Long short-term memory networks for proton dose calculation in highly heterogeneous tissues.

PURPOSE To investigate the feasibility and accuracy of proton dose calculations with artificial neural networks (ANN) in challenging 3D anatomies. METHODS A novel proton dose calculation approach was designed based on the application of a long short-term memory (LSTM) network. It processes the 3D geometry as a sequence of two-dimensional (2D) computed tomography slices and outputs a corresponding sequence of (2D) slices that forms the 3D dose distribution. The general accuracy of the approach is investigated in comparison to Monte Carlo reference simulations and pencil beam dose calculations. We consider both artificial phantom geometries and clinically realistic lung cases for three different pencil beam energies. RESULTS For artificial phantom cases, the trained LSTM model achieved a 98:57% γ -index pass rate ([1%, 3mm]) in comparison to MC simulations for a pencil beam with initial energy 104:25MeV. For a lung patient case, we observe pass rates of 98:56%, 97:74%, and 94:51% for an initial energy of 67:85MeV, 104:25MeV, and 134:68MeV, respectively. Applying the LSTM dose calculation on patient cases that were fully excluded from the training process yields an average - γ index pass rate of 97:85%. CONCLUSIONS LSTM networks are well suited for proton dose calculation tasks. Further research, especially regarding model generalization and computational performance in comparison to established dose calculation methods, is warranted.

[1]  Steve B. Jiang,et al.  GPU-based fast Monte Carlo dose calculation for proton therapy , 2012, Physics in medicine and biology.

[2]  Matthew J. Hausknecht,et al.  Beyond short snippets: Deep networks for video classification , 2015, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[3]  A. Dell'Acqua,et al.  Geant4 - A simulation toolkit , 2003 .

[4]  Trevor Darrell,et al.  Long-term recurrent convolutional networks for visual recognition and description , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[5]  Julia Bauer,et al.  Fast robust dose calculation on GPU for high-precision 1H, 4He, 12C and 16O ion therapy: the FRoG platform , 2018, Scientific Reports.

[6]  Jürgen Schmidhuber,et al.  Framewise phoneme classification with bidirectional LSTM and other neural network architectures , 2005, Neural Networks.

[7]  Wonjun Hwang,et al.  An Efficient Three-Dimensional Convolutional Neural Network for Inferring Physical Interaction Force from Video , 2019, Sensors.

[8]  Yaozong Gao,et al.  Estimating CT Image from MRI Data Using 3D Fully Convolutional Networks , 2016, LABELS/DLMIA@MICCAI.

[9]  Fabio Viola,et al.  The Kinetics Human Action Video Dataset , 2017, ArXiv.

[10]  Vasant Kearney,et al.  DoseNet: a volumetric dose prediction algorithm using 3D fully-convolutional neural networks , 2018, Physics in medicine and biology.

[11]  Sasa Mutic,et al.  Development of a fast Monte Carlo dose calculation system for online adaptive radiation therapy quality assurance , 2017, Physics in medicine and biology.

[12]  M Bangert,et al.  Efficiency of analytical and sampling-based uncertainty propagation in intensity-modulated proton therapy , 2017, Physics in medicine and biology.

[13]  Jan J W Lagendijk,et al.  DeepDose: Towards a fast dose calculation engine for radiation therapy using deep learning , 2020, Physics in medicine and biology.

[14]  Pierre Baldi,et al.  Autoencoders, Unsupervised Learning, and Deep Architectures , 2011, ICML Unsupervised and Transfer Learning.

[15]  S. Webb,et al.  Use of artificial neural networks to predict biological outcomes for patients receiving radical radiotherapy of the prostate. , 2004, Radiotherapy and oncology : journal of the European Society for Therapeutic Radiology and Oncology.

[16]  Steve B. Jiang,et al.  A feasibility study for predicting optimal radiation therapy dose distributions of prostate cancer patients from patient anatomy using deep learning , 2017, Scientific Reports.

[17]  Mark Bangert,et al.  Design and Selection of Machine Learning Methods Using Radiomics and Dosiomics for Normal Tissue Complication Probability Modeling of Xerostomia , 2018, Front. Oncol..

[18]  Mubarak Shah,et al.  An End-to-end 3D Convolutional Neural Network for Action Detection and Segmentation in Videos , 2017, ArXiv.

[19]  D. Low,et al.  A technique for the quantitative evaluation of dose distributions. , 1998, Medical physics.

[20]  Navdeep Jaitly,et al.  Towards End-To-End Speech Recognition with Recurrent Neural Networks , 2014, ICML.

[21]  U. Oelfke,et al.  Two-dimensional pencil beam scaling: an improved proton dose algorithm for heterogeneous media. , 2002, Physics in medicine and biology.

[22]  K Parodi,et al.  Integration and evaluation of automated Monte Carlo simulations in the clinical practice of scanned proton and carbon ion beam therapy , 2014, Physics in medicine and biology.

[23]  David S Followill,et al.  Pencil Beam Algorithms Are Unsuitable for Proton Dose Calculations in Lung. , 2017, International journal of radiation oncology, biology, physics.

[24]  Radhe Mohan,et al.  Monte Carlo simulations for configuring and testing an analytical proton dose-calculation algorithm , 2007, Physics in medicine and biology.

[25]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[26]  Munchurl Kim,et al.  3DSRnet: Video Super-resolution using 3D Convolutional Neural Networks , 2018, ArXiv.

[27]  Katia Parodi,et al.  Development of the open‐source dose calculation and optimization toolkit matRad , 2017, Medical physics.

[28]  Yaozong Gao,et al.  7T‐guided super‐resolution of 3T MRI , 2017, Medical physics.

[29]  M. Fippel,et al.  A Monte Carlo dose calculation algorithm for proton therapy. , 2004, Medical physics.

[30]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[31]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[32]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[33]  Matthias Fippel,et al.  A pencil beam algorithm for intensity modulated proton therapy derived from Monte Carlo simulations , 2005, Physics in medicine and biology.

[34]  Le Song,et al.  Deep Fried Convnets , 2014, 2015 IEEE International Conference on Computer Vision (ICCV).

[35]  Jürgen Schmidhuber,et al.  Recurrent nets that time and count , 2000, Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium.

[36]  Hanne M Kooy,et al.  Dose uncertainties in IMPT for oropharyngeal cancer in the presence of anatomical, range, and setup errors. , 2013, International journal of radiation oncology, biology, physics.

[37]  Thomas Bortfeld,et al.  Reducing the sensitivity of IMPT treatment plans to setup errors and range uncertainties via probabilistic treatment planning. , 2008, Medical physics.

[38]  Volker Tresp,et al.  Tensor-Train Recurrent Neural Networks for Video Classification , 2017, ICML.

[39]  Andrey V. Savchenko,et al.  Compression of Recurrent Neural Networks for Efficient Language Modeling , 2019, Appl. Soft Comput..

[40]  Timothy C. Y. Chan,et al.  Automated Treatment Planning in Radiation Therapy using Generative Adversarial Networks , 2018, MLHC.

[41]  Philipp Hennig,et al.  Analytical probabilistic modeling for radiation therapy treatment planning , 2013, Physics in medicine and biology.

[42]  Zenglin Xu,et al.  Learning Compact Recurrent Neural Networks with Block-Term Tensor Decomposition , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[43]  K Parodi,et al.  A Monte Carlo-based treatment planning tool for proton therapy , 2013, Physics in medicine and biology.

[44]  Jürgen Schmidhuber,et al.  Learning to Forget: Continual Prediction with LSTM , 2000, Neural Computation.

[45]  Steve B. Jiang,et al.  Improving proton dose calculation accuracy by using deep learning , 2020, Mach. Learn. Sci. Technol..

[46]  H Paganetti,et al.  TOPAS: an innovative proton Monte Carlo platform for research and clinical applications. , 2012, Medical physics.

[47]  Philipp Hennig,et al.  Analytical probabilistic modeling of RBE-weighted dose for ion therapy , 2017, Physics in medicine and biology.

[48]  J. Debus,et al.  Dosimetric validation of Monte Carlo and analytical dose engines with raster-scanning 1H, 4He, 12C, and 16O ion-beams using an anthropomorphic phantom. , 2019, Physica medica : PM : an international journal devoted to the applications of physics to medicine and biology : official journal of the Italian Association of Biomedical Physics.

[49]  Wei Liu,et al.  Statistical assessment of proton treatment plans under setup and range uncertainties. , 2013, International journal of radiation oncology, biology, physics.

[50]  Uwe Titt,et al.  Monte Carlo simulations of a nozzle for the treatment of ocular tumours with high-energy proton beams , 2005, Physics in medicine and biology.

[51]  Song Han,et al.  Deep Compression: Compressing Deep Neural Network with Pruning, Trained Quantization and Huffman Coding , 2015, ICLR.

[52]  E. Pedroni,et al.  Dose calculation models for proton treatment planning using a dynamic beam delivery system: an attempt to include density heterogeneity effects in the analytical dose calculation. , 1999, Physics in medicine and biology.

[53]  Quoc V. Le,et al.  Sequence to Sequence Learning with Neural Networks , 2014, NIPS.