rikixy/virtual-staining

How rikixy/virtual-staining works

Virtual Staining is a reproducible histopathology image-translation framework. Users configure a run, prepare slide sets into patches, train Pix2Pix or CycleGAN, generate images, then evaluate or compare results. It also supports single-image and directory inference, TIFF conversion, comparison panels, run organization, and queued staged runs. The graph reflects the sampled pipeline, inference, and metadata code plus README-documented capabilities; unsampled subsystem wiring is left unspecified.

Main entry points: pipeline.py Stage Orchestrator · runner.py Inference Runner · evaluate.py Run Evaluation · methods Translation Methods

Components 24

Run Workflow

Stage Orchestrator virtual_staining/applications/pipeline.py
Run Configuration virtual_staining/config/run.py
Dataset Preparation virtual_staining/applications/prepare.py
Slide Data Processing virtual_staining/data
Run Session virtual_staining/experiment/session.py

Translation Runtime

Model Training virtual_staining/applications/train.py
Training Runtime virtual_staining/training/trainer.py
Translation Methods virtual_staining/methods
Network Components virtual_staining/models
Checkpoint Selection virtual_staining/checkpoint_selection.py
Run Inference virtual_staining/applications/infer.py
Inference Runner virtual_staining/inference/runner.py
Image Inference virtual_staining/inference/single.py
Generated Artifacts virtual_staining/inference/outputs.py

Evaluation

Run Evaluation virtual_staining/applications/evaluate.py
Image Metrics virtual_staining/metrics.py
Evaluation Reports virtual_staining/evaluation/reports.py
Run Comparison virtual_staining/applications/compare.py

Run Utilities

TIFF Conversion virtual_staining/applications/convert.py
Comparison Panels virtual_staining/applications/compare_panels.py
Output Organization virtual_staining/applications/organize.py
Queued Runs virtual_staining/applications/run_queue.py
Dependency Status virtual_staining/applications/status.py

Other components

User

Connections 25

User → Stage Orchestrator: runs stages
Stage Orchestrator → Run Configuration: loads config
Stage Orchestrator → Dataset Preparation: dispatches prepare
Stage Orchestrator → Model Training: dispatches train
Stage Orchestrator → Run Inference: dispatches infer
Stage Orchestrator → Run Evaluation: dispatches evaluate
Dataset Preparation → Slide Data Processing: builds patches
Model Training → Training Runtime: starts training
Training Runtime → Translation Methods: uses method
Translation Methods → Network Components: constructs networks
Run Inference → Inference Runner: runs inference
Inference Runner → Checkpoint Selection: selects checkpoint
Inference Runner → Translation Methods: builds predictor
Inference Runner → Generated Artifacts: writes outputs
User → Image Inference: infers images
Image Inference → Generated Artifacts: writes outputs
Run Evaluation → Image Metrics: computes metrics
Run Evaluation → Evaluation Reports: writes reports
User → Run Comparison: compares runs
User → TIFF Conversion: converts TIFFs
User → Comparison Panels: builds panels
User → Output Organization: organizes outputs
User → Queued Runs: queues runs
Queued Runs → Stage Orchestrator: dispatches stages
User → Dependency Status: checks dependencies

More Python diagrams: tryingtobeingnikhil/vllm_inference_engine · tue-bmd/zea · uuembodiedsocialai/probtalk3d · wannabeasmartguy/gpt-gradio-agent · zhangfengcdt/memoir

Drawn by GitDiagram on September 28, 2026 from the repository's file tree, README and source files. Read as Markdown ·