Dicom Series Preflight

skill_manifest.yaml

# SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

id: medagent.dicom_series_preflight
version: 0.1.0
upstream_refs:
  - kind: pypi_package
    name: pydicom
    version_constraint: ">=2.4.5,<4,!=3.0.0,!=3.0.1"
  - kind: pypi_package
    name: nibabel
    version_constraint: ">=5.4,<6"
license: Apache-2.0
intended_use:
  summary: >
    Engineering-time DICOM folder preflight. Header-only scan of a series
    directory for corruption, orientation, PHI-tag presence, and consistency
    before conversion or model inference. No GPU required.
  scope: development
  not_for:
    - clinical deployment
    - regulatory de-identification
    - autonomous diagnosis
    - production ingestion without a vetted converter
inputs:
  - name: dicom_dir
    type: directory_path
    formats: [dicom_series]
outputs:
  - name: preflight_json
    type: json
    schema: validators/output_schema.json
runtime:
  language: python
  python: ">=3.10"
  entrypoint: scripts/preflight_series.py
  args:
    - "${python}"
    - "${script}"
    - "${fixture}"
  dependencies:
    pydicom: ">=2.4.5,<4,!=3.0.0,!=3.0.1"
    nibabel: ">=5.4,<6"
    numpy: ">=1.25,<3"
    typer: ">=0.9"
  side_effects:
    pip_packages:
      - pydicom>=2.4.5,<4,!=3.0.0,!=3.0.1
      - nibabel>=5.4,<6
      - numpy>=1.25,<3
      - typer>=0.9
    local_writes: []
    home_writes: []
    network_endpoints: []
    requires_docker: false
    requires_gpu: none
    env_required: []
paired_verifiers:
  - id: medagent.verifiers.dicom_preflight_quality_v1
    status: implemented
    consumes: evidence_pack_dir
    purpose: >
      Second-pass preflight gate with pass/warn/fail verdict and explicit
      findings list. Surfaces corruption, orientation, PHI, and consistency
      issues without running conversion or segmentation.
limitations:
  - Header-only; does not decode pixel data or detect burnt-in PHI.
  - Canonical orientation gate assumes LPS-derived CT axcodes L,P,S.
  - Compressed transfer syntax and multi-frame instances are warned, not decoded.
  - Single-directory scan; does not reconcile multiple studies in one tree.
validation:
  expected_runtime_seconds:
    max: 15.0
    inference_path: runtime.scan_seconds
  sanity_checks:
    - {path: skill, eq: dicom_series_preflight}
    - {path: series.single_series, eq: true}
    - {path: inventory.n_corrupt, eq: 0}
    - {path: orientation.axcodes_match, eq: true}
    - {path: preflight.acceptable, eq: true}
  expected_cost:
    wall_seconds: {max: 15}
    cpu_seconds: {max: 15}
    rss_mb_peak: {max: 300}
    gpu_seconds: {max: 0}
    gpu_memory_mb_peak: {max: 0}
  env_pin:
    pydicom: ">=2.4.5,<4,!=3.0.0,!=3.0.1"
    nibabel: ">=5.4,<6"
    numpy: ">=1.25,<3"
  reproducibility:
    mode: repeat
    fixture: fixtures/clean_no_phi
    fixture_builder: fixtures/generate_fixtures.py
    runs: 2
negative_fixtures:
  - path: fixtures/flipped_lr
    expected_overall: failed
    expected_failed_gate: sanity
    failure_reason: >
      LR-flipped ImageOrientationPatient should fail orientation.axcodes_match
      and preflight.verdict sanity gates.