README.md

metadata

title: LaborView AI
emoji: 👶
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 4.44.0
python_version: '3.11'
app_file: app.py
pinned: false
license: apache-2.0
models:
  - samwell/laborview-medsiglip
  - google/medsiglip-448
tags:
  - medical
  - ultrasound
  - segmentation
  - medgemma
  - labor-monitoring

LaborView AI - Intrapartum Ultrasound Analysis

AI-powered labor monitoring using MedSigLIP from Google's MedGemma family.

Features

• Automated Segmentation - Identifies pubic symphysis and fetal head
• Angle of Progression (AoP) - Key clinical indicator for labor progress
• Head-Symphysis Distance (HSD) - Distance measurement between structures
• Clinical Assessment - AI-generated labor progress evaluation

Model

Built on MedSigLIP (878M parameters), fine-tuned on transperineal ultrasound images.

• Validation IoU: 90%
• Input: 448x448 RGB ultrasound images
• Output: 3-class segmentation (background, symphysis, fetal head) + plane classification

Clinical Metrics

Angle of Progression (AoP)

• < 110: Early labor - head not engaged
• 110-120: Active labor - head descending
• 120-140: Advanced labor - good progress

• 140: Late labor - delivery imminent

Head-Symphysis Distance (HSD)

Measures the distance between inferior symphysis and closest point of fetal head.

Disclaimer

This is a research prototype developed for the MedGemma Impact Challenge. It is NOT intended for clinical use. Always consult qualified medical professionals for clinical decisions.

Citation

If you use this work, please cite:

@misc{laborview2025,
  title={LaborView AI: MedSigLIP-based Intrapartum Ultrasound Analysis},
  author={Team LaborView},
  year={2025},
  publisher={HuggingFace}
}