Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.
No entries.
Section 2 · machine-generated
Variations
Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.
Other field of use
Aortic calcium measurement in veterinary studies
A three-dimensional convolutional network locates the aortic valve center and estimates the valve plane in noncontrast canine thoracic computed tomography volumes. A second network segments the aortic root in a crop aligned to that plane and sized to the estimated annular diameter. A scoring module counts voxels within the root mask that exceed an attenuation threshold calibrated on labeled canine scans and multiplies the count by the physical voxel volume.
Component substitution
Graph cuts for root segmentation
A convolutional localizer supplies the valve center and plane for an oriented three-dimensional crop. An operator marks a few root and background voxels on multiplanar views, and a graph-cut algorithm estimates foreground and background attenuation distributions from those marks. The graph assigns a lower separation cost to adjacent voxels with larger attenuation differences and partitions the crop into root and background. A separate attenuation threshold identifies calcified voxels inside the root mask for scoring.
Combination with existing systems
Calcium scoring in an imaging archive
A medical imaging archive routes eligible noncontrast cardiac computed tomography series to a scoring service using series metadata and a configured acquisition filter. The service locates the valve, segments the aortic root, and computes the score from calcified voxels inside the mask. It writes a structured measurement and a segmentation overlay linked to the source images so an imaging workstation can display the score and its contributing voxels.
Automated or data-driven variant
Automatic review routing using model agreement
A valve localizer defines a shared crop for three independently trained root segmentation networks. The scoring software applies the same attenuation threshold to each mask on the original voxel grid and calculates the calcium volume for each result. It reports the median volume and queues the scan for review when the spread among volumes or the disagreement among calcium masks exceeds limits set on a labeled validation set.
Automatic separation of adjacent calcium deposits
After valve localization, a convolutional segmentation network labels the valve leaflets, aortic wall, coronary origins, and mitral annulus in a valve-aligned crop. The software collects voxels above the calcium attenuation threshold and groups face-connected voxels into deposits on the original scan grid. For deposits that cross label boundaries, the software assigns each voxel to the structure with the highest predicted probability. The valve score uses only the voxels assigned to the leaflets.
Cite aspatent.fyi, US 2026/0354783 A1, claim 1, first published 2026-11-19T03:31:31-05:00, https://patent.fyi/US20260354783/, SHA-256 eee27cc18b9a7d1eb51d49365a388b22aa81be2c74fa5ccbede0d008860e972d