Soft Temporal Scoring Using a Foundation Model: Optimal Frame Selection for Improved ONSD Measurement in Ultrasound Videos
- 1University of Central Florida
Abstract
Medical ultrasound is a portable, non-invasive, and cost effective imaging modality that is particularly well suited for resource-limited settings. Optic nerve sheath diameter (ONSD) measurement from ultrasound is used as a point-of-care assessment tool for intracranial pressure (ICP), which is associated with several neurological conditions. However, valid measurement depends on selecting a frame in which the optic nerve sheath is clearly visible. Manual selection requires a high level of expertise. To enable the use of ONSD in low resource settings by medical personnel of all levels, we present a sparsely supervised AI frame-work that scores each frame of an ONSD ultrasound video and selects the optimal frame for ONSD assessment. Per-frame embeddings from an ultrasound foundation model (USFM) are passed to a lightweight temporal head and trained using Gaussian soft labels, which assign graded targets around labeled key frames, rather than hard binary (0/1) per-frame targets. We evaluate the model performance using subject-level cross-validation on 18 subjects and 323 ultrasound videos spanning nine acquisition sweep types. Top-k frame selection is used as a metric for direct comparison against the baselines. While only requiring sparse labels, our method can identify an optimal frame in 82.2% of the evaluated sweeps, exceeding the strongest training-free baseline (49.2%) and hard-label classification (71.9%).
Related
- Standardized Transducer Placement Representation for Improved Reviewability of Ultrasound Examinations
- Core needle biopsy washout biomarkers combined with ultrasound for preoperative axillary risk stratification in breast cancer: a retrospective nomogram study with temporal validation
- Data-Efficient Prostate Micro-Ultrasound Image Analysis using a Modality-specific Foundation Model
- Multi-feature Response Attention Network for kidney stone detection and measurement of stone size in ultrasound images
- Relative and absolute reliability of medial knee joint space width measurement using ultrasound imaging during valgus stress testing
BibTeX
@misc{ali2026soft,
title = {{Soft Temporal Scoring Using a Foundation Model: Optimal Frame Selection for Improved ONSD Measurement in Ultrasound Videos}},
author = {M. Ali and L. M. Castaneda and C. Wu and C. E. Escamilla-Ocanas and M. Hirzallah and L. J. Brattain},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.21.26362872},
url = {https://www.medrxiv.org/content/10.64898/2026.09.21.26362872v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Soft Temporal Scoring Using a Foundation Model: Optimal Frame Selection for Improved ONSD Measurement in Ultrasound Videos
AU - M. Ali
AU - L. M. Castaneda
AU - C. Wu
AU - C. E. Escamilla-Ocanas
AU - M. Hirzallah
AU - L. J. Brattain
PY - 2026
DA - 2026/09/21
PB - medRxiv
DO - 10.64898/2026.09.21.26362872
UR - https://www.medrxiv.org/content/10.64898/2026.09.21.26362872v1
AB - Medical ultrasound is a portable, non-invasive, and cost effective imaging modality that is particularly well suited for resource-limited settings. Optic nerve sheath diameter (ONSD) measurement from ultrasound is used as a point-of-care assessment tool for intracranial pressure (ICP), which is associated with several neurological conditions. However, valid measurement depends on selecting a frame in which the optic nerve sheath is clearly visible. Manual selection requires a high level of expertise. To enable the use of ONSD in low resource settings by medical personnel of all levels, we present a sparsely supervised AI frame-work that scores each frame of an ONSD ultrasound video and selects the optimal frame for ONSD assessment. Per-frame embeddings from an ultrasound foundation model (USFM) are passed to a lightweight temporal head and trained using Gaussian soft labels, which assign graded targets around labeled key frames, rather than hard binary (0/1) per-frame targets. We evaluate the model performance using subject-level cross-validation on 18 subjects and 323 ultrasound videos spanning nine acquisition sweep types. Top-k frame selection is used as a metric for direct comparison against the baselines. While only requiring sparse labels, our method can identify an optimal frame in 82.2% of the evaluated sweeps, exceeding the strongest training-free baseline (49.2%) and hard-label classification (71.9%).
ER -