spacenet

SpaceNet Model Doc

Overview

SpaceNet is a classical non-deep-learning method for neuroimaging-based disease classification.

• Model family: non-deep-learning supervised classification method
• Typical objectives:
  • classify patient vs control groups from voxel-wise neuroimaging maps
  • build sparse discriminative models in aligned image space
  • export predictive scores, evaluation metrics, and interpretable weight maps
• Primary input: aligned subject images, labels, optional covariates, optional mask
• Primary output: class predictions, decision scores, cross-validation metrics, coefficient maps

In NeuroClaw, this document is model-level guidance for SpaceNet-based disease classification workflows rather than deep learning phenotype prediction.

Upstream preparation should usually be delegated to:

• fmri-skill for fMRI preprocessing and voxel-wise feature preparation
• smri-skill for structural feature extraction when disease classification uses sMRI
• nilearn-tool for concrete SpaceNet fitting and coefficient map export

Research use only.

---

Quick Start

1) Prepare disease classification inputs

Expected inputs:

• subject-level labels such as patient / control
• aligned subject-level voxel maps
• optional covariates such as age, sex, site
• optional train / validation / test split definition

If features are not ready, delegate preprocessing to fmri-skill or smri-skill first.

2) SpaceNet route

Representative operations:

• prepare subject-level voxel maps in aligned space
• fit SpaceNet for sparse discriminative disease classification
• export predictions and coefficient maps
• visualize discriminative regions for interpretation

Example execution route:

# delegated through claw-shell after voxel maps are prepared
python skills/nilearn-tool/scripts/spacenet_classifier_reference.py \
  --input-list path/to/image_list.txt \
  --labels path/to/labels.csv \
  --target diagnosis \
  --mask path/to/group_mask.nii.gz \
  --output-dir run_models_output/spacenet