AI Fetal Ultrasound Biometry Analysis System
Budget: ₹12,500 – ₹37,500 INR
Building an AI-powered fetal ultrasound biometry analysis system. Looking for an experienced ML/computer vision engineer or team.
The project involves developing a two-phase deep learning pipeline for automated measurement of fetal biometry from 2D ultrasound images.
Ultrasound plane detection and classification (head, abdomen, femur planes)
Semantic segmentation of anatomical structures using CNN/U-Net architecture
Ellipse fitting and geometry extraction for biometry calculations
Automated measurement of HC, BPD, AC, FL, OFD, EFW, and derived ratios
Scan quality scoring, measurement consistency validation
Basic explainability (GradCAM overlays) and rule-based report generation
Training datasets: HC18, FETAL_PLANES_DB, FPUS23, INTERGROWTH growth charts
Multi-task learning model (single backbone, multiple biometry outputs)
Confidence-weighted measurements with calibrated uncertainty estimates
Longitudinal growth modeling across multiple scans
Anomaly and risk scoring against population growth curves
Active learning loop for continuous improvement from clinician corrections
Bias and fairness monitoring across demographic subgroups
Regulatory audit trace engine (MDR / FDA 510(k) ready)
The project involves developing a two-phase deep learning pipeline for automated measurement of fetal biometry from 2D ultrasound images.
Ultrasound plane detection and classification (head, abdomen, femur planes)
Semantic segmentation of anatomical structures using CNN/U-Net architecture
Ellipse fitting and geometry extraction for biometry calculations
Automated measurement of HC, BPD, AC, FL, OFD, EFW, and derived ratios
Scan quality scoring, measurement consistency validation
Basic explainability (GradCAM overlays) and rule-based report generation
Training datasets: HC18, FETAL_PLANES_DB, FPUS23, INTERGROWTH growth charts
Multi-task learning model (single backbone, multiple biometry outputs)
Confidence-weighted measurements with calibrated uncertainty estimates
Longitudinal growth modeling across multiple scans
Anomaly and risk scoring against population growth curves
Active learning loop for continuous improvement from clinician corrections
Bias and fairness monitoring across demographic subgroups
Regulatory audit trace engine (MDR / FDA 510(k) ready)
Related categories:
Machine Learning (ML)
Computer Vision
Deep Learning
Anomaly Detection
Convolutional Neural Network