Automated vertebral landmarks and spinal curvature estimation using non-directional part affinity fields

Zhang, Cheng; Wang, Jun; Jian, He; Gao, Peng; Xie, Guotong

doi:10.1016/j.neucom.2020.05.120

Cited by 32 publications

(21 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These landmarks were manually annotated based on visual cues. We utilize two different training-test splits for evaluation, i.e., (1) Official Split 1 , where the dataset is split into 481 images for training and the rest 128 images for validation, as well as additional 98 challenge-private images with no annotations for test, (2) Consistent Split [9], where only the official training set is used and split into 431 for training and 50 for test, which is consistent with existing state-of-the-arts [2], [3], [9], [10].…”

Section: A Dataset and Evaluation Metricmentioning

confidence: 92%

“…To the best of our knowledge, it is the first try to utilize multi-stage cascaded CNNs for shape-constrained vertebral landmark localization instead of resorting to those overused heatmap-based approaches. 3) Extensive experimental results on the public dataset demonstrate that our method enjoys a benefit of much fewer false positives as well as missing landmarks, and thus achieves superior performance to other state-of-the-arts [2], [3], [5], [9], [10] by decreasing the relative error from 3.2e−3 to 7.2e−4.…”

Section: Introductionmentioning

confidence: 93%

“…Comparison. We compare our proposed framework with several state-of-the-art methods [2], [3], [5], [9], [10]. Among them, only Yi el al.…”

Section: ) Comparison With State-of-the-arts For Vertebral Landmark L...mentioning

confidence: 99%

“…Among them, only Yi el al. [5] have released their source code, therefore, we used , BoostNet [9] and SLSN [3] are borrowed from the paper of SLSN, and we all followed the same standard of data split.…”

Section: ) Comparison With State-of-the-arts For Vertebral Landmark L...mentioning

confidence: 99%

“…Those vertebral landmarks are then obtained by finding points with a local maximum response in the heatmap. Zhang et al [3] utilized four different fullyconnected layers to predict four heatmaps, each of which indicates a group of corner points with different semantics, i.e., top left, top right, bottom left, and bottom right points. The coordinates of landmarks are obtained indirectly by post-processing the predicted heatmap, i.e., finding local maximums' column and row indexes as landmark's coordinates.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Accurate Scoliosis Vertebral Landmark Localization on X-ray Images via Shape-constrained Multi-stage Cascaded CNNs

Wang¹,

Lv²,

Yang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Vertebral landmark localization is a crucial step for variant spine-related clinical applications, which requires detecting the corner points of 17 vertebrae. However, the neighbor landmarks often disturb each other for the homogeneous appearance of vertebrae, which makes vertebral landmark localization extremely difficult. In this paper, we propose multi-stage cascaded convolutional neural networks (CNNs) to split the single task into two sequential steps, i.e., center point localization to roughly locate 17 center points of vertebrae, and corner point localization to find 4 corner points for each vertebra without distracted by others. Landmarks in each step are located gradually from a set of initialized points by regressing offsets via cascaded CNNs. Principal Component Analysis (PCA) is employed to preserve a shape constraint in offset regression to resist the mutual attraction of vertebrae. We evaluate our method on the AASCE dataset that consists of 609 tight spinal anterior-posterior X-ray images and each image contains 17 vertebrae composed of the thoracic and lumbar spine for spinal shape characterization. Experimental results demonstrate our superior performance of vertebral landmark localization over other state-of-the-arts with the relative error decreasing from 3.2e−3 to 7.2e−4.

show abstract

Section: A Dataset and Evaluation Metricmentioning

confidence: 92%

Section: Introductionmentioning

confidence: 93%

“…Comparison. We compare our proposed framework with several state-of-the-art methods [2], [3], [5], [9], [10]. Among them, only Yi el al.…”

Section: ) Comparison With State-of-the-arts For Vertebral Landmark L...mentioning

confidence: 99%

Section: ) Comparison With State-of-the-arts For Vertebral Landmark L...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Accurate Scoliosis Vertebral Landmark Localization on X-ray Images via Shape-constrained Multi-stage Cascaded CNNs

Wang¹,

Lv²,

Yang³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

Definition of Normal Vertebral Morphometry Using NHANES‐II Radiographs

Hipp¹,

Grieco²,

Newman³

et al. 2022

JBMR Plus

View full text Add to dashboard Cite

A robust definition of normal vertebral morphometry is required to confidently identify abnormalities such as fractures. The Second National Health and Nutrition Examination Survey (NHANES-II) collected a nationwide probability sample to document the health status of the United States. Over 10,000 lateral cervical spine and 7,000 lateral lumbar spine X-rays were collected. Demographic, anthropometric, health, and medical history data were also collected. The coordinates of the vertebral body corners were obtained for each lumbar and cervical vertebra using previously validated, automated technology consisting of a pipeline of neural networks and coded logic. These landmarks were used to calculate six vertebral body morphometry metrics. Descriptive statistics were generated and used to identify and trim outliers from the data. Descriptive statistics were tabulated using the trimmed data for use in quantifying deviation from average for each metric. The dependency of these metrics on sex, age, race, nation of origin, height, weight, and body mass index (BMI) was also assessed. There was low variation in vertebral morphometry after accounting for vertebrae (eg, L1, L2), and the R 2 was high for ANOVAs. Excluding outliers, age, sex, race, nation of origin, height, weight, and BMI were statistically significant for most of the variables, though the F-statistic was very small compared to that for vertebral level. Excluding all variables except vertebra changed the ANOVA R 2 very little. Reference data were generated that could be used to produce standardized metrics in units of SD from mean. This allows for easy identification of abnormalities resulting from vertebral fractures, atypical vertebral body morphometries, and other congenital or degenerative conditions. Standardized metrics also remove the effect of vertebral level, facilitating easy interpretation and enabling data for all vertebrae to be pooled in research studies.

show abstract

Automated measurement of spine indices on axial MR images for lumbar spinal stenosis diagnosis using segmentation‐guided regression network

Pang

Lin

et al. 2022

Medical Physics

View full text Add to dashboard Cite

Purpose Automated measurement of spine indices on axial magnetic resonance (MR) images plays a significant role in lumbar spinal stenosis diagnosis. Existing direct spine indices measurement approaches fail to explicitly focus on the task‐specific region or feature channel with the additional information for guiding. We aim to achieve accurate spine indices measurement by introducing the guidance of the segmentation task. Methods In this paper, we propose a segmentation‐guided regression network (SGRNet) to achieve automated spine indices measurement. SGRNet consists of a segmentation path for generating the spine segmentation prediction and a regression path for producing spine indices estimation. The segmentation path is a U‐Net‐like network which includes a segmentation encoder and a decoder which generates multilevel segmentation features and segmentation prediction. The proposed segmentation‐guided attention module (SGAM) in the regression encoder extracts the attention‐aware regression feature under the guidance of the segmentation feature. Based on the attention‐aware regression feature, a fully connected layer is utilized to output the accurate spine indices estimation. Results Experiments on the open‐access Lumbar Spine MRI data set show that SGRNet achieves state‐of‐the‐art performance with a mean absolute error of 0.49 mm and mean Pearson correlation coefficient of 0.956 for four indices estimation. Conclusions The proposed SGAM in SGRNet is capable of improving the performance of spine indices measurement by focusing on the task‐specific region and feature channel under the guidance of the segmentation task.

show abstract

Automated vertebral landmarks and spinal curvature estimation using non-directional part affinity fields

Cited by 32 publications

References 6 publications

Accurate Scoliosis Vertebral Landmark Localization on X-ray Images via Shape-constrained Multi-stage Cascaded CNNs

Accurate Scoliosis Vertebral Landmark Localization on X-ray Images via Shape-constrained Multi-stage Cascaded CNNs

Definition of Normal Vertebral Morphometry Using NHANES‐II Radiographs

Automated measurement of spine indices on axial MR images for lumbar spinal stenosis diagnosis using segmentation‐guided regression network

Contact Info

Product

Resources

About