Kidney segmentation in neck-to-knee body MRI of 40,000 UK Biobank participants

Abstract

The UK Biobank is collecting extensive data on health-related characteristics of over half a million volunteers. The biological samples of blood and urine can provide valuable insight on kidney function, with important links to cardiovascular and metabolic health. Further information on kidney anatomy could be obtained by medical imaging. In contrast to the brain, heart, liver, and pancreas, no dedicated Magnetic Resonance Imaging (MRI) is planned for the kidneys. An image-based assessment is nonetheless feasible in the neck-to-knee body MRI intended for abdominal body composition analysis, which also covers the kidneys. In this work, a pipeline for automated segmentation of parenchymal kidney volume in UK Biobank neck-to-knee body MRI is proposed. The underlying neural network reaches a relative error of 3.8%, with Dice score 0.956 in validation on 64 subjects, close to the 2.6% and Dice score 0.962 for repeated segmentation by one human operator. The released MRI of about 40,000 subjects can be processed within one day, yielding volume measurements of left and right kidney. Algorithmic quality ratings enabled the exclusion of outliers and potential failure cases. The resulting measurements can be studied and shared for large-scale investigation of associations and longitudinal changes in parenchymal kidney volume.

Figure 1

Among all UK Biobank subjects two subsets, A and B, were manually segmented. A neural network was evaluated in two cross-validation experiments on these and applied for inference to all remaining subjects. After excluding 5% of results in two quality control stages, about 37,500 measurements remain as final result.

Figure 2

Segmented parenchymal tissue of right (red) and left (blue) kidney in MRI of a male subject.

Figure 3

Main validation result for 64 subjects of dataset A, with images from dataset B added for training. The diagonal line in the scatter plot on the left represents a hypothetical perfect result, whereas the dashed lines in the Bland–Altman plot on the right give the 95% Limits of Agreement. When compared to the reference, the network appears to emulate a tendency towards oversegmentation which is also seen in Table 1 for the operators who provided reference segmentations for B.

Figure 4

Distribution of algorithmic quality ratings by subject, sorted separately for each rating. High values of each cost term indicate low quality. In stage one (a–c) and stage two (d,e) of quality controls, the highlighted top one or two percent of subjects were accordingly flagged for exclusion as potential failure cases.

Figure 5

Inferred UK Biobank parenchymal kidney volume (left + right) in ${\text{cm}}^3$ for 17,846 male and 19,622 female subjects.