rphammonds/scalesurfer-stats-v7

ScaleSurfer FreeSurfer v7 Stats Prediction Model

This repository contains a ScaleSurfer multi-head model for predicting FreeSurfer-style .stats targets from a T1w image and an aparc+aseg segmentation.

Files

• stats_model.safetensors: model weights and inference metadata in safetensors format.
• config.json: architecture and feature-schema metadata needed by ScaleSurferStatsPredictor.
• metadata.json: checksums and training/evaluation metadata.
• summary.csv, history.csv, target_metrics.csv: copied training diagnostics.

Test Summary

Group	Targets	Values	Normalized MAE	Median absolute percent error
aseg	113	12321	0.45587876439194475	7.503098487854004
global	64	5638	0.1726450031340259	1.3967947363853455
lh_aparc	306	33914	0.4840144798817026	8.386276721954346
rh_aparc	306	33912	0.49182673176802866	8.2892165184021

Loading

from scalesurfer.stats import ScaleSurferStatsPredictor

predictor = ScaleSurferStatsPredictor.from_pretrained(7)
features = predictor.predict_subjects(subjects_dir, subjects, return_format="wide")

This model is intended for research workflows and is not a clinical diagnostic device.