Abstract:This paper presents a fully automated method for segmenting articular knee cartilage and bone from in vivo 3-D dual echo steady state images. The magnetic resonance imaging (MRI) datasets were obtained from the Osteoarthritis Initiative (OAI) pilot study and include longitudinal images from controls and subjects with knee osteoarthritis (OA) scanned twice at each visit (baseline, 24 month). Initially, human experts segmented six MRI series. Five of the six resultant sets served as reference atlases for a multi… Show more
“…18,36,[38][39][40][41][42] Other methods achieve a similar coarse-to-fine effect by applying atlas-based registration before either nonrigid registration 21,40,43 and/or classifier-based segmentation. 10,22,43 In general, it appears that integration of global information and local features is essential for solving the challenging problem of segmentation of cartilage on the background of multiple other tissue types. This explicitly or implicitly allows a zoom to several, simpler, local segmentation tasks such as cartilage versus subchondral bone, cartilage versus meniscus, cartilage versus cartilage, and cartilage versus synovial fluid.…”
Abstract. Clinical studies including thousands of magnetic resonance imaging (MRI) scans offer potential for pathogenesis research in osteoarthritis. However, comprehensive quantification of all bone, cartilage, and meniscus compartments is challenging. We propose a segmentation framework for fully automatic segmentation of knee MRI. The framework combines multiatlas rigid registration with voxel classification and was trained on manual segmentations with varying configurations of bones, cartilages, and menisci. The validation included high-and low-field knee MRI cohorts from the Center for Clinical and Basic Research, the osteoarthritis initiative (QAI), and the segmentation of knee images10 (SKI10) challenge. In total, 1907 knee MRIs were segmented during the evaluation. No segmentations were excluded. Our resulting OAI cartilage volume scores are available upon request. The precision and accuracy performances matched manual reader re-segmentation well. The cartilage volume scan-rescan precision was 4.9% (RMS CV). The Dice volume overlaps in the medial/lateral tibial/femoral cartilage compartments were 0.80 to 0.87. The correlations with volumes from independent methods were between 0.90 and 0.96 on the OAI scans. Thus, the framework demonstrated precision and accuracy comparable to manual segmentations. Finally, our method placed second for cartilage segmentation in the SKI10 challenge. The comprehensive validation suggested that automatic segmentation is appropriate for cohorts with thousands of scans.
“…18,36,[38][39][40][41][42] Other methods achieve a similar coarse-to-fine effect by applying atlas-based registration before either nonrigid registration 21,40,43 and/or classifier-based segmentation. 10,22,43 In general, it appears that integration of global information and local features is essential for solving the challenging problem of segmentation of cartilage on the background of multiple other tissue types. This explicitly or implicitly allows a zoom to several, simpler, local segmentation tasks such as cartilage versus subchondral bone, cartilage versus meniscus, cartilage versus cartilage, and cartilage versus synovial fluid.…”
Abstract. Clinical studies including thousands of magnetic resonance imaging (MRI) scans offer potential for pathogenesis research in osteoarthritis. However, comprehensive quantification of all bone, cartilage, and meniscus compartments is challenging. We propose a segmentation framework for fully automatic segmentation of knee MRI. The framework combines multiatlas rigid registration with voxel classification and was trained on manual segmentations with varying configurations of bones, cartilages, and menisci. The validation included high-and low-field knee MRI cohorts from the Center for Clinical and Basic Research, the osteoarthritis initiative (QAI), and the segmentation of knee images10 (SKI10) challenge. In total, 1907 knee MRIs were segmented during the evaluation. No segmentations were excluded. Our resulting OAI cartilage volume scores are available upon request. The precision and accuracy performances matched manual reader re-segmentation well. The cartilage volume scan-rescan precision was 4.9% (RMS CV). The Dice volume overlaps in the medial/lateral tibial/femoral cartilage compartments were 0.80 to 0.87. The correlations with volumes from independent methods were between 0.90 and 0.96 on the OAI scans. Thus, the framework demonstrated precision and accuracy comparable to manual segmentations. Finally, our method placed second for cartilage segmentation in the SKI10 challenge. The comprehensive validation suggested that automatic segmentation is appropriate for cohorts with thousands of scans.
“…Rather, the performance of their segmentation methods was assessed by using a leaveone-out method. 4,7 In our study, with a large number of training cases, we were able to use 40 cases as the cross validation set. We believe this approach strengthens and improves the rigor for the development and performance test of our proposed scheme.…”
Section: Discussionmentioning
confidence: 99%
“…However, several groups recently proposed "fully automated methods with no user interaction" for segmenting cartilage and bone from MR images. [4][5][6][7][8] Folkesson et al proposed k-nearest neighbor framework to perform tissue classification by selecting features such as voxel position, raw and Gaussian smoothed intensities, and intensity derivatives. 8 Fripp et al presented a segmentation scheme with three-dimensional active shape models.…”
Section: Introductionmentioning
confidence: 99%
“…12,13 This limitation may be overcome, though, by using a multi-atlas segmentation approach, 14,15 derived from separately registering multiple atlases to the target and resolving voxel-wise label conflicts between the registered atlases through label fusion. While Tamez-Pena et al reported promising results using the multi-atlas based cartilage segmentation algorithm, 4 they incorporated only a simple labelfusion approach and did not use any further correction methods to address abnormalities, such as osteophytes that form along joint margins. In the label fusion, it is important to find the corresponding position in multiple atlases, which in previous approaches has been accomplished by using the signal intensity of voxels.…”
Purpose: To develop a fully automated method to segment cartilage from the magnetic resonance (MR) images of knee and to evaluate the performance of the method on a public, open dataset. Methods: The segmentation scheme consisted of three procedures: multiple-atlas building, applying a locally weighted vote (LWV), and region adjustment. In the atlas building procedure, all training cases were registered to a target image by a nonrigid registration scheme and the best matched atlases selected. A LWV algorithm was applied to merge the information from these atlases and generate the initial segmentation result. Subsequently, for the region adjustment procedure, the statistical information of bone, cartilage, and surrounding regions was computed from the initial segmentation result. The statistical information directed the automated determination of the seed points inside and outside bone regions for the graph-cut based method. Finally, the region adjustment was conducted by the revision of outliers and the inclusion of abnormal bone regions. Results: A total of 150 knee MR images from a public, open dataset (available at www.ski10.org) were used for the development and evaluation of this approach. The 150 cases were divided into the training set (100 cases) and the test set (50 cases). The cartilages were segmented successfully in all test cases in an average of 40 min computation time. The average dice similarity coefficient was 71.7% ± 8.0% for femoral and 72.4% ± 6.9% for tibial cartilage.
Conclusions:The authors have developed a fully automated segmentation program for knee cartilage from MR images. The performance of the program based on 50 test cases was highly promising.
“…In the last decade, scoring of BMLs in MRI systematically appears in the various osteoarthritis knee evaluation systems contributing to the final grade of an overall knee condition [9][10][11][12][13] commonly used in longitudinal studies. In parallel, the combination of MRI and image processing techniques has allowed several semiautomatic and automatic systems to be developed for articular tissue segmentation including bone [14][15][16][17][18] , cartilage [19][20][21][22][23] , menisci [24] , and synovitis [25] for quantitative and semi-quantitative evaluation. However, very few technologies were developed looking at automatically assessing the volume of BMLs.…”
Background/Objective: Bone marrow lesions (BMLs) have been associated with pain and cartilage degeneration in patients with knee osteoarthritis; their specific detection and quantification is therefore of primary importance. This study aimed at developing a fully automated quantitative BML assessment technology for human knee osteoarthritis using magnetic resonance images (MRI) and two sequences, a T1/T2*-weighted gradient echo (DESS) and a water-sensitive intermediate-weighted turbo spin echo (IW-TSE).
Methods:The automated BML quantification first characterizes the bone and cartilage domains in the DESS sequence using our already published automated technology, then proceeds to the BML quantification which was developed as a four-stage process: selection of structured bright areas corresponding to BMLs, geometric filtering of unrelated structures, segmentation of the BML, and quantification of BML proportion within bone regions. For the IW-TSE sequence, the first step consists of the transfer of the bone and cartilage objects from the DESS to the IW-TSE images, followed by the BML detection and quantification as for the DESS. Validation was performed on 154 OA patients from a subset of the Osteoarthritis Initiative (OAI) cohort (public data sets) in which BML manual segmentation intra-and inter-reader reliability was done for each sequence (DESS and IW-TSE) using the intraclass correlation (ICC). BML comparison between the newly developed automated method with a manual segmentation was performed with ICC for the proportion of BML and Dice similarity coefficient (DSC) for BML localization and geometric extent. Finally, comparisons between the DESS and the IW-TSE sequences were performed for BML incidence and proportion.Results: Excellent to very good correlations were obtained for both MRI sequences for intra-and inter-reader reliability of the manual BML segmentation. Comparison between the developed automated method and the manual BML segmentation showed excellent to very good correlations in the global knee and regions (ICC=0.99 to 0.68 for DESS and 0.99 to 0.77 for IW-TSE sequences) as well as very good to good similarity for the BML geometrical agreement (DESS, 0.60 to 0.41; IW-TSE, 0.59 to 0.41). Data revealed greater BML incidence at the sites of high articular constraints: lateral femoropatellar and medial tibiofemoral articulation. Average BML proportion revealed a scaling factor of about 4.5-fold more for the IW-TSE compared to the DESS. www.sciedu.ca/jbgc Journal of Biomedical Graphics and Computing, 2013, Vol. 3, No. 1 ISSN 1925-4008 E-ISSN 1925 52 Conclusions: The newly developed fully automated MRI based BML assessment technology not only detects the absence/ presence of these pathological signals in the osteoarthritic human knee, but also provides accurate quantitative assessment of BMLs in the global knee and knee regions. Such automated system will enable large scale studies to be conducted within shorter durations, as well as increase stability of the reading.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.