Advancing a tracked two-dimensional ultrasound system for assessing dysplasia metrics toward clinical deployment

Developmental Dysplasia of the Hip (DDH) is commonly screened using 2D ultrasound (2DUS), but diagnostic metrics such as the alpha (α) angle and femoral head coverage (FHC) have been shown to be highly variable due to operator experience, probe placement, and patient positioning. Although 3D ultrasound (3DUS) offers improved reliability, its high cost and limited availability restrict widespread clinical use. To address this gap, we developed an Optically Tracked Ultrasound (OTUS) system that reconstructs 3D volumes from tracked sweeps of 2DUS images. While an initial OTUS prototype demonstrated feasibility in a clinical pilot study, clinical integration was limited by reliance on a Non-Planar marker system with strict line-of-sight requirements. In this study, we evaluate the accuracy and usability of a redesigned planar marker system intended to improve clinical practicality by improving robustness to partial occlusion. We conducted a clinical pilot study acquiring OTUS scans of 47 hips from 30 infants (mean age 8.02 weeks, range 4.42-12 weeks). Scans were reconstructed into 3D volumes, and relevant anatomy was segmented using a 3D U-Net model trained on expert-labeled 3DUS data. Three-dimensional equivalents of the α angle (α3D) and femoral head coverage (FHC3D) were extracted by fitting planes to the pelvis. We assessed repeatability across repeated OTUS scans and evaluated agreement with reference 3DUS. The updated system increased the proportion of high-quality scans compared to the previous prototype (42.5% vs. 9.7%). For high-quality plane fits, α3D variability was 3.15°, higher than 3DUS (1.18°) but improved relative to the prior OTUS prototype. OTUS-derived α3D showed low bias compared to 3DUS (2.98°), but this bias was variable (standard deviation: 7.54°). As in the earlier study, OTUS scans exhibited lower ultrasound volume and anatomical segmentation similarity than repeated 3DUS acquisitions, and FHC3D values showed clinically implausible dispersion. Overall, the updated OTUS system improves scan robustness and increases acquisition success rates. When high-quality intermediates are achieved, α3D repeatability appears improved relative to standard 2DUS approaches, suggesting potential clinical utility. However, persistent variability and segmentation limitations indicate that further refinement is required before clinical deployment.