Accurate Pelvis and Femur Segmentation in Hip CT With a Novel Patch-Based Refinement

Chang, Yun Hee; Yuan, Ye; Guo, Cheng; Wang, Yadong; Cheng, Yuanzhi; Tamura, Shinichi

doi:10.1109/jbhi.2018.2834551

Cited by 21 publications

(20 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The findings of this study may be enhanced by automating the segmentation of volumes and derivation of ratios. Automatic segmentation of the femur has been widely reported on in recent years and indicates strong potential for our volume ratio measurements to become automated [14][15][16][17][18]. This meets the last of the criteria we set forth for this method.…”

Section: Discussionmentioning

confidence: 64%

Improving the quantitative classification of Erlenmeyer flask deformities

et al. 2020

View full text Add to dashboard Cite

The Erlenmeyer flask deformity is a common skeletal modeling deformity, but current classification systems are binary and may restrict its utility as a predictor of associated skeletal conditions. A quantifiable 3-point system of severity classification could improve its predictive potential in disease. Ratios were derived from volumes of regions of interests drawn in 50 Gaucher's disease patients. ROIs were drawn from the distal physis to 2 cm proximal, 2 cm to 4 cm, and 4 cm to 6 cm. Width was also measured at each of these boundaries. Two readers rated these 100 femurs using a 3-point scale of severity classification. Weighted kappa indicated reliability and one-way analysis of variance characterized ratio differences across the severity scale. Accuracy analyses allowed determination of clinical cutoffs for each ratio. Pearson's correlations assessed the associations of volume and width with a shape-based concavity metric of the femur. The volume ratio incorporating the metaphyseal region from 0 to 2 cm and the diametaphyseal region at 4-6 cm was most accurate at distinguishing femurs on the 3-point scale. Receiver operating characteristic curves for this ratio indicated areas of 0.95 to distinguish normal and mild femurs and 0.93 to distinguish mild and severe femurs. Volume was moderately associated with the degree of femur concavity. The proposed volume ratio method is an objective, proficient method at distinguishing severities of the Erlenmeyer flask deformity with the potential for automation. This may have application across diseases associated with the deformity and deficient osteoclast-mediated modeling of growing bone.

show abstract

Section: Discussionmentioning

confidence: 64%

Improving the quantitative classification of Erlenmeyer flask deformities

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Compared to traditional CNN nets and manual segmentation [ 18 ], the segmentation time of the diseased hip joints using our proposed method was significantly reduced. Shinichi et al [ 4 ] showed a coarse-to-fine hip CT segmentation framework that consisted of regional growth-based preprocessing, conditional random field-based coarse segmentation, and patch-based refinement. Radiology experts expend considerable effort in completing the training samples.…”

Section: Discussionmentioning

confidence: 99%

“…Preoperative computed tomography (CT)-based 3D planning is essential for total hip arthroplasty. Precise localisation and segmentation of the hip joint on CT images are necessary to simulate the placement of joint implants [ 4 , 5 ]. CT images from diseased hips exhibit image degradation, noise, non-homogeneous intensities and obscure boundaries between the femoral head and acetabulum; because of these features, automatic CT hip-joint segmentation is challenging [ 6 , 7 ].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Utility of a novel integrated deep convolutional neural network for the segmentation of hip joint from computed tomography images in the preoperative planning of total hip arthroplasty

et al. 2022

View full text Add to dashboard Cite

Purpose Preoperative three-dimensional planning is important for total hip arthroplasty. To simulate the placement of joint implants on computed tomography (CT), pelvis and femur must be segmented. Accurate and rapid segmentation of the hip joint is challenging. This study aimed to develop a novel deep learning network, named Changmugu Net (CMG Net), which could achieve accurate segmentation of the femur and pelvis. Methods The overall deep neural network architecture of CMG Net employed three interrelated modules. CMG Net included the 2D U-net to separate the bony and soft tissues. The modular hierarchy method was used for the main femur segmentation to achieve better performance. A layer classifier was adopted to localise femur layers among a series of CT scan images. The first module was a modified 2D U-net, which separated bony and soft tissues; it provided intermediate supervision for the main femur segmentation. The second module was the main femur segmentation, which was used to distinguish the femur from the acetabulum. The third module was the layer classifier, which served as a post-processor for the second module. Results There was a much greater overlap in accuracy results with the “gold standard” segmentation than with competing networks. The dice overlap coefficient was 93.55% ± 5.57%; the mean surface distance was 1.34 ± 0.24 mm, and the Hausdorff distance was 4.19 ± 1.04 mm in the normal and diseased hips, which indicated greater accuracy than the other four competing networks. Moreover, the mean segmentation time of CMG Net was 25.87 ± 2.73 s, which was shorter than the times of the other four networks. Conclusions The prominent segmentation accuracy and run-time of CMG Net suggest that it is a reliable method for clinicians to observe anatomical structures of the hip joints, even in severely diseased cases.

show abstract

“…A number of methods have been implemented to surmount the problem of bone segmentation, ranging from thresholding techniques to graph-cut methods. [8][9][10][11] Current segmentation methods often require a "user-in-theloop" paradigm in order to manually correct segmentations to produce acceptable masks for FE modeling and/or their processing time is too long for clinical use. This lack of robustness is costly in terms of time and the need for a highly trained specialist to manually correct the segmentations.…”

Section: Introductionmentioning

confidence: 99%

Automated femur segmentation from computed tomography images using a deep neural network

Bjornsson,

Helgason,

Palsson

et al. 2021

Preprint

View full text Add to dashboard Cite

Osteoporosis is a common bone disease that occurs when the creation of new bone does not keep up with the loss of old bone, resulting in increased fracture risk. Adults over the age of 50 are especially at risk and see their quality of life diminished because of limited mobility, which can lead to isolation and depression. We are developing a robust screening method capable of identifying individuals predisposed to hip fracture to address this clinical challenge. The method uses finite element analysis and relies on segmented computed tomography (CT) images of the hip. Presently, the segmentation of the proximal femur requires manual input, which is a tedious task, prone to human error, and severely limits the practicality of the method in a clinical context. Here we present a novel approach for segmenting the proximal femur that uses a deep convolutional neural network to produce accurate, automated, robust, and fast segmentations of the femur from CT scans. The network architecture is based on the renowned u-net, which consists of a downsampling path to extract increasingly complex features of the input patch and an upsampling path to convert the acquired low resolution image into a high resolution one. Skipped connections allow us to recover critical spatial information lost during downsampling. The model was trained on 30 manually segmented CT images and was evaluated on 200 ground truth manual segmentations. Our method delivers a mean Dice similarity coefficient (DSC) and 95 th percentile Hausdorff distance (HD95) of 0.990 and 0.981 mm, respectively.

show abstract

Accurate Pelvis and Femur Segmentation in Hip CT With a Novel Patch-Based Refinement

Cited by 21 publications

References 29 publications

Improving the quantitative classification of Erlenmeyer flask deformities

Improving the quantitative classification of Erlenmeyer flask deformities

Utility of a novel integrated deep convolutional neural network for the segmentation of hip joint from computed tomography images in the preoperative planning of total hip arthroplasty

Automated femur segmentation from computed tomography images using a deep neural network

Contact Info

Product

Resources

About