Segmenting Articular Cartilage Automatically Using a Voxel Classification Approach

Abstract

Abstract-We present a fully automatic method for articular cartilage segmentation from magnetic resonance imaging (MRI) which we use as the foundation of a quantitative cartilage assessment. We evaluate our method by comparisons to manual segmentations by a radiologist and by examining the interscan reproducibility of the volume and area estimates. Training and evaluation of the method is performed on a data set consisting of 139 scans of knees with a status ranging from healthy to severely osteoarthritic. This is, to our knowledge, the only fully automatic cartilage segmentation method that has good agreement with manual segmentations, an interscan reproducibility as good as that of a human expert, and enables the separation between healthy and osteoarthritic populations. While high-field scanners offer high-quality imaging from which the articular cartilage have been evaluated extensively using manual and automated image analysis techniques, low-field scanners on the other hand produce lower quality images but to a fraction of the cost of their high-field counterpart. For low-field MRI, there is no well-established accuracy validation for quantitative cartilage estimates, but we show that differences between healthy and osteoarthritic populations are statistically significant using our cartilage volume and surface area estimates, which suggests that low-field MRI analysis can become a useful, affordable tool in clinical studies.

Automatic Articular Cartilage Segmentation Based on Pattern Recognition from Knee MRI Images

“…Based on the experiments and the literature [1,3], the following features were finally determined for use in pixel classification: the intensity f(x,y); the distance from the BCI d(x,y) Fig. 5 Optimized results of edge detection Fig.…”

“…5 Optimized results of edge detection Fig. 6 Results of corner detection [1]; the gradient norm determined by the Prewitt operator |∇f(x,y)| and position o(x,y) [3]; and the intensity variance of the eight neighboring pixels σ 8 (x,y). All features were calculated after denoising by a Gaussian low-pass filter in the section of image preprocessing.…”

“…Folkesson et al designed a cartilage pixel classifier based on the k-nearest neighbor method using pixel features for lowfield MRI images. In her study, 139 knee joints were examined with T1 sequences, and the mean Dice similarity coefficient (DSC) was determined to be 0.80 [3]. Zhang designed a support vector machine (SVM) model using the gray value of pixels, cartilage thickness, the mean curvature of the boundary, and spatial features for cartilage segmentation, and the model was validated based on a dataset of multiple MR sequences, including FS SPGR, FIESTA, and IDEAL GRE, whose mean DSC was determined to be 0.913 [4].…”

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

“…During training, the number of candidates at each non-leaf node is set to 1000. The dice similarity coefficient (DSC) is used to measure the performance of our method since it is commonly reported in previous literature [2,4,5,6,23].…”

Voxel classification and graph cuts for automated segmentation of pathological periprosthetic hip anatomy

“…Similar to previous studies [25][26][27] we used multiscale image and image gradient values as features. Isotropic Gaussian low-pass (blurring) with standard deviations of 0.5, 1 and 2 mm, along with their image gradient magnitudes, were computed from the original CT volume.…”