CAIR IIT Mandi
Surface2Anatomy
Centre for AI & Robotics (CAIR) • IIT Mandi Research Initiative

3D Internal Anatomy Localization from External Body Geometry

Predicting 3D internal organ centroids and spatial uncertainty directly from optical surface scans and depth cameras using deep multi-scale cross-attention decoders. Developed at CAIR, IIT Mandi by Khushi Mhamane (Project Lead) & Sharon Melhi under the supervision of Dr. Deepak Raina.

Interactive 3D Digital Twin
CAIR Scanner
Heart & Aorta
Zero Radiation: Optical 3D surface scanning replaces ionizing CT.
IIT Mandi
Anatomical Targets121Female Reproductive, Cranial, Thorax & Spine
Internal MRE23.34 mm3-Seed Ensemble Consensus
Brain Error (Retrained)5.5 mmPhase 16 Whole-Body Model on CT-ORG
Inference Latency11.3 FPS88 ms End-to-End GPU Latency
Radiation Dose0 mSvOptical Surface Photogrammetry

Multi-Scale PointNet++ Encoder

Extracts hierarchical geometric surface features across SA1 (1,024 points), SA2 (256 points), and SA3 (64 points) scales, preserving local anatomical contours.

Target-Query Cross-Attention

104 learned organ queries attend to multi-scale body surface tokens, predicting spatial coordinate residuals relative to a canonical anatomical atlas.

3-Seed Ensemble Uncertainty

Predicts consensus centroids and computes root-mean-square seed disagreement across independent initializations, providing localized spatial confidence bounds.

Comparative Paradigm Analysis

How external geometric localization compares to standard diagnostic and procedural modalities.

ModalityIonizing DoseLatencyHardware CostPortability
Surface2Anatomy (Ours)0.0 mSv< 100 msOptical / RGB-D ScannerBedside / Field
Diagnostic Whole-Body CT10 - 20 mSvMinutes to Hours\$500,000 - \$2,000,000Fixed Gantry Suite
Diagnostic MRI0.0 mSv30 - 60 Minutes\$1,000,000+Shielded Magnet Room
Handheld Ultrasound0.0 mSvReal-Time\$5,000 - \$20,000Operator Acoustic Window
Official Python Release

Open Source Package & Checkpoints

Surface2Anatomy is packaged for production and research environments with zero placeholder weights.

Installation via PyPI
pip install surface2anatomy

Installs frozen PointNet++ multi-scale encoder, cross-attention decoders, and Ridge alignment models.

Python Quickstart
from surface2anatomy import SurfaceAnatomyModel

model = SurfaceAnatomyModel.from_pretrained()
result = model.predict("patient_surface.ply", target="spleen")
print(result.centroid_mm)  # (-61.2, 34.8, 105.4)
Weights cached automatically to standard user cache with SHA256 integrity verification.