Click here to

Session: Real-time Tracking and Adaptive Radiation Therapy [Return to Session]

Dual-Energy Real-Time Stereoscopic X-Ray Image Guidance for Markerless Lung Tumor Tracking

C Peacock, M Sattarivand*, Nova Scotia Health Authority


SU-F-TRACK 6-2 (Sunday, 7/25/2021) 4:30 PM - 5:30 PM [Eastern Time (GMT-4)]

Purpose: The purpose of this work is to develop real-time markerless tracking of lung tumors during patient free-breathing, guided by stereoscopic dual energy (DE) x-ray imaging.

Methods: Incremental and motorized motion stages were designed and used to simulate lung tumor respiratory motion within an anthropomorphic phantom. A set of tumor motion experiments were performed using spherical tumor models with diameters ranging from 6.4 to 25.4 mm. A series of stereoscopic high (140 kVp) and low (60 kVp) energy x-ray images of the phantom were acquired with the moving tumor insert using the ExacTrac stereoscopic imaging system (Brainlab AG, Germany). The high and low energy images were combined by weighted log subtraction to create stereoscopic DE images. Conventional stereoscopic single energy (SE) (120 kVp) images were also acquired for comparison. Digitally reconstructed radiographs (DRRs) were calculated from tumor-only volumes within CT images of the phantom, and tumor templates were cropped from these DRRs. Normalized template-image cross correlation and stereoscopic triangulation were used to calculate the 3D tumor positions from stereoscopic image pairs, which were compared to a ground truth from known tumor locations and the accuracy, precision and success rate of localization were calculated.

Results: The success rate of tumor localization using DE imaging was consistently higher than that of the SE technique for both incremental and dynamic motion experiments with most dramatic improvement for the smaller tumors (<20mm). The accuracy of tumor localization using the DE technique was generally sub-millimeter (range 0.53 - 1.10 mm).

Conclusion: DE stereoscopic imaging outperforms the SE technique to localize lung tumors in real-time particularly for smaller tumors. With some alterations to existing stereoscopic image guidance systems, markerless lung tumor tracking would be feasible.

Funding Support, Disclosures, and Conflict of Interest: This work was supported by the ACOA (Atlantic Canada Opportunities Agency) Atlantic Innovation Fund.



    Not Applicable / None Entered.


    Not Applicable / None Entered.

    Contact Email