Program Information
Improving CBCT Image Quality Using Machine Learning and Auto-Context Model
X Yang*, Y Lei , K Higgins , X Dong , A Dhabaan , E Elder , X Jiang , W Curran , X Tang , Emory Univ, Atlanta, GA
Presentations
SU-F-201-4 (Sunday, July 30, 2017) 2:05 PM - 3:00 PM Room: 201
Purpose: Quantitative cone beam CT (CBCT) imaging is on increasing demand for precise image-guided radiation therapy because it provides a foundation for advanced image-guided techniques, including accurate treatment setup, online tumor delineation and patient dose calculation. However, due to severe artifacts, the current CBCT is limited to patient setup. This study’s purpose is to develop a learning-based approach to improve CBCT image quality for quantitative analysis during adaptive radiotherapy.
Methods: We propose to integrate auto-context model and anatomical features into a machine learning framework to iteratively predict the corrected CBCT with high image quality. The first step is to remove the uninformative regions, reduce noise, and perform an alignment between CT and CBCT. We then partition a given CBCT image into a set of patches. The most informative and salient anatomical features are extracted to train random forests. For each patch, we use the random forest to directly predict a corrected CBCT patch as an output. Moreover, we utilize an auto-context model to iteratively refine the prediction. Finally, we combine all of the predicted CBCT patches to obtain the final corrected CBCT image. We performed leave-one-out cross-validation method to evaluate our leaning-based correction algorithm.
Results: This prediction-based correction algorithm was evaluated using 12 patients with CBCT and CT images. The mean absolute error (MAE) and structural similarity (SSIM) indexes were used to quantify the correction accuracy of the prediction algorithm. The mean MAE and SSIM were 16.16±2.04 and 0.92±0.01, which demonstrated the corrected CBCT prediction accuracy of the proposed learning-based method.
Conclusion: To improve CBCT imaging, we have developed a novel learning-based method demonstrated that this method could significantly improve CBCT image quality. The proposed method has great potential in improving CBCT image quality to a level close to planning CT, therefore, allowing its quantitative use in adaptive radiotherapy.
Contact Email: