Image algorithm validator

Skill a5c-ai/babysitter/library/specializations/domains/science/biomedical-engineering/skills/image-algorithm-validator

Babysitter enforces obedience on agentic workforces and enables them to manage extremely complex tasks and workflows through deterministic, hallucination-free self-orchestration

Install

npx -y skills add a5c-ai/babysitter --skill image-algorithm-validator

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Medical image processing algorithm validation skill for segmentation, detection, and analysis algorithms

SKILL.md

2.9 KB, as published. Nobody here has run it

Image Algorithm Validator Skill

Purpose

The Image Algorithm Validator Skill supports validation of medical image processing algorithms, including segmentation, detection, and analysis algorithms, ensuring performance meets clinical requirements.

Capabilities

Ground truth dataset curation guidance
Performance metric calculation (Dice, IoU, sensitivity, specificity)
Inter-observer variability analysis
Statistical comparison methods
Validation dataset stratification
Multi-reader multi-case study design
FDA AI/ML guidance alignment
Failure case analysis
Edge case identification
Performance boundary testing
Cross-validation methodology

Usage Guidelines

When to Use

Validating image analysis algorithms
Curating validation datasets
Designing reader studies
Preparing regulatory submissions

Prerequisites

Algorithm development complete
Ground truth established
Validation dataset available
Performance criteria defined

Best Practices

Use representative, diverse datasets
Establish robust ground truth methodology
Assess performance across subgroups
Document failure modes

Process Integration

This skill integrates with the following processes:

Medical Image Processing Algorithm Development
AI/ML Medical Device Development
Clinical Evaluation Report Development
Software Verification and Validation

Dependencies

SimpleITK library
scikit-image
MONAI framework
Evaluation frameworks
Statistical analysis tools

Configuration

image-algorithm-validator:
  algorithm-types:
    - segmentation
    - detection
    - classification
    - registration
    - quantification
  metrics:
    - Dice
    - IoU
    - sensitivity
    - specificity
    - AUC
    - Hausdorff-distance
  validation-methods:
    - holdout
    - cross-validation
    - external-validation

Output Artifacts

Dataset curation protocols
Ground truth documentation
Performance reports
Statistical analyses
Reader study results
Failure mode catalogs
Regulatory submission sections
Validation summaries

Quality Criteria

Ground truth methodology validated
Metrics appropriate for algorithm type
Dataset representative of intended use
Statistical analysis rigorous
Subgroup performance assessed
Documentation supports regulatory review