Automatic analysis of trabecular bone structure from knee MRI

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningfagfællebedømt

Joselene Marques, Rabia Granlund, Martin Lillholm, Paola C. Pettersen, Erik B. Dam

We investigated the feasibility of quantifying osteoarthritis (OA) by analysis of the trabecular bone
structure in low-¿eld knee MRI. Generic texture features were extracted from the images and
subsequently selected by sequential ¿oating forward selection (SFFS), following a fully automatic,
uncommitted machine-learning based framework. Six different classi¿ers were evaluated in crossvalidation schemes and the results showed that the presence of OA can be quanti¿ed by a bone
structure marker. The performance of the developed marker reached a generalization area-under-theROC (AUC) of 0.82, which is higher than the established cartilage markers known to relate to the OA diagnosis.
OriginalsprogEngelsk
TidsskriftComputers in Biology and Medicine
Vol/bind42
Udgave nummer7
Sider (fra-til)735-742
Antal sider8
ISSN0010-4825
DOI
StatusUdgivet - 2012

ID: 40995413