Pancreas Segmentation-Derived Biomarkers: Volume and Shape Metrics in the UK Biobank Imaging Study

Bagur, Alexandre Triay; Ridgway, Gerard R.; McGonigle, John; Brady, Michael; Bulte, Daniel

doi:10.1007/978-3-030-52791-4_11

Cited by 8 publications

(6 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2) Quantity of training data: For segmentation with neural networks, successful training on UK Biobank neckto-knee body MRI has been reported with annotated data of between 90 and 220 subjects [17], [15]. Each subject can effectively supply multiple training sample in the form of two-or three-dimensional patches.…”

Section: Resultsmentioning

confidence: 99%

“…Architectures such as the U-Net [11] can be trained on carefully prepared ground truth segmentation images of dozens or hundreds of subjects to automatically label structures and tissues in two-dimensional representations of medical imaging data. In UK Biobank body MRI, related techniques have been proposed for segmentation of various organs [12], [13], [14], [15], [16], muscles [17], and other tissues [18].…”

Section: A Segmentation and Regressionmentioning

confidence: 99%

See 1 more Smart Citation

Deep regression for uncertainty-aware and interpretable analysis of large-scale body MRI

Langner,

Strand,

Ahlström

et al. 2021

Preprint

View full text Add to dashboard Cite

Large-scale medical studies such as the UK Biobank examine thousands of volunteer participants with medical imaging techniques. Combined with the vast amount of collected metadata, anatomical information from these images has the potential for medical analyses at unprecedented scale. However, their evaluation often requires manual input and long processing times, limiting the amount of reference values for biomarkers and other measurements available for research. Recent approaches with convolutional neural networks for regression can perform these evaluations automatically. On magnetic resonance imaging (MRI) data of more than 40,000 UK Biobank subjects, these systems can estimate human age, body composition and more. This style of analysis is almost entirely data-driven and no manual intervention or guidance with manually segmented ground truth images is required. The networks often closely emulate the reference method that provided their training data and can reach levels of agreement comparable to the expected variability between established medical gold standard techniques. The risk of silent failure can be individually quantified by predictive uncertainty obtained from a mean-variance criterion and ensembling. Saliency analysis furthermore enables an interpretation of the underlying relevant image features and showed that the networks learned to correctly target specific organs, limbs, and regions of interest.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: A Segmentation and Regressionmentioning

confidence: 99%

Deep regression for uncertainty-aware and interpretable analysis of large-scale body MRI

Langner,

Strand,

Ahlström

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Manual analyses [4] and semi-automated multi-atlas segmentations [5] are being conducted, but require manual guidance or quality control amounting to weeks or months of work at the given quantity of data. For faster processing, automation with neural networks for segmentation has been proposed, yielding measurements of the heart [6], kidney [7], pancreas [8], [9], and liver [10], but also the iliopsoas muscles [11], spleen, adipose tissues and more [12].…”

Section: Introductionmentioning

confidence: 99%

Uncertainty-Aware Body Composition Analysis with Deep Regression Ensembles on UK Biobank MRI

Langner¹,

Gustafsson²,

Avelin³

et al. 2021

Preprint

View full text Add to dashboard Cite

To enable fast and automated analysis of body composition from UK Biobank MRI with accurate estimates of individual measurement errors. Methods: In an ongoing large-scale imaging study the UK Biobank has acquired MRI of over 40,000 men and women aged 44-82. Phenotypes derived from these images, such as body composition, can reveal new links between genetics, cardiovascular disease, and metabolic conditions. In this retrospective study, neural networks were trained to provide six measurements of body composition from UK Biobank neck-to-knee body MRI. A ResNet50 architecture can automatically predict these values by image-based regression, but may also produce erroneous outliers. Predictive uncertainty, which could identify these failure cases, was therefore modeled with a mean-variance loss and ensembling. Its estimates of individual prediction errors were evaluated in cross-validation on over 8,000 subjects, tested on another 1,000 cases, and finally applied for inference. Results: Relative measurement errors below 5% were achieved on all but one target, for intra-class correlation coefficients (ICC) above 0.97 both in validation and testing. Both mean-variance loss and ensembling yielded improvements and provided uncertainty estimates that highlighted some of the worst outlier predictions. Combined, they reached the highest quality, but also exhibited a consistent bias towards high uncertainty in heavyweight subjects. Conclusion: Mean-variance regression and ensembling provided complementary benefits for automated body composition measurements from UK Biobank MRI, reaching high speed and accuracy. These values were inferred for the entire cohort, with uncertainty estimates that can approximate the measurement errors and identify some of the worst outliers automatically.

show abstract

“…Computational image analysis, by which machine learning is used to annotate and segment the images, is gaining traction as a means of representing detailed three-dimensional (3D) mesh-derived phenotypes related to shape variations at thousands of vertices in a standardised coordinate space. One approach to inference is to transform the spatially correlated data into a smaller number of uncorrelated principal components [4], while the modes from PCA are useful in exploratory data analysis they do not provide an explicit model relating 3D shape to other phenotypic measures. A more powerful approach may be to estimate parameters at each vertex of the 3D surface mesh, hence creating a so-called statistical parametric map (SPM), a concept widely used in functional neuroimaging [20] and cardiac imaging [9].…”

Section: Introductionmentioning

confidence: 99%

Mass Univariate Regression Analysis for Three-Dimensional Liver Image-Derived Phenotypes

Thanaj

Basty

Liu

et al. 2021

Medical Image Understanding and Analysis

View full text Add to dashboard Cite

Image-derived phenotypes of abdominal organs from magnetic resonance imaging reveal variations in volume and shape and may be used to model changes in a normal versus pathological organ and improve diagnosis. Computational atlases of anatomical organs provide many advantages in quantifying and modeling differences in shape and size of organs for population imaging studies. Here we made use of liver segmentations derived from Dixon MRI for 2,730 UK Biobank participants to create 3D liver meshes. We computed the signed distances between a reference and subject-specific meshes to define the surface-tosurface (S2S) phenotype. We employed mass univariate regression analysis to compare the S2S values from the liver meshes to image-derived phenotypes specific to the liver, such as proton density fat fraction and iron concentration while adjusting for age, sex, ethnicity, body mass index and waist-to-hip ratio. Vertex-based associations in the 3D liver mesh were extracted and threshold-free cluster enhancement was applied to improve the sensitivity and stability of the statistical parametric maps. Our findings show that the 3D liver meshes are a robust method for modeling the association between anatomical, anthropometric, and phenotypic variations across the liver. This approach may be readily applied to different clinical conditions as well as extended to other abdominal organs in a larger population.

show abstract

Pancreas Segmentation-Derived Biomarkers: Volume and Shape Metrics in the UK Biobank Imaging Study

Cited by 8 publications

References 21 publications

Deep regression for uncertainty-aware and interpretable analysis of large-scale body MRI

Deep regression for uncertainty-aware and interpretable analysis of large-scale body MRI

Uncertainty-Aware Body Composition Analysis with Deep Regression Ensembles on UK Biobank MRI

Mass Univariate Regression Analysis for Three-Dimensional Liver Image-Derived Phenotypes

Contact Info

Product

Resources

About