Stanford Advisors


  • Lei Xing, Postdoctoral Faculty Sponsor

All Publications


  • Calibrated uncertainty estimation for interpretable proton computed tomography image correction using Bayesian deep learning. Physics in medicine and biology Nomura, Y. n., Tanaka, S. n., Wang, J. n., Shirato, H. n., Shimizu, S. n., Xing, L. n. 2021

    Abstract

    Integrated-type proton computed tomography (pCT) measures proton stopping power ratio (SPR) images for proton therapy treatment planning, but its image quality is degraded due to noise and scatter. Although several correction methods have been proposed, techniques that include estimation of uncertainty are limited. This study proposes a novel uncertainty-aware pCT image correction method using a Bayesian convolutional neural network (BCNN). A DenseNet-based BCNN was constructed to predict both a corrected SPR image and its uncertainty from a noisy SPR image. A total 432 noisy SPR images of 6 non-anthropomorphic and 3 head phantoms were collected with Monte Carlo simulations, while true noise-free images were calculated with known geometric and chemical components. Heteroscedastic loss and deep ensemble techniques were performed to estimate aleatoric and epistemic uncertainties by training 25 unique BCNN models. 200-epoch end-to-end training was performed for each model independently. Feasibility of the predicted uncertainty was demonstrated after applying two post-hoc calibrations and calculating spot-specific path length uncertainty distribution. For evaluation, accuracy of head SPR images and water-equivalent thickness (WET) corrected by the trained BCNN models was compared with a conventional method and non-Bayesian CNN model. BCNN-corrected SPR images represent noise-free images with high accuracy. Mean absolute error in test data was improved from 0.263 for uncorrected images to 0.0538 for BCNN-corrected images. Moreover, the calibrated uncertainty represents accurate confidence levels, and the BCNN-corrected calibrated WET was more accurate than non-Bayesian CNN with high statistical significance. Computation time for calculating one image and its uncertainties with 25 BCNN models is 0.7 seconds with a consumer grade GPU. Our model is able to predict accurate pCT images as well as two types of uncertainty. These uncertainties will be useful to identify potential cause of SPR errors and develop a spot-specific range margin criterion, toward elaboration of uncertainty-guided proton therapy.

    View details for DOI 10.1088/1361-6560/abe956

    View details for PubMedID 33626513

  • Fast spot-scanning proton dose calculation method with uncertainty quantification using a three-dimensional convolutional neural network. Physics in medicine and biology Nomura, Y. n., Wang, J. n., Shirato, H. n., Shimizu, S. n., Xing, L. n. 2020

    Abstract

    This study proposes a near-real-time spot-scanning proton dose calculation method with probabilistic uncertainty estimation using a three-dimensional convolutional neural network (3D-CNN).CT images and clinical target volume contours of 215 head and neck cancer patients were collected from a public database. 1,484 and 488 plans were extracted for training and testing the 3D-CNN model, respectively. Spot beam data and single-field uniform dose (SFUD) labels were calculated for each plan using an open-source dose calculation toolkit. Variable spot data were converted into a fixed-size volume hereby called a "peak map" (PM). 300 epochs of end-to-end training was implemented using sets of stopping power ratio and PM as input. Moreover, transfer learning techniques were used to adjust the trained model to SFUD doses calculated with different beam parameters and calculation algorithm using only 7.95% of training data used for the base model. Finally, accuracy of the 3D-CNN-calculated doses and model uncertainty was reviewed with several evaluation metrics.The 3D-CNN model calculates 3D proton dose distributions accurately with a mean absolute error of 0.778 cGyE. The predicted uncertainty is correlated with dose errors at high contrast edges. Averaged Sørensen-Dice similarity coefficients between binarized outputs and ground truths are mostly above 80%. Once the 3D-CNN model was well-trained, it can be efficiently fine-tuned for different proton doses by transfer learning techniques. Inference time for calculating one dose distribution is around 0.8 seconds for a plan using 1,500 spot beams with a consumer grade GPU.A novel spot-scanning proton dose calculation method using 3D-CNN was developed. The 3D-CNN model is able to calculate 3D doses and uncertainty with any SFUD spot data and beam irradiation angles. Our proposed method should be readily extendable to other setups and plans and be useful for dose verification, image-guided proton therapy, or other applications.

    View details for DOI 10.1088/1361-6560/aba164

    View details for PubMedID 32604078

  • Projection-domain scatter correction for cone beam computed tomography using a residual convolutional neural network MEDICAL PHYSICS Nomura, Y., Xu, Q., Shirato, H., Shimizu, S., Xing, L. 2019; 46 (7): 3142–55

    View details for DOI 10.1002/mp.13583

    View details for Web of Science ID 000475671900020

  • Modified fast adaptive scatter kernel superposition (mfASKS) correction and its dosimetric impact on CBCT-based proton therapy dose calculation. Medical physics Nomura, Y. n., Xu, Q. n., Peng, H. n., Takao, S. n., Shimizu, S. n., Xing, L. n., Shirato, H. n. 2019

    Abstract

    While cone beam computed tomography (CBCT) is able to provide patient anatomical information, its image quality is severely degraded due to scatter contamination, which degrades the accuracy of CBCT-based dose distribution estimation in proton therapy. In this work, we combined two existing scatter kernel correction methods: the point-spread function (PSF)-based scatter kernel derivation method and the fast adaptive scatter kernel superposition (fASKS) model, and evaluated the impact of the modified fASKS (mfASKS) correction on the accuracy of proton dose distribution estimation. To evaluate feasibility of the mfASKS approach using accurate scatter distributions, both Monte Carlo simulations and experiments were performed for an on-board CBCT machine integrated with a proton therapy machine.We developed a strategy to modify central intensity, constant intensity, and amplitude of the scatter kernels derived from PSFs for the fASKS model. A parameter required for the fASKS model was derived by optimizing uniformity in the mfASKS-corrected reconstructed images. Subsequently, the mfASKS model was used to remove scatter in CBCT imaging. We quantitatively compared the Hounsfield Unit (HU) and proton stopping power ratio (SPR) images for five different phantoms. To assess improvement of dose calculation accuracy, a series of proton treatment plans were produced using the CBCT images with and without the mfASKS correction.The accuracies of both HU and SPR intensity quantifications are improved as a result of the mfASKS correction. Mean absolute water-equivalent path length difference to the true value decreases from 10.3 to 0.934 mm for the Gammex phantom (simulation). At the same time, mfASKS is able to offer more accurate dose distributions, especially at the distal fall-off region where noticeable dose overestimation is observed in the uncorrected scenario. Mean absolute relative error of proton range in the pelvic phantom improves from 5.03% to 2.57% (experiment).mfASKS enables more accurate CBCT-based proton dose calculation. This technique has significant implications in image-guided radiotherapy and dose verifications in adaptive proton therapy.

    View details for DOI 10.1002/mp.13878

    View details for PubMedID 31661161