Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

VLM-in-the-Loop: A Plug-In Quality Assurance Module for ECG Digitization Pipelines

Jiachen Li, Shihao Li, Soovadeep Bakshi, Wei Li, Dongmei Chen

Abstract

ECG digitization could unlock billions of archived clinical records, yet existing methods collapse on real-world images despite strong benchmark numbers. We introduce \textbf{VLM-in-the-Loop}, a plug-in quality assurance module that wraps any digitization backend with closed-loop VLM feedback via a standardized interface, requiring no modification to the underlying digitizer. The core mechanism is \textbf{tool grounding}: anchoring VLM assessment in quantitative evidence from domain-specific signal analysis tools. In a controlled ablation on 200 records with paired ground truth, tool grounding raises verdict consistency from 71\% to 89\% and doubles fidelity separation ($Δ$PCC 0.03 $\rightarrow$ 0.08), with the effect replicating across three VLMs (Claude Opus~4, GPT-4o, Gemini~2.5 Pro), confirming a pattern-level rather than model-specific gain. Deployed across four backends, the module improves every one: 29.4\% of borderline leads improved on our pipeline; 41.2\% of failed limb leads recovered on ECG-Digitiser; valid leads per image doubled on Open-ECG-Digitizer (2.5 $\rightarrow$ 5.8). On 428 real clinical HCM images, the integrated system reaches 98.0\% Excellent quality. Both the plug-in architecture and tool-grounding mechanism are domain-parametric, suggesting broader applicability wherever quality criteria are objectively measurable.

VLM-in-the-Loop: A Plug-In Quality Assurance Module for ECG Digitization Pipelines

Abstract

ECG digitization could unlock billions of archived clinical records, yet existing methods collapse on real-world images despite strong benchmark numbers. We introduce \textbf{VLM-in-the-Loop}, a plug-in quality assurance module that wraps any digitization backend with closed-loop VLM feedback via a standardized interface, requiring no modification to the underlying digitizer. The core mechanism is \textbf{tool grounding}: anchoring VLM assessment in quantitative evidence from domain-specific signal analysis tools. In a controlled ablation on 200 records with paired ground truth, tool grounding raises verdict consistency from 71\% to 89\% and doubles fidelity separation (PCC 0.03 0.08), with the effect replicating across three VLMs (Claude Opus~4, GPT-4o, Gemini~2.5 Pro), confirming a pattern-level rather than model-specific gain. Deployed across four backends, the module improves every one: 29.4\% of borderline leads improved on our pipeline; 41.2\% of failed limb leads recovered on ECG-Digitiser; valid leads per image doubled on Open-ECG-Digitizer (2.5 5.8). On 428 real clinical HCM images, the integrated system reaches 98.0\% Excellent quality. Both the plug-in architecture and tool-grounding mechanism are domain-parametric, suggesting broader applicability wherever quality criteria are objectively measurable.

Paper Structure

This paper contains 29 sections, 7 figures, 17 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. System Overview
  4. Digitization Pipeline (Testbed)
  5. Tool Grounding Modules
  6. Experiments
  7. Judge Ablation: Tool Grounding Is the Key Mechanism
  8. Multi-VLM Robustness: Tool Grounding Is Model-Invariant
  9. VLM Verdicts Predict Reconstruction Fidelity
  10. End-to-End VLM Value Ablation
  11. Pipeline and QA Evaluation on Real Clinical Data
  12. Baseline Comparisons and Backend Separability
  13. Discussion
  14. Conclusion
  15. Pipeline Implementation Details
  16. ...and 14 more sections

Figures (7)

  • Figure 1: Representative challenging ECGs from the 428-image HCM clinical archive. All three state-of-the-art baselines fail on these images.
  • Figure 2: System overview.Left: Digitization pipeline with Optimizer Agent closing the loop via 12-action correction space; backend-agnostic interface supports drop-in replacement. Right: Hierarchical QA routing (Tiers 1--3) producing per-lead verdicts. Tool grounding lifts consistency to 89%; 98.0% Excellent on 428 HCM images.
  • Figure 3: Judge ablation: tool grounding produces the largest gains in consistency and fidelity separation.
  • Figure 4: VLM quality score distribution (428 HCM images).
  • Figure 5: Qualitative examples from PTB-XL (high / median / low fidelity).
  • ...and 2 more figures

Theorems & Definitions (1)

  • Definition 1: Plug-in Interface