Medical imaging
Grading prenatal hydronephrosis from ultrasound
Can a pipeline use segmented kidneys across several ultrasound views to estimate a subject-level severity grade?
Expert Systems with Applications · 2024
Approach
The study curated 2,062 ultrasound images from 163 subjects. The workflow locates the kidney region, outlines it with KUSNet, and predicts one of five severity grades with PHCNet. Weighted voting combines grades from different ultrasound views at the subject level.
Reported results
The paper reports a 92.1% Dice score for kidney segmentation. For five-grade classification with segmented views, Table 3 reports 93.9% subject-level accuracy and an 89.0% F1 score. The two tasks were evaluated separately.
- Kidney segmentation Dice
- 92.1%
- Subject-level grading accuracy
- 93.9%
- Subjects in the curated cohort
- 163
My contribution
I contributed to study design and data curation, developed the methods and software, and wrote the original draft with the research team.
Study scope
The study used five subject-independent folds on the retrospective QU–HMC–AHN cohort. Reported grading performance reflects the curated views and reference grades in that cohort. Prospective external validation would establish how the approach performs across sites and acquisition protocols.
Illustrated example
From ultrasound views to one severity grade
- Locate the kidney
Find the kidney region in each view.
- Segment and grade each view
KUSNet outlines the kidney. PHCNet grades the segmented views. In this example all three give grade II.
- Combine the grades
Weighted voting gives one subject-level result: Society for Fetal Urology grade II.