Virtual-Eyes: Quantitative Validation of a Lung CT Quality-Control Pipeline for Foundation-Model Cancer Risk Prediction

Abstract

Robust preprocessing is rarely quantified in deep-learning pipelines for low-dose CT (LDCT) lung cancer screening. We develop and validate Virtual-Eyes, a clinically motivated, 16-bit CT quality-control pipeline for NLST, and measure its differential impact on generalist foundation models versus specialist models. Virtual-Eyes enforces strict 512 $\times$ 512 resolution, rejects short or non-diagnostic series, and extracts a contiguous lung block using Hounsfield-unit filtering and bilateral lung-coverage scoring while preserving the original 16-bit DICOM grid. Using 765 NLST patients (182 cancer, 583 non-cancer), we evaluate RAD-DINO, Merlin, Sybil, and ResNet-18 under a leakage-free protocol. For RAD-DINO, preprocessing improves slice-level AUC from 0.576 to 0.610 and patient-level AUC from 0.646 to 0.683 (mean pooling) and 0.619 to 0.735 (max pooling), with improved calibration (Brier score 0.188 $\rightarrow$ 0.112). In contrast, Sybil and ResNet-18 degrade under Virtual-Eyes, revealing reliance on contextual or shortcut features, while Merlin shows limited transferability. Sensitivity analysis and uncertainty estimation confirm the robustness and stability of these findings.