Dual-energy x-ray imaging for AI-based lung tumor localization using foundation models: a phantom study

Objectives.This study assessed the performance of artificial intelligence (AI)-based tumor tracking using foundation models in planar dual-energy (DE) x-ray compared with the corresponding high-energy (HE) images from the same DE acquisition.Approach.An in-house phantom with pork ribs and artificial tumors was scanned using fast kVp-switching DE planar imaging using two kVp pairs (60/120 kVp and 80/120 kVp), with two tube current settings (15 mA and 20 mA) applied to both DE protocols. Weighted logarithmic subtraction was applied to generate soft-tissue-enhanced DE images by evaluating weighting factors in the range of 0.55-0.85. The contrast-to-noise ratio (CNR) between bone and background regions was calculated to determine the optimal weighting factor for bone suppression. An AI foundation model was then evaluated for frame-by-frame tumor localization using first-frame initialization, and its predictions were compared with manually generated ground-truth (GT) segmentation.Main result.The optimal weighting factor ranged from 0.57 to 0.79. Increasing the tube current improved DE CNR, with median improvements of 4.7% for the 60/120 pair kVp and 11.5% for the 80/120 kVp pair. The AI model achieved high localization accuracy across DE and HE images (Dice scores0.9, HD952-3 mm). Although global metrics showed limited differences between the two image types and across imaging protocols, analysis of the missed tumor fraction, defined as the proportion of the GT tumor area not captured by the predicted segmentation, demonstrated improved localization with DE imaging, particularly for the smallest spherical tumor. For the 60/120 and 80/120 kVp pairs, the median missed tumor fraction was 35%-48% and 50%-55% for HE images, compared with 5%-10% and15% for DE images, respectively.Significance.This study demonstrates that integrating DE imaging with AI foundation models may improve tumor localization in anatomically challenging regions, particularly when tumor-bone overlap occurs.