docker

Docker Environments

This directory provides dedicated container setups for:

• Core Python workflows (no SIRF/STIR/PyTomography)
• STIR workflows
• SIRF workflows
• PyTomography workflows

Quick start

Build and run a specific environment:

docker compose -f docker/compose.yaml build stir
docker compose -f docker/compose.yaml run --rm stir

Run the dedicated connector + OSEM examples directly:

docker compose -f docker/compose.yaml run --rm stir python examples/07A_stir_adaptor_osem.py
docker compose -f docker/compose.yaml run --rm sirf python examples/07B_sirf_adaptor_osem.py
docker compose -f docker/compose.yaml run --rm pytomography python examples/07C_pytomography_adaptor_osem.py

Run the containerized test matrix:

bash scripts/run_container_validation.sh

Include SIMIND-dependent examples/integration tests:

bash scripts/run_container_validation.sh --with-simind

Run selected example groups:

bash scripts/run_container_examples.sh --only-core
bash scripts/run_container_examples.sh --only-pytomography
bash scripts/run_container_examples.sh --only-osem

Set a specific Docker platform when needed:

bash scripts/run_container_examples.sh --only-core --docker-platform linux/amd64
bash scripts/run_container_validation.sh --with-simind --docker-platform linux/amd64

Require SIMIND explicitly (fail fast if missing):

bash scripts/run_container_examples.sh --only-osem --require-simind
bash scripts/run_container_validation.sh --with-simind --require-simind

Services

• python: Core Python image for non-backend examples/tests.
• stir: STIR-focused image (based on synerbi/sirf) with this repository installed.
• sirf: SIRF-focused image (based on synerbi/sirf) with this repository installed.
• pytomography: Python image with PyTomography + this repository installed.
• Each service enables an import guard so only its target library is importable during tests/examples (stir, sirf.STIR, or pytomography respectively; python blocks all three backend libraries).

Notes

• The SIRF container build arg is SIRF_BASE_IMAGE.
• The STIR container build arg is STIR_BASE_IMAGE.
• Install SIMIND by following the official installation instructions on the SIMIND website.
• For quick local use with these containers, put SIMIND at ./simind in this repo so the executable is available as ./simind/simind.
• SIMIND-dependent examples require the simind executable inside the container. Runner scripts now auto-check ./simind/simind and skip SIMIND-dependent checks when missing (or fail fast if --require-simind is used).
• For mixed host/runtime architectures, the runner scripts auto-detect SIMIND binary architecture (x86_64/aarch64) and set Docker platform automatically; use --docker-platform to override.
• input.smc is bundled in simind_python_connector/configs; SIMIND runtime checks only gate external executable availability.
• The toy adaptor examples use simind_python_connector/configs/Example.yaml, a reduced projection-space profile derived from AnyScan.yaml.
• These images are intended to run examples and interactive exploration.