Omentum

Morphology and diagnostic patterns in omental specimens — peritoneal surface staging, metastatic disease, and benign mimics

← Home · Onboarding

What this project is

Omental specimens — biopsies, partial resections, and “omentectomy for staging” — cross almost every oncologic sign-out queue. This project consolidates the group’s omental morphology work: what’s being submitted, how it’s sampled, and which findings actually change management.

Objectives

• Describe the case mix of omental specimens by indication (staging, symptomatic, incidental).
• Quantify gross and microscopic sampling adequacy.
• Characterize the spectrum of metastatic disease and benign mimics (reactive mesothelium, torsion / infarct, inflammatory changes).
• Identify cases in which omental findings changed stage or management.

Status

Stream	Status
Cohort identification	Active
Sampling adequacy review	Draft rubric
Metastatic case series	Planning
Benign-mimic teaching set	Drafted

Data

• Omental resections and biopsies with accompanying gross descriptions.
• De-identified WSIs on the Memorial share when available.
• Linked primary tumor diagnoses via hashed accession IDs.

Tools and repositories

• omentum — primary repo.
• Histology review in QuPath; annotation SOP on the Home page.
• Statistics via ClinicoPathDescriptives; agreement via meddecide.

Related work

• Pancreas Morphology — upstream primaries for many omental metastases.
• HER2 Gastro — gastric primaries contribute to the omental cohort.

How to contribute

1. Review the sampling-adequacy rubric and flag ambiguous rules.
2. Pilot-code 20 cases against the rubric and compare to the reference reader.
3. Annotate one benign-mimic teaching case in QuPath and submit the GeoJSON.

Pitfalls

• “Positive for metastatic carcinoma” in a staging omentectomy changes stage; confirm the linkage to the primary tumor diagnosis before any aggregation.
• Don’t conflate true metastatic deposits with transcoelomic surface implants without the gross correlate — they are prognostically different.
• Sample-size-driven conclusions are tempting in omentum studies. Keep the denominator (number of tissue blocks sampled) in every table.

In the research wiki

• Omentum — the repository’s pointer page.
• Record Linkage Under Pseudonymisation — the hashed-accession join back to the primary tumour diagnosis is exactly this problem.
• Descriptive Tables — the sampling denominator belongs in every table, and this is where that habit is written down.
• Interobserver Agreement — the pilot-coding step against a reference reader is an agreement study.
• Memorial LIS Extracts — where the case mix and indications come from.