VIRDX

VIRDX builds MRI-based diagnostics for prostate cancer, spanning the full pipeline from data platform to synthetic data generation to deep-learning inference:

• ve2e (the “inverse” workstream) — infers disease state (csPCa, ISUP grade, lesion location) from bi-parametric MRI using deep inverse models (DINO/ViT-based, e.g. vx1, FlexDINO). Competes on public benchmarks (PI-CAI Open Tuning Leaderboard). See workstreams/inverse/index.md.
• diffsim — a biophysical tissue-simulation engine that generates synthetic MR imaging data from tissue microstructure models, used to train and stress-test the ve2e inverse models and to explore MR sequence design. See workstreams/diffsim/index.md.
• histo — histopathology imaging/annotation work (whole-slide images, tissue segmentation, vxAnnotate-produced pathology assessments) that grounds the biophysical tissue models used by diffsim and provides ground truth for ve2e. See workstreams/histo/index.md.
• infrastructure — the data platform (vxData), Kubernetes cluster, CI/CD, and internal tooling (dashboard, vxAnnotate) that the science workstreams run on. See workstreams/infrastructure/index.md.

Other workstreams tracked by the org (product, clinical, regulatory, business) have no dedicated wiki pages yet — no monthly reports or inbox sessions have been filed for them so far either; there is nothing to synthesize until material shows up in inbox/.

Where to look for more

• Each workstream’s sops/ directory — the actionable how-to material. Start at workstreams/infrastructure/sops/ and workstreams/histo/sops/.
• Monthly reports overview — cross-workstream progress digest by month. Monthly reports are the only thing left under sources/; the legacy guides/, investigations/ and incidents/ trees were audited against the repos on 2026-08-06 and either promoted into wiki/ or deleted.