Abstract
PURPOSE: Two-dimensional (2D) X-ray imaging is routinely used in radiotherapy for patient alignment but offers limited tumor visibility due to anatomical overlap. We propose a novel technique, prior information-based subtraction (PIBS) radiography, to enhance tumor visualization on 2D X-ray and facilitate more accurate patient alignment. MATERIALS AND METHODS: PIBS X-ray improves tumor visualization on 2D X-ray by subtracting signal contributions from non-essential (NE) tissues, which can be obtained from the planning CT (prior information). To demonstrate the feasibility, we acquired four CT scans using a thorax phantom with a spherical lung target, representing the planning CT and three treatment-day CTs capturing the patient's positions with the lung tumor at three different locations, respectively. NE tissues, defined as structures outside the thoracic cavity, were segmented from the planning CT. On treatment days, we aligned NE tissues based on bony structures via 3D/3D or 2D/2D registration and applied the calculated shifts so the patient's position matched that in the planning CT. To simulate 2D X-rays, digitally reconstructed radiographs (DRRs) were generated from the planning CT of NE tissues and aligned with treatment-day CTs. Subtracting NE tissue contributions from conventional 2D X-ray yielded PIBS X-ray. PIBS X-ray images were compared with conventional X-ray images to assess tumor visibility. RESULTS: PIBS X-ray provided better tumor visualization across all projection angles and all tumor locations compared to conventional X-ray, even for certain cases where the tumor was completely indistinguishable on conventional X-ray. Both 3D/3D and 2D/2D registrations produced acceptable alignment of NE tissues because of the clear visibility of bony structures. CONCLUSION: PIBS X-ray is a promising technique to enhance tumor visualization on 2D X-ray. In this study, we demonstrated its potential in improving radiotherapy on-board imaging using a thorax phantom. In the future, we will explore strategies for integrating this technology into clinical workflows and evaluate its clinical benefits in radiotherapy patients.