Scanning Time

Operational analysis of Leica + Sectra scanner utilization, transfer, and turnaround

← Home

GitHub

https://github.com/MemorialPath/scan-time-in-real-life

Pages

https://scanningtime.patoloji.dev

https://scanningtime.patoloji.dev/

— main analysis repo (R + Quarto). - Supporting QC work: grandQC_quality_study, HistoQC, PathProfiler.

How to contribute

  1. Clone scan-time-in-real-life and reproduce the hour-of-day figure from the sample CSV.
  2. Review the architecture diagram; flag any transfer step that isn’t instrumented.
  3. Propose one new figure or one new metric with a clear operational question it answers.

Pitfalls

  • Scanner log timestamps are in local time with DST; the repo normalizes to UTC — don’t double-correct.
  • Some “failed transfers” in the logs are benign retries. The analysis de-duplicates by (scan_id, final_outcome); preserve that logic.

In the research wiki

  • Scanner Log Analysis — the method page: two documented cleaning traps that decide whether these timings mean anything.
  • Scanner Logs — the dataset itself, and what each event in it actually records.
  • Turnaround Time — the department-level metric that scanner timing feeds into.
  • WSI Quality Control — rescans driven by quality failures are a real component of scanner load, so QC and utilisation are the same question from two ends.
  • Preparation Quality Versus Image Quality — measured here on 502 slides: about 70% of the defects technicians record happen before the scanner sees the slide, so an artefact model is not a substitute for the preparation check.
  • Laboratory Workload Measurement — scanning is laboratory work, and comparable units are what let it be argued for in staffing terms.
  • Scanning Time in Real Life — this project’s own pointer page.
  • GrandQC Quality Study — the group’s only measurement of a QC model against its own technicians: 502 slides, and the source of the preparation-versus-image distinction above.