An essential imaging biomarker for autosomal dominant polycystic kidney disease (ADPKD) is total kidney volume (TKV). However, manual TKV computation takes a lot of time and effort, especially when exophytic cysts are excluded.

In order to selectively segment kidney areas while avoiding exophytic cysts, researchers created a completely automated segmentation approach for TKV. They used abdominal T2-weighted magnetic resonance images from 210 ADPKD patients who were split into 2 groups: a training group of 157 and a testing group of 53. The network was trained via K-fold cross-validation with a 3D U-Net architecture using dataset fingerprints, where 80% of the 157 examples were used for training, and the remaining 20% were used for validation. They evaluated the performance of the automated segmentation approach to the manual method using the Dice similarity coefficient, intraclass correlation coefficient, and Bland-Altman analysis.

The test datasets included kidney volumes ranging from 178.9 to 2,776.0 ml (mean±SD, 1,058.5±706.8 ml) and exophytic cysts ranging from 113.4 to 2497.6 ml (mean±SD, 549.0±559.1 ml), the automated and manual reference methods showed excellent geometric concordance (Dice similarity coefficient: mean±SD, 0.962±0.018) with a minimum bias of 2.424 ml (95% limits of agreement, -49.80 to 44.95) and an intraclass correlation coefficient of 0.9994 (95% CI, 0.9991 to 0.9996; P< 0.001), the study was statistically significant.

They created a segmentation approach for measuring TKV that was totally automated, eliminated exophytic cysts, and had accuracy comparable to that of a human expert. The method could be helpful in clinical trials that need to automatically compute the TKV to assess the progression of ADPKD and therapy response.