholm, Erik Dam

Datalogisk Institut

2 Citationer (Scopus)

Abstract

We present a texture analysis methodology that combined uncommitted machine-learning techniques and partial least square (PLS) in a fully automatic framework. Our approach introduces a robust PLS-based dimensionality reduction (DR) step to specifically address outliers and high-dimensional feature sets. The texture analysis framework was applied to diagnosis of knee osteoarthritis (OA). To classify between healthy subjects and OA patients, a generic bank of texture features was extracted from magnetic resonance images of tibial knee bone. The features were used as input to the DR algorithm, which first applied a PLS regression to rank the features and then defined the best number of features to retain in the model by an iterative learning phase. The outliers in the dataset, that could inflate the number of selected features, were eliminated by a pre-processing step. To cope with the limited number of samples, the data were evaluated using Monte Carlo cross validation (CV). The developed DR method demonstrated consistency in selecting a relatively homogeneous set of features across the CV iterations. Per each CV group, a median of 19 % of the original features was selected and considering all CV groups, the methods selected 36 % of the original features available. The diagnosis evaluation reached a generalization area-under-the-ROC curve of 0.92, which was higher than established cartilage-based markers known to relate to OA diagnosis.

Originalsprog	Engelsk
Tidsskrift	Machine Vision & Applications
Vol/bind	24
Udgave nummer	7
Sider (fra-til)	1435-1444
Antal sider	10
ISSN	0932-8092
DOI	https://doi.org/10.1007/s00138-012-0461-1
Status	Udgivet - okt. 2013

Adgang til dokumentet

10.1007/s00138-012-0461-1

Linear feature selection in texture analysis - A PLS based methodForlagets udgivne version, 433 KB

Citationsformater

@article{6d652fbb3590492ba09fdbd38ab3b9cd,

title = "Linear feature selection in texture analysis - A PLS based method",

abstract = "We present a texture analysis methodology that combined uncommitted machine-learning techniques and partial least square (PLS) in a fully automatic framework. Our approach introduces a robust PLS-based dimensionality reduction (DR) step to specifically address outliers and high-dimensional feature sets. The texture analysis framework was applied to diagnosis of knee osteoarthritis (OA). To classify between healthy subjects and OA patients, a generic bank of texture features was extracted from magnetic resonance images of tibial knee bone. The features were used as input to the DR algorithm, which first applied a PLS regression to rank the features and then defined the best number of features to retain in the model by an iterative learning phase. The outliers in the dataset, that could inflate the number of selected features, were eliminated by a pre-processing step. To cope with the limited number of samples, the data were evaluated using Monte Carlo cross validation (CV). The developed DR method demonstrated consistency in selecting a relatively homogeneous set of features across the CV iterations. Per each CV group, a median of 19 % of the original features was selected and considering all CV groups, the methods selected 36 % of the original features available. The diagnosis evaluation reached a generalization area-under-the-ROC curve of 0.92, which was higher than established cartilage-based markers known to relate to OA diagnosis.",

author = "Joselene Marques and Christian Igel and Martin Lillholm and Erik Dam",

year = "2013",

month = oct,

doi = "10.1007/s00138-012-0461-1",

language = "English",

volume = "24",

pages = "1435--1444",

journal = "Machine Vision and Applications",

issn = "0932-8092",

publisher = "Springer",

number = "7",

}

TY - JOUR

T1 - Linear feature selection in texture analysis - A PLS based method

AU - Marques, Joselene

AU - Igel, Christian

AU - Lillholm, Martin

AU - Dam, Erik

PY - 2013/10

Y1 - 2013/10

N2 - We present a texture analysis methodology that combined uncommitted machine-learning techniques and partial least square (PLS) in a fully automatic framework. Our approach introduces a robust PLS-based dimensionality reduction (DR) step to specifically address outliers and high-dimensional feature sets. The texture analysis framework was applied to diagnosis of knee osteoarthritis (OA). To classify between healthy subjects and OA patients, a generic bank of texture features was extracted from magnetic resonance images of tibial knee bone. The features were used as input to the DR algorithm, which first applied a PLS regression to rank the features and then defined the best number of features to retain in the model by an iterative learning phase. The outliers in the dataset, that could inflate the number of selected features, were eliminated by a pre-processing step. To cope with the limited number of samples, the data were evaluated using Monte Carlo cross validation (CV). The developed DR method demonstrated consistency in selecting a relatively homogeneous set of features across the CV iterations. Per each CV group, a median of 19 % of the original features was selected and considering all CV groups, the methods selected 36 % of the original features available. The diagnosis evaluation reached a generalization area-under-the-ROC curve of 0.92, which was higher than established cartilage-based markers known to relate to OA diagnosis.

AB - We present a texture analysis methodology that combined uncommitted machine-learning techniques and partial least square (PLS) in a fully automatic framework. Our approach introduces a robust PLS-based dimensionality reduction (DR) step to specifically address outliers and high-dimensional feature sets. The texture analysis framework was applied to diagnosis of knee osteoarthritis (OA). To classify between healthy subjects and OA patients, a generic bank of texture features was extracted from magnetic resonance images of tibial knee bone. The features were used as input to the DR algorithm, which first applied a PLS regression to rank the features and then defined the best number of features to retain in the model by an iterative learning phase. The outliers in the dataset, that could inflate the number of selected features, were eliminated by a pre-processing step. To cope with the limited number of samples, the data were evaluated using Monte Carlo cross validation (CV). The developed DR method demonstrated consistency in selecting a relatively homogeneous set of features across the CV iterations. Per each CV group, a median of 19 % of the original features was selected and considering all CV groups, the methods selected 36 % of the original features available. The diagnosis evaluation reached a generalization area-under-the-ROC curve of 0.92, which was higher than established cartilage-based markers known to relate to OA diagnosis.

U2 - 10.1007/s00138-012-0461-1

DO - 10.1007/s00138-012-0461-1

M3 - Journal article

SN - 0932-8092

VL - 24

SP - 1435

EP - 1444

JO - Machine Vision and Applications

JF - Machine Vision and Applications

IS - 7

ER -