What question did this study set out to answer?

The aim is to evaluate a new method for associating data in cell tracking, using surprise-gated techniques.

July 8, 20260 citationsOpen Access

CIP-CT-01 — Surprise-Gated Data Association for Cell Tracking: a pre-registered negative result

Key Points

The aim is to evaluate a new method for associating data in cell tracking, using surprise-gated techniques.
Employs a boundary-surprise gate with geometric confirmation to evaluate cell division tracking.
Uses synthetic data to benchmark performance metrics under controlled conditions.
Evaluates the method's pass-rule based on locked aggregate metrics and division handling.
Achieved a false-split rate reduction of approximately 90%.
Increased selectivity from 0.29 to 2.15.
The system fails to meet its pre-registered pass-rule, with 84% of performance affected by ID-switches.

Abstract

A pre-registered junction-invention benchmark under BIT-OS/CIP v1. 2. A boundary-surprise gate (|ΔArea|>θ) augmented with geometric confirmation (two ~½-area daughters + midpoint symmetry) cuts false-split rate ~90% and lifts selectivity Rₛel 0. 29→2. 15. The case nonetheless fails its pre-registered system pass-rule: the locked aggregate metric is dominated (84%) by steady-state ID-switch, capping any division handler at ~15% and actively de-prioritizing division recall under tuning. Published as a first-class negative result with full harness and per-arm-per-seed receipts. Synthetic data only; no novelty claim (prior-art search pending).

问 AI

Bookmark

View Full Paper