Surgical Environment Error Prevention

Research question and target

Evaluate tracking within the visible field. Use fixed sequences with occlusion to test arrival and removal counts.

Internal target

Zero instrument count omissions on the fixed test

Programme

Patient safety

Evidence on this record

Common synthetic checks

Evaluation data

• CAMMA Cholec80 videos and tool annotations

  Cholec80 release · CAMMA overlap index 8347b9f4cb02ebe739747903e6eada272ee9d25e

  Manifest locked
• CholecTrack20 CVPR 2025

  2025-02-28 Synapse release

  Split rule locked
• Independent instrument and sponge counts linked to video

  Eligible reference set pending

  Blocked

Evaluation registry · Labels and split definitions (KR)

Sources and eligibility

• CAMMA Cholec80

  Public · Noncommercial terms

  CC BY-NC-SA 4.0 applies. Tool presence in visible frames covers a different endpoint from instrument movement throughout the room or retained objects.

• EndoVis / ROBUST-MIS

  Dataset-specific conditions

  Supports tool presence and segmentation in surgical video. Coverage is limited to the observable field.

• HeiChole

  Public · Noncommercial terms

  The repository specifies CC BY-NC-SA 4.0. It provides phase, instrument and action labels, with views outside the abdominal cavity blurred.

Boundary cases

• Tracking and duplicate counts after smoke, occlusion or departure from the camera field
• A retained-item mismatch flagged before closure in synthetic sequences
• Single-frame latency of at most 33 ms on fixed hardware

Common synthetic checks

• Occlusion, arrival and removal tracking · 5/5 matched

Expected and observed values for these cases

What the official call asks

Automatic counting of surgical instruments and patient identification before and after surgery are separate areas. Validation takes place in real operating rooms at three or more hospitals, with designated types of institution taking part.

Proposal requirements, printed pages 101–104 · All twelve requirements

Interpretation and next evaluation

In-frame presence and segmentation cover a different observation space from total instrument inventories. Report the synthetic missed-count denominator.

Next check

Lock visibility, occlusion and tracking ID criteria and define separate video and synthetic count metrics.