Feasibility of a real‐time dual energy markerless monitoring of lung tumors using a clinical room‐mounted stereoscopic and monoscopic x‐ray imaging system
Bibliographic record
Abstract
BACKGROUND: The motion of lung tumors during breathing poses challenges in stereotactic body radiotherapy (SBRT), warranting improved monitoring techniques. Breathing complicates SBRT by creating positional uncertainty in the lungs, traditionally managed with PTV margins, respiratory gating, or breath hold, each with significant drawbacks. While external and implanted markers for tracking have limitations, dual energy (DE) imaging offers a noninvasive, markerless solution that enhances soft tissue contrast and improves real-time tumor localization accuracy and precision. PURPOSE: This study aims to develop a markerless real-time DE tumor localization technique on a clinical room-mounted x-ray image guidance system to allow precise 3D stereoscopic and monoscopic lung tumor motion monitoring during radiotherapy. METHODS: A motorized programmable breathing phantom combined with an anthropomorphic phantom was developed to simulate a lung tumor's respiratory motion, with various asymmetric 3D printed tumor models from lung patients. Tumor sizes ranged between 1.0 and 3.3 cm, with some having varying densities and imaged with varying doses. Real-time images were acquired with a clinical ExacTrac stereoscopic imaging system at a rate of 1.67 Hz with high and low energies (140 and 60 kVp). Weighted logarithmic subtraction and an anti-correlated noise reduction algorithm were used to generate DE images. Conventional single energy images (120 kVp) were acquired for comparison. Digital reconstructed radiographs from x-ray imaging views were created to serve as templates for a template-matching algorithm developed to localize tumor locations on x-ray views. For the stereoscopic case where both imaging views were available, 3D triangulation was performed to localize the tumor. In the monoscopic case, when only one x-ray view was available, the 3D tumor position was estimated using a single 2D localization, combined with a 3D probability density function (PDF) describing tumor motion. RESULTS: Stereoscopic DE techniques demonstrated accurate localizations. The monoscopic view obstructed by the spine showed lower success rates than the view obstructed only by the rib bone. In stereoscopic cases, the localization success rates were similar (>96%) between single and DE techniques for large tumor sizes. As tumor sizes decreased, the localization success rates were higher for DE than the single energy technique showing an improvement of up to 25%. Monoscopic results demonstrated the same trend, with DE localization success rates improvement versus single energy by up to 53% for small tumors. DE showed more successful localization for less dense tumors by up to 60% compared to single energy. Tumors imaged with varying mAs values while remaining at optimal kVp settings showed similar localization success rates between single and DE techniques. CONCLUSION: A real-time markerless tumor monitoring technique was developed utilizing a clinical room-mounted stereoscopic/monoscopic image guidance system. DE increases the accuracy of successful tumor localization as compared to the conventional single energy approach, especially for smaller, less dense tumors. The use of PDFs may be a viable approach to monoscopic estimates when only one view is available.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".