How to CALMaR

CALMaR: Co-designed, Automated Lesion Mapping and Reporting

Understanding neuroimaging for CALMaR

CALMaR connects neuroimaging tools into a reproducible workflow for stroke lesion mapping and clinically interpretable reporting. This guide explains the concepts needed to work on that workflow without turning every contributor into a radiologist or neuroimaging methodologist.

The focus is practical: what each processing stage contributes, what its outputs mean, which assumptions can break, and how errors affect later analyses.

Scope

This guide supports development and research. CALMaR outputs require appropriate quality control and clinical interpretation. Associations derived from atlases, normative data, or published groups do not establish an individual patient's diagnosis, prognosis, or ideal treatment.

CALMaR at a glance

flowchart TD
    A[Clinical or research MRI] --> B[Prepare images and verify spatial information]
    B --> C[Automatic lesion segmentation]
    C --> D[Automated QC]
    D --> E{Human review available?}
    E -->|Yes| F[Review or correct mask]
    E -->|No| G[Retain automated result with QC status and uncertainty]
    F --> H[Register and resample]
    G --> H
    H --> I[Lesion, atlas, tract and network analyses]
    I --> J[Evidence-linked interpretation]
    J --> K[Traceable report]

Figure 1. CALMaR at a glance. The automatic path continues with explicit quality-control status and uncertainty when human review is unavailable. Human review or correction is an optional branch.

Human review can improve confidence and provide a corrected or reference mask, but the workflow must remain capable of producing an explicitly qualified result when no human-traced mask is available.

Start where the work takes you

• I am new to neuroimaging

  Start with why neuroimaging matters, then learn about the images used by CALMaR.

• I am working on preprocessing

  Read images as spatial data and the workflow overview.

• I am working on lesion masks

  Go to automatic lesion segmentation and quality control.

• I am implementing an analysis

  Compare atlas overlap, disconnection, and meta-analytic decoding.

• I am interpreting an output

  Read clinical interpretation before translating a result into language about impairment, recovery, or therapy.

• I keep hearing the same question

  Check the recurring questions, then add a new one through the repository issue template if it is missing.

Authoritative project sources

• CALMaR repository for the current notebooks, code, QC rubric, knowledge base, and changing implementation details
• Lesion interpretation notebook for the implemented workflow
• Lesion segmentation benchmark for the current tool comparison and benchmarking outputs
• NeurodeskEDU for general tutorials and executable neuroimaging examples

This guide explains stable concepts. When a tool, version, threshold, or benchmark result changes, CALMaR itself remains the source of truth.