DINOv3 with Test-Time Calibration for Automated Carotid Intima-Media Thickness Measurement on CUBS v1

2026-03-10

AI Summary

This study addresses the lack of robust and transferable deep learning models for automated carotid intima-media thickness (CIMT) measurement by proposing a segmentation framework based on the DINOv3 vision foundation model, introduced here for the first time to CIMT estimation. The method predicts membrane boundary locations column-wise at a fixed resolution and converts segmentation outputs into physical CIMT values using an image calibration factor. A test-time threshold calibration mechanism is further incorporated to mitigate systematic bias. Evaluated on the CUBS v1 dataset, the model achieves a Dice score of 0.7739 and an IoU of 0.6384, with a mean absolute CIMT error of 181.16 μm, which reduces to 101.1 μm after test-time calibration—approaching the clinically acceptable precision of approximately 0.1 mm. This approach enables interpretable, calibration-aware automated CIMT assessment.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Carotid intima-media thickness (CIMT) measured from B-mode ultrasound is an established vascular biomarker for atherosclerosis and cardiovascular risk stratification. Although a wide range of computerized methods have been proposed for carotid boundary delineation and CIMT estimation, robust and transferable deep models that jointly address segmentation and measurement remain underexplored, particularly in the era of vision foundation models. Motivated by recent advances in adapting DINOv3 to medical segmentation and exploiting DINOv3 in test-time optimization pipelines, we investigate a DINOv3-based framework for carotid intima-media complex segmentation and subsequent CIMT measurement on the Carotid Ultrasound Boundary Study (CUBS) v1 dataset. Our pipeline predicts the intima-media band at a fixed image resolution, extracts upper and lower boundaries column-wise, corrects for image resizing using the per-image calibration factor provided by CUBS, and reports CIMT in physical units. Across three patient-level test splits, our method achieved a mean test Dice of 0.7739 $\pm$ 0.0037 and IoU of 0.6384 $\pm$ 0.0044. The mean CIMT absolute error was 181.16 $\pm$ 11.57 $μ$m, with a mean Pearson correlation of 0.480 $\pm$ 0.259. In a held-out validation subset ($n=28$), test-time threshold calibration reduced the mean absolute CIMT error from 141.0 $μ$m at the default threshold to 101.1 $μ$m at the measurement-optimized threshold, while simultaneously reducing systematic bias toward zero. Relative to the error ranges reported in the original CUBS benchmark for classical computerized methods, these results place a DINOv3-based approach within the clinically relevant $\sim$0.1 mm measurement regime. Together, our findings support the feasibility of using vision foundation models for interpretable, calibration-aware CIMT measurement.

Problem

• carotid intima-media thickness
• automated measurement
• ultrasound segmentation
• vision foundation models
• clinical biomarker

Innovation

• DINOv3
• test-time calibration
• carotid intima-media thickness
• vision foundation model
• ultrasound segmentation