Radiation doses accumulated during very complicated image-guided x-ray procedures have the potential to cause stochastic, but also deterministic effects, such as skin rashes or even hair loss. To monitor and reduce radiation-related risks to patients’ skin, x-ray imaging devices are equipped with online air kerma monitoring components. Traditionally, such measurements have been used to estimate skin entrance dose by (a) estimating air kerma at the interventional reference point (IRP), (b) forward projecting the dose distribution, and (c) considering a backscatter factor among other correction factors. Unfortunately, the complicated interaction between incident x-ray photons, secondary electrons, and skin tissue cannot be properly accounted for by assuming a linear relationship between forward projected air kerma and a backscatter factor. Gold standard skin dose models are therefore determined using Monte Carlo (MC) techniques. However, MC simulations are computationally complex in general and possible acceleration mainly depends on the employed hardware and variance reduction techniques. To obtain reliable and fast dose estimates, we propose to combine MC-based simulations with learning-based methods.
The basic idea of our method is to approximate the radiation physics to calculate a first-order exposure estimate quickly. This initial estimate is then refined using prior knowledge derived from MC simulations. To this end, the primary photon propagation inside a voxelized patient model is estimated using a less accurate but fast photon ray casting (RC) simulation based on the Beer–Lambert law. The results of the RC simulation are then fed into a convolutional neural network (CNN), which maps the propagation of primary photons to the dose deposition inside the patient model. Additionally, the patient model itself including anatomy and material properties, such as mass density and mass energy-absorption coefficients, are fed into the CNN as well. The CNN is trained using smoothed results of MC simulations as output and RC simulations of identical imaging settings and patient models as input.
In total, 163 MC and associated RC simulations are carried out for the head, thorax, abdomen, and pelvis in three different voxel phantoms. We used or primarily emitted photons sampled from a 125 kV peak voltage spectrum, respectively. Edge-preserving smoothing (EPS) is applied to reduce (a) general stochastic uncertainties and (b) stochastic uncertainty concerning MC simulations of less primary photons. The CNN is trained using seven imaging settings of the abdomen in a single phantom. Testing its performance on the remaining datasets, the CNN is capable of estimating skin dose with an error of below 10% for the majority of test cases.
The combination of deep neural networks and MC simulation of particle physics has the potential to decrease the computational complexity of accurate skin dose estimation. The proposed approach can provide dose distributions in under one second when running on high-end hardware. On lower cost hardware, it took up to 2 min to arrive at the same result. This makes our approach applicable in high-end environments as well as in budget solutions. Furthermore, the number of primary photons only affects the training time, while the execution time is independent of the number of primary photons.
|mp13758-sup-0001-SupinfoS1.pdfPDF document, 285.2 KB|
|mp13758-sup-0002-SupinfoS2.pdfPDF document, 84.1 KB|
|mp13758-sup-0003-SupinfoS3.pdfPDF document, 94.4 KB|
Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
- 1, , , et al. Radiation-induced temporary hair loss after endovascular embolization of the cerebral arteries: six cases. Ann Dermatol Venereol. 2002; 129: 703.
- 2, , , Skin injuries from fluoroscopically guided procedures: part 1, characteristics of radiation injury. Am J Roentgenol. 2001; 177: 3.
- 3, , , Severe skin reactions from interventional fluoroscopy: case report and review of the literature. Radiology. 1999; 213: 773.
- 4 Radiation dose management for fluoroscopically guided interventional medical procedures. Med Phys. 2012; 39: 5789.
- 5, , , et al. STROBE-radiation ulcer: an overlooked complication of fluoroscopic intervention: a cross-sectional study. Medicine. 2015; 94: e2178.
- 6, , , et al. The appropriate and justified use of medical radiation in cardiovascular imaging: a position document of the ESC Associations of Cardiovascular Imaging, Percutaneous Cardiovascular Interventions and Electrophysiology. Eur Heart J. 2014; 35: 665.
- 7, , , et al. Use of diagnostic imaging studies and associated radiation exposure for patients enrolled in large integrated health care systems, 1996–2010. JAMA. 2012; 307: 2400.
- 8, , , et al. Chronic radiodermatitis following cardiac catheterization. Arch Dermatol. 1996; 123: 663.
- 9 Challenges in radiation protection of patients for the 21st century. Am J Roentgenol. 2013; 200: 762.
- 10, , , , , Radiation-induced skin injury after complex endovascular procedures. J Vasc Surg. 2014; 60: 742.
- 11, Characterizing the phenomenon of radiation recall dermatitis. Radiother Oncol. 2001; 59: 237.
- 12, , Radiation dermatitis: clinical presentation, pathophysiology, and treatment. J Am Acad Dermatol. 2006; 54: 28.
- 13, , , et al. A new method of realtime skin dose visualization. Clinical evaluation of fluoroscopically guided interventions. Neuroradiology. 2014; 56: 971.
- 14 Methods for measuring fluoroscopic skin dose. Pediatr Radiol. 2006; 36: 136.
- 15, , A tracking system to calculate patient skin dose in real-time during neurointerventional procedures using a biplane x-ray imaging system. Med Phys. 2016; 43: 5131.
- 16, , Patient dose map indications on interventional x-ray systems and validation with Gafchromic xr-rv3 film. Radiat Prot Dosimetry. 2014; 163: 306.
- 17, , , , , Skin dose mapping for fluoroscopically guided interventions. Med Phys. 2011; 38: 5490.
- 18, , , , Calculation of backscatter factors for diagnostic radiology using Monte Carlo methods. Phys Med Biol. 1998; 43: 2237.
- 19, , , Backscatter factors and mass energy-absorption coefficient ratios for diagnostic radiology dosimetry. Phys Med Biol. 2011; 56: 7179.
- 20, , , A system to track skin dose for neuro-interventional cone-beam computed tomography (CBCT) Proc SPIE Int Soc Opt Eng. 2016; 9783. https://doi.org/doi.org/10.1117/12.2216931
- 21, , , , Skin dose mapping for non-uniform xray fields using a backscatter point spread function. Proc SPIE Int Soc Opt Eng. 2017; 10132. https://doi.org/10.1117/12.2254257
- 22, , , Monte Carlo investigation of backscatter point spread function for x-ray imaging examinations. Proc SPIE Int Soc Opt Eng. 2017; 10132. https://doi.org10.1117/12.2254064
- 23, , , A dose point kernel database using gate monte carlo simulation toolkit for nuclear medicine applications: comparison with other monte carlo codes. Med Phys. 2012; 39: 5238.
- 24, , A dose kernel based analytical algorithm for voxel dose calculation in nuclear medicine. Physica Med. 2016; 32: 313.
- 25, , , Evaluating the application of tissue-specific dose kernels instead of water dose kernels in internal dosimetry: a monte carlo study. Cancer Biother Radiopharm. 2016; 31: 367.
- 26, Accelerating Monte Carlo simulations of photon transport in a voxelized geometry using a massively parallel graphics processing unit. Med Phys. 2009; 36: 4878.
- 27, , , et al. Geant4-based Monte Carlo simulations on GPU for medical applications. Phys Med Biol. 2013; 58: 5593.
- 28, , Brox T. U-Net: convolutional networks for biomedical image segmentation. CoRR. 2015. https://www.abs/1505.04597
- 29 Institute of Radiation Medicine. Virtual human database. https://www.helmholtz-muenchen.de/en/irm/service/virtual-human-download-portal/virtual-humandatabase/index.html. Accessed 20.12.2018.
- 30, , , , , Generation of personalized computational phantoms using only patient metadata. In: IEEE Nuclear Science Symposium and Medical Imaging Conference. 2017.
- 31, , , , , A machine learning pipeline for internal anatomical landmark embedding based on a patient surface model. Int J Comput Assist Radiol Surg. 2019; 14: 53.
- 32 International Commission on Radiological Protection. ICRP publication 110: adult reference computation phantoms. Ann ICRP 2009; 39.
- 33, An accurate method for computer-generating tungsten anode x-ray spectra from 30 to 140 kv. Med Phys. 1997; 24: 1661.
- 34, , , et al. Geant4-a simulation toolkit. Nucl Instrum Meth A. 2003; 506: 250.
- 35 , , et al. A review of the use and potential of the GATE Monte Carlo code for radiation therapy and dosimetry applications. Med Phys. 2014; 41: 064301.
- 36, , , et al. Recent developments in geant4. Nucl Instrum Meth A. 2016; 835: 186.
- 37, . Bilateral filtering for gray and color images. In: Sixth International Conference on Computer Vision. 1998: 839–846.
- 38, . A fast voxel traversal algorithm for ray tracing. In: Proceedings of Eurographics. 1987.
- 39, . Adam: a method for stochastic optimization. CoRR. 2014. https://www.abs/1412.6980:arXiv:1412.6980
- 40https://www.anleitungen.rrze.fau.de/hpc/woodycluster. Accessed 01.08.2019.
- 41, , How accurately can the peak skin dose in fluoroscopy be determined using indirect dose metrics. Med Phys. 2014; 41: 071913.
- 42, , , , Fluoroscopically guided interventional procedures: a review of radiation effects on patients’ skin and hair. Radiology. 2010; 254: 326.
- 43, , , , , , . Precision learning: towards use of known operators in neural networks. CoRR. 2017. https://www.abs/1712.00374:arXiv:1712.00374
- 44, , . Conditional random fields: probabilistic models for segmenting and labeling sequence data, In: Proceedings of the Eighteenth International Conference on Machine Learning. 2001: 282–289.
- 45, , . V-net: Fully convolutional neural networks for volumetric medical image segmentation, 2016 Fourth International Conference on 3D Vision (3DV), 2016: 565.
- 46, , , et al. Automatic Multi-Organ Segmentation in Dual Energy CT using 3D Fully Convolutional Network. In: Proceedings of the first International Conference on Medical Imaging with Deep Learning, edited by van Ginneken B and Welling M, 2018.
- 47, , , et al. Enabling machine learning in x-ray-based procedures via realistic simulation of image formation. Int J Comput Assist Radiol Surg. 2019; 10. https://doi.org/10.1007/s11548-019-02011-2