Automatic segmentation of the glenohumeral cartilages from magnetic resonance images

Neubert, A., Yang, Z., Engstrom, C., Xia, Y., Strudwick, M. W., Chandra, S. S., Fripp, J. and Crozier, S. (2016) Automatic segmentation of the glenohumeral cartilages from magnetic resonance images. Medical Physics, 43 10: 5370-5379. doi:10.1118/1.4961011


Author Neubert, A.
Yang, Z.
Engstrom, C.
Xia, Y.
Strudwick, M. W.
Chandra, S. S.
Fripp, J.
Crozier, S.
Title Automatic segmentation of the glenohumeral cartilages from magnetic resonance images
Journal name Medical Physics   Check publisher's open access policy
ISSN 0094-2405
2473-4209
Publication date 2016-10-01
Year available 2016
Sub-type Article (original research)
DOI 10.1118/1.4961011
Open Access Status Not yet assessed
Volume 43
Issue 10
Start page 5370
End page 5379
Total pages 10
Place of publication Melville, NY United States
Publisher A I P Publishing
Language eng
Subject 1304 Biophysics
2741 Radiology Nuclear Medicine and imaging
Formatted abstract
Purpose: Magnetic resonance (MR) imaging plays a key role in investigating early degenerative disorders and traumatic injuries of the glenohumeral cartilages. Subtle morphometric and biochemical changes of potential relevance to clinical diagnosis, treatment planning, and evaluation can be assessed from measurements derived from in vivo MR segmentation of the cartilages. However, segmentation of the glenohumeral cartilages, using approaches spanning manual to automated methods, is technically challenging, due to their thin, curved structure and overlapping intensities of surrounding tissues. Automatic segmentation of the glenohumeral cartilages from MR imaging is not at the same level compared to the weight-bearing knee and hip joint cartilages despite the potential applications with respect to clinical investigation of shoulder disorders. In this work, the authors present a fully automated segmentation method for the glenohumeral cartilages using MR images of healthy shoulders.
Methods: The method involves automated segmentation of the humerus and scapula bones using 3D active shape models, the extraction of the expected bone–cartilage interface, and cartilage segmentation using a graph-based method. The cartilage segmentation uses localization, patient specific tissue estimation, and a model of the cartilage thickness variation. The accuracy of this method was experimentally validated using a leave-one-out scheme on a database of MR images acquired from 44 asymptomatic subjects with a true fast imaging with steady state precession sequence on a 3 T scanner (Siemens Trio) using a dedicated shoulder coil. The automated results were compared to manual segmentations from two experts (an experienced radiographer and an experienced musculoskeletal anatomist) using the Dice similarity coefficient (DSC) and mean absolute surface distance (MASD) metrics.
Results: Accurate and precise bone segmentations were achieved with mean DSC of 0.98 and 0.93 for the humeral head and glenoid fossa, respectively. Mean DSC scores of 0.74 and 0.72 were obtained for the humeral and glenoid cartilage volumes, respectively. The manual interobserver reliability evaluated by DSC was 0.80±0.03 and 0.76±0.04 for the two cartilages, implying that the automated results were within an acceptable 10% difference. The MASD between the automatic and the corresponding manual cartilage segmentations was less than 0.4 mm (previous studies reported mean cartilage thickness of 1.3 mm).
Conclusions: This work shows the feasibility of volumetric segmentation and separation of the glenohumeral cartilages from MR images. To their knowledge, this is the first fully automated algorithm for volumetric segmentation of the individual glenohumeral cartilages from MR images. The approach was validated against manual segmentations from experienced analysts. In future work, the approach will be validated on imaging datasets acquired with various MR contrasts in patients.
Keyword glenohumeral cartilage
image segmentation
magnetic resonance imaging
morphological analysis
statistical shape models
Q-Index Code C1
Q-Index Status Provisional Code
Institutional Status UQ

 
Versions
Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 0 times in Thomson Reuters Web of Science Article
Scopus Citation Count Cited 0 times in Scopus Article
Google Scholar Search Google Scholar
Created: Wed, 05 Oct 2016, 21:50:08 EST by Sandrine Ducrot on behalf of Learning and Research Services (UQ Library)