Semantic Context Forests for Learning-Based Knee Cartilage Segmentation in 3D MR Images

Wang, Quan; Wu, Dijia; Lü, Le; Liu, Meizhu; Boyer, Kim L.; Zhou, S. Kevin

doi:10.1007/978-3-319-05530-5_11

Cited by 10 publications

(7 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, an automatic method combining a bone ASM segmentation, a semantic context forest cartilage classifier, and smoothing by a graph cut optimization was presented at a workshop. 37 The method was validated on the OAI subcohort with manual segmentations by iMorphics also used in this paper. The Dice volume overlaps for the cartilages were between 0.79 and 0.85.…”

Section: Previous Results On the Validation Collectionsmentioning

confidence: 99%

“…Therefore, our results could possibly be improved by a regularization postprocessing step that refines the segmentation boundaries using either local or global shape information. This could be using a multiobject shape model or possibly by a more local regularization process like the graph cut optimization used by Wang et al 37 However, markers that are to be quantified from the segmentations may be robust to these fine-scale details. A cartilage volume marker will likely not be significantly affected and markers that are quantified with explicit or implicit regularization will also be robust to this, such as surface smoothness 38 and joint congruity.…”

Section: Segmentation Errorsmentioning

confidence: 99%

See 1 more Smart Citation

Automatic segmentation of high- and low-field knee MRIs using knee image quantification with data from the osteoarthritis initiative

Dam¹,

Lillholm²,

Marques³

et al. 2015

J. Med. Imag

View full text Add to dashboard Cite

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.

show abstract

Section: Previous Results On the Validation Collectionsmentioning

confidence: 99%

Section: Segmentation Errorsmentioning

confidence: 99%

Automatic segmentation of high- and low-field knee MRIs using knee image quantification with data from the osteoarthritis initiative

Dam¹,

Lillholm²,

Marques³

et al. 2015

J. Med. Imag

View full text Add to dashboard Cite

show abstract

“…Table 9 presents learning-based studies. The patellar cartilage (cPT) is the most difficult structure to segment, with a mean DSC of 79.16% on OAI dataset [179], while the structure that seems to attain better results is the femoral cartilage (cF).…”

Section: Methodsmentioning

confidence: 99%

“…Apart from the features mentioned above for knee and other medical images, many other hand-crafted features have been designed specifically for knee cartilage segmentation. Wang et al [179] designed various features to segment cartilage without relying on the BCI extraction step. They developed a learning-based bone segmentation from anatomical correspondence mesh points and calculated the 3D Euclidean distance from the voxels found on the bone boundaries.…”

Section: Semantic Context and Contextmentioning

confidence: 99%

A review on segmentation of knee articular cartilage: from conventional methods towards deep learning

Ebrahimkhani

Jaward

Cicuttini

et al. 2020

Artificial Intelligence in Medicine

View full text Add to dashboard Cite

In this paper, we review the state-of-the-art approaches for knee articular cartilage segmentation from conventional techniques to deep learning (DL) based techniques. Knee articular cartilage segmentation on magnetic resonance (MR) images is of great importance in early diagnosis of osteoarthritis (OA). Besides, segmentation allows estimating the articular cartilage loss rate which is utilised in clinical practice for assessing the disease progression and morphological changes. It has been traditionally applied in quantifying longitudinal knee OA progression pattern to detect and assess the articular cartilage thickness and volume. Topics covered include various image processing algorithms and major features of different segmentation techniques, feature computations and the performance evaluation metrics. This paper is intended to provide researchers with a broad overview of the currently existing methods in the field, as well as to highlight the shortcomings and potential considerations in the application at clinical practice. The survey showed that state-of-the-art techniques based on DL outperform the other segmentation methods. The analysis of the existing methods reveals that integration of DL-based algorithms with other traditional model-based approaches has achieved the best results (mean Dice similarity coefficient (DSC) between 85.8% and 90%).

show abstract

“…For example, in [4] and [2], methods are proposed for the annotation of the spine. In [5], a method is proposed to segment cartilage in the knee. In [6], a method is proposed to annotate the ribs.…”

mentioning

confidence: 99%

Anatomical triangulation: from sparse landmarks to dense annotation of the skeleton in CT images

Bieth

Donner

Schwaiger

et al. 2015

Procedings of the British Machine Vision Conference 2015

View full text Add to dashboard Cite

The automated annotation of bones that are visible in CT images of the skeleton is a challenging task which has, so far, been approached for only certain subregions of the skeleton, such as the spine. For example,in [4] and [2], methods are proposed for the annotation of the spine. In [5], a method is proposed to segment cartilage in the knee. In [6], a method is proposed to annotate the ribs. However, these methods are specific to a particular region of the skeleton and generalizing them to the whole skeleton, with a much larger field of view, a higher variability of the anatomy, and also variations in the positioning of the patient, remains an open task.In this paper, we propose an approach for fully automatic dense annotation of substructures of the whole skeleton in CT images for staging bone tumour patients, where we want to localize anatomical substructures in order to estimate the local tumour load as visible from PET/CT images. As shown in Fig. 1, our method contains four main steps: the preprocessing, a first Random Forest classifier, a parts-based model and a second Random Forest classifier. The last two steps can be iterated.Preprocessing: The bones are segmented from the CT image by adapting [3]. The intensities in the image are modelled as a mixture of two Gaussians and the resulting thresholding is regularized using graphcut. Landmarks distributed on the whole body are pre-filtered by a classifier and refined through a Hough regression model and a parts-based model of the global landmark topology as explained in [1].Anatomical triangulation: A Random Forest classifier provides a first annotation of the skeleton. It uses as features: the triangulation features (signed distance to each landmarks in each direction), geodesic distances to selected landmarks, mean image intensity and bone proportion around the center voxel, position in the image.Centroid regularization: The position of the centroid of each substructure of the skeleton is disambiguated by a Conditional Random Field (CRF), whose unary and binary potentials use global anatomical knowledge. The best candidate centroid for each substructure is computed by minimizing the energy of the CRF.

show abstract

Semantic Context Forests for Learning-Based Knee Cartilage Segmentation in 3D MR Images

Cited by 10 publications

References 21 publications

Automatic segmentation of high- and low-field knee MRIs using knee image quantification with data from the osteoarthritis initiative

Automatic segmentation of high- and low-field knee MRIs using knee image quantification with data from the osteoarthritis initiative

A review on segmentation of knee articular cartilage: from conventional methods towards deep learning

Anatomical triangulation: from sparse landmarks to dense annotation of the skeleton in CT images

Contact Info

Product

Resources

About