Korean Journal of Radiology

  • 제21권10호
  • /
  • Pages.1165-1177
  • /
  • 2020
  • /
  • 1229-6929(pISSN)
  • /
  • 2005-8330(eISSN)
대한영상의학회 (The Korean Society of Radiology)
권호 표지
DOI QR코드

DOI QR Code

Incremental Image Noise Reduction in Coronary CT Angiography Using a Deep Learning-Based Technique with Iterative Reconstruction

  • Jung Hee Hong (Department of Radiology, Seoul National University College of Medicine, Seoul National University Hospital) ;
  • Eun-Ah Park (Department of Radiology, Seoul National University College of Medicine, Seoul National University Hospital) ;
  • Whal Lee (Department of Radiology, Seoul National University College of Medicine, Seoul National University Hospital) ;
  • Chulkyun Ahn (Department of Transdisciplinary Studies, Graduate School of Convergence Science and Technology, Seoul National University) ;
  • Jong-Hyo Kim (Department of Radiology, Seoul National University College of Medicine, Seoul National University Hospital)
  • 투고 : 2019.09.24
  • 심사 : 2020.03.20
  • 발행 : 2020.10.01
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

Objective: To assess the feasibility of applying a deep learning-based denoising technique to coronary CT angiography (CCTA) along with iterative reconstruction for additional noise reduction. Materials and Methods: We retrospectively enrolled 82 consecutive patients (male:female = 60:22; mean age, 67.0 ± 10.8 years) who had undergone both CCTA and invasive coronary artery angiography from March 2017 to June 2018. All included patients underwent CCTA with iterative reconstruction (ADMIRE level 3, Siemens Healthineers). We developed a deep learning based denoising technique (ClariCT.AI, ClariPI), which was based on a modified U-net type convolutional neural net model designed to predict the possible occurrence of low-dose noise in the originals. Denoised images were obtained by subtracting the predicted noise from the originals. Image noise, CT attenuation, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were objectively calculated. The edge rise distance (ERD) was measured as an indicator of image sharpness. Two blinded readers subjectively graded the image quality using a 5-point scale. Diagnostic performance of the CCTA was evaluated based on the presence or absence of significant stenosis (≥ 50% lumen reduction). Results: Objective image qualities (original vs. denoised: image noise, 67.22 ± 25.74 vs. 52.64 ± 27.40; SNR [left main], 21.91 ± 6.38 vs. 30.35 ± 10.46; CNR [left main], 23.24 ± 6.52 vs. 31.93 ± 10.72; all p < 0.001) and subjective image quality (2.45 ± 0.62 vs. 3.65 ± 0.60, p < 0.001) improved significantly in the denoised images. The average ERDs of the denoised images were significantly smaller than those of originals (0.98 ± 0.08 vs. 0.09 ± 0.08, p < 0.001). With regard to diagnostic accuracy, no significant differences were observed among paired comparisons. Conclusion: Application of the deep learning technique along with iterative reconstruction can enhance the noise reduction performance with a significant improvement in objective and subjective image qualities of CCTA images.

키워드

참고문헌

  1. Hamilton-Craig CR, Friedman D, Achenbach S. Cardiac computed tomography-Evidence, limitations and clinical application. Heart Lung Circ 2012;21:70-81 https://doi.org/10.1016/j.hlc.2011.08.070
  2. Task Force Members; Montalescot G, Sechtem U, Achenbach S, Andreotti F, Arden C, Budaj A, et al. 2013 ESC guidelines on the management of stable coronary artery disease: the task force on the management of stable coronary artery disease of the European Society of Cardiology. Eur Heart J 2013;34:2949-3003 https://doi.org/10.1093/eurheartj/eht296
  3. Deseive S, Chen MY, Korosoglou G, Leipsic J, Martuscelli E, Carrascosa P, et al. Prospective randomized trial on radiation dose estimates of CT angiography applying iterative image reconstruction: the PROTECTION V study. JACC Cardiovasc Imaging 2015;8:888-896 https://doi.org/10.1016/j.jcmg.2015.02.024
  4. Halliburton SS, Abbara S, Chen MY, Gentry R, Mahesh M, Raff GL, et al. SCCT guidelines on radiation dose and dose-optimization strategies in cardiovascular CT. J Cardiovasc Comput Tomogr 2011;5:198-224 https://doi.org/10.1016/j.jcct.2011.06.001
  5. Hausleiter J, Meyer TS, Martuscelli E, Spagnolo P, Yamamoto H, Carrascosa P, et al. Image quality and radiation exposure with prospectively ECG-triggered axial scanning for coronary CT angiography: the multicenter, multivendor, randomized PROTECTION-III study. JACC Cardiovasc Imaging 2012;5:484-493 https://doi.org/10.1016/j.jcmg.2011.12.017
  6. Hirshfeld JW Jr, Ferrari VA, Bengel FM, Bergersen L, Chambers CE, Einstein AJ, et al. 2018 ACC/HRS/NASCI/SCAI/SCCT expert consensus document on optimal use of ionizing radiation in cardiovascular imaging: best practices for safety and effectiveness: a report of the American College of Cardiology task force on expert consensus decision pathways. J Am Coll Cardiol 2018;71:e283-e351 https://doi.org/10.1016/j.jacc.2018.02.016
  7. Stocker TJ, Deseive S, Leipsic J, Hadamitzky M, Chen MY, Rubinshtein R, et al. Reduction in radiation exposure in cardiovascular computed tomography imaging: results from the PROspective multicenter registry on radiaTion dose Estimates of cardiac CT angIOgraphy iN daily practice in 2017 (PROTECTION VI). Eur Heart J 2018;39:3715-3723 https://doi.org/10.1093/eurheartj/ehy546
  8. Chen H, Zhang Y, Kalra MK, Lin F, Chen Y, Liao P, et al. Low-dose CT with a residual encoder-decoder convolutional neural network. IEEE Trans Med Imaging 2017;36:2524-2535 https://doi.org/10.1109/TMI.2017.2715284
  9. Kang E, Chang W, Yoo J, Ye JC. Deep convolutional framelet denosing for low-dose CT via wavelet residual network. IEEE Trans Med Imaging 2018;37:1358-1369 https://doi.org/10.1109/TMI.2018.2823756
  10. Ahn CK, Jin H, Heo C, Kim JH. Combined low-dose simulation and deep learning for CT denoising: application of ultra-low-dose cardiac CTA. In: Schmidt TG, Chen GH, Bosmans H, eds. Proceedings of SPIE, Volume 10948. Bellingham: SPIE, 2019
  11. Ronneberger O, Fischer P, Brox T. U-net: convolutional networks for biomedical image segmentation. Cornell University, 2015. Available at: https://arxiv.org/abs/1505.04597. Accessed March 8, 2019
  12. Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift [updated Mar 2015]. Cornell University, 2015. Available at: https://arxiv.org/abs/1502.03167. Accessed March 8, 2019
  13. Nair V, Hinton GE. Rectified linear units improve restricted boltzmann machines. In: Furnkranz J, Joachims T. Proceedings of the 27th international conference on international conference on machine learning. Madison: Omnipress, 2010
  14. Kingma DP, Ba J. Adam: a method for stochastic optimization [updated Jan 2017]. Cornell University, 2014. Available at: https://arxiv.org/abs/1412.6980. Accessed March 8, 2019
  15. Abadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, et al.; Google Brain. Tensorflow: a system for large-scale machine learning. In: USENIX Association, eds. 12th USENIX symposium on operating systems design and implementation (Proceedings of OSDI '16). Berkeley: USENIX Association, 2016:265-283
  16. Kim CW, Kim JH. Realistic simulation of reduced-dose CT with noise modeling and sinogram synthesis using DICOM CT images. Med Phys 2014;41:011901
  17. Mangold S, Wichmann JL, Schoepf UJ, Caruso D, Tesche C, Steinberg DH, et al. Diagnostic accuracy of coronary CT angiography using 3rd-generation dual-source CT and automated tube voltage selection: clinical application in a non-obese and obese patient population. Eur Radiol 2017;27:2298-2308 https://doi.org/10.1007/s00330-016-4601-2
  18. Achenbach S, Paul JF, Laurent F, Becker HC, Rengo M, Caudron J, et al. Comparative assessment of image quality for coronary CT angiography with iobitridol and two contrast agents with higher iodine concentrations: iopromide and iomeprol. A multicentre randomized double-blind trial. Eur Radiol 2017;27:821-830 https://doi.org/10.1007/s00330-016-4437-9
  19. Yi Y, Wu W, Lin L, Zhang HZ, Qian H, Shen ZJ, et al. Single-phase coronary artery CT angiography extracted from stress dynamic myocardial CT perfusion on third-generation dual-source CT: validation by coronary angiography. Int J Cardiol 2018;269:343-349 https://doi.org/10.1016/j.ijcard.2018.06.112
  20. Smith SW. Digital signal processors. In: Smith SW, ed. The scientist and engineer's guide to diginal signal processing. San Diego: California Technical Publishing, 1997:503-534
  21. Leipsic J, Nguyen G, Brown J, Sin D, Mayo JR. A prospective evaluation of dose reduction and image quality in chest CT using adaptive statistical iterative reconstruction. AJR Am J Roentgenol 2010;195:1095-1099 https://doi.org/10.2214/AJR.09.4050
  22. Marin D, Nelson RC, Schindera ST, Richard S, Youngblood RS, Yoshizumi TT, et al. Low-tube-voltage, high-tube-current multidetector abdominal CT: improved image quality and decreased radiation dose with adaptive statistical iterative reconstruction algorithm-Initial clinical experience. Radiology 2010;254:145-153 https://doi.org/10.1148/radiol.09090094
  23. Chun EJ, Lee W, Choi YH, Koo BK, Choi SI, Jae HJ, et al. Effects of nitroglycerin on the diagnostic accuracy of electrocardiogram-gated coronary computed tomography angiography. J Comput Assist Tomogr 2008;32:86-92 https://doi.org/10.1097/rct.0b013e318059befa
  24. Moskowitz CS, Pepe MS. Comparing the predictive values of diagnostic tests: sample size and analysis for paired study designs. Clin Trials 2006;3:272-279 https://doi.org/10.1191/1740774506cn147oa
  25. Kang E, Koo HJ, Yang DH, Seo JB, Ye JC. Cycle-consistent adversarial denoising network for multiphase coronary CT angiography. Med Phys 2019;46:550-562 https://doi.org/10.1002/mp.13284
  26. Ahn C, Heo C, Kim JH. Combined low-dose simulation and deep learning for CT denoising: application in ultra-low-dose chest CT. In: Lin F, Fujita H, Kim JH, eds. International forum on medical imaging in Asia 2019, Volume 11050. Bellingham: SPIE, 2019
  27. Ahn CK, Yang Z, Heo C, Jin H, Park B, Kim JH. A deep learning-enabled iterative reconstruction of ultra-low-dose CT: Use of synthetic sinogram-based noise simulation technique. In: Lo JY, Schmidt TG, Chen GH, eds. Medical imaging 2018: physics of medical imaging, Volume 10573. Bellingham: SPIE, 2018
  28. Yang Q, Yan P, Zhang Y, Yu H, Shi Y, Mou X, et al. Low-dose CT image denoising using a generative adversarial network with wasserstein distance and perceptual loss. IEEE Trans Med Imaging 2018;37:1348-1357 https://doi.org/10.1109/TMI.2018.2827462
  29. Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek W. Iterative reconstruction technique for reducing body radiation dose at CT: feasibility study. AJR Am J Roentgenol 2009;193:764-771 https://doi.org/10.2214/AJR.09.2397
  30. Katsura M, Matsuda I, Akahane M, Sato J, Akai H, Yasaka K, et al. Model-based iterative reconstruction technique for radiation dose reduction in chest CT: comparison with the adaptive statistical iterative reconstruction technique. Eur Radiol 2012;22:1613-1623 https://doi.org/10.1007/s00330-012-2452-z
  31. Li T, Tang T, Yang L, Zhang X, Li X, Luo C. Coronary CT angiography with knowledge-based iterative model reconstruction for assessing coronary arteries and non-calcified predominant plaques. Korean J Radiol 2019;20:729-738 https://doi.org/10.3348/kjr.2018.0435
  32. Park C, Choo KS, Kim JH, Nam KJ, Lee JW, Kim JY. Image quality and radiation dose in CT venography using model-based iterative reconstruction at 80 kVp versus adaptive statistical iterative reconstruction-V at 70 kVp. Korean J Radiol 2019;20:1167-1175 https://doi.org/10.3348/kjr.2018.0897
  33. Lim WH, Choi YH, Park JE, Cho YJ, Lee S, Cheon JE, et al. Application of vendor-neutral iterative reconstruction technique to pediatric abdominal computed tomography. Korean J Radiol 2019;20:1358-1367 https://doi.org/10.3348/kjr.2018.0715
  34. Yoo RE, Park EA, Lee W, Shim H, Kim YK, Chung JW, et al. Image quality of adaptive iterative dose reduction 3D of coronary CT angiography of 640-slice CT: comparison with filtered back-projection. Int J Cardiovasc Imaging 2013;29:669-676 https://doi.org/10.1007/s10554-012-0113-6
  35. Wang R, Schoepf UJ, Wu R, Nance JW Jr, Lv B, Yang H, et al. Diagnostic accuracy of coronary CT angiography: comparison of filtered back projection and iterative reconstruction with different strengths. J Comput Assist Tomogr 2014;38:179-184 https://doi.org/10.1097/RCT.0000000000000005
  36. Precht H, Gerke O, Thygesen J, Egstrup K, Auscher S, Waaler D, et al. Image quality in coronary computed tomography angiography: influence of adaptive statistical iterative reconstruction at various radiation dose levels. Acta Radiol 2018;59:1194-1202 https://doi.org/10.1177/0284185117753657
  37. Fuchs TA, Fiechter M, Gebhard C, Stehli J, Ghadri JR, Kazakauskaite E, et al. CT coronary angiography: impact of adapted statistical iterative reconstruction (ASIR) on coronary stenosis and plaque composition analysis. Int J Cardiovasc Imaging 2013;29:719-724 https://doi.org/10.1007/s10554-012-0134-1
  38. Leipsic J, Labounty TM, Heilbron B, Min JK, Mancini GB, Lin FY, et al. Adaptive statistical iterative reconstruction: assessment of image noise and image quality in coronary CT angiography. AJR Am J Roentgenol 2010;195:649-654 https://doi.org/10.2214/AJR.10.4285
  39. Yin WH, Lu B, Li N, Han L, Hou ZH, Wu RZ, et al. Iterative reconstruction to preserve image quality and diagnostic accuracy at reduced radiation dose in coronary CT angiography: an intraindividual comparison. JACC Cardiovasc Imaging 2013;6:1239-1249 https://doi.org/10.1016/j.jcmg.2013.08.008
  40. Moscariello A, Takx RA, Schoepf UJ, Renker M, Zwerner PL, O'Brien TX, et al. Coronary CT angiography: image quality, diagnostic accuracy, and potential for radiation dose reduction using a novel iterative image reconstruction technique-comparison with traditional filtered back projection. Eur Radiol 2011;21:2130-2138 https://doi.org/10.1007/s00330-011-2164-9
  41. Leipsic J, Labounty TM, Heilbron B, Min JK, Mancini GB, Lin FY, et al. Estimated radiation dose reduction using adaptive statistical iterative reconstruction in coronary CT angiography: the ERASIR study. AJR Am J Roentgenol 2010;195:655-660 https://doi.org/10.2214/AJR.10.4288
  42. Renker M, Ramachandra A, Schoepf UJ, Raupach R, Apfaltrer P, Rowe GW, et al. Iterative image reconstruction techniques: applications for cardiac CT. J Cardiovasc Comput Tomogr 2011;5:225-230 https://doi.org/10.1016/j.jcct.2011.05.002
  43. Song JS, Choi EJ, Kim EY, Kwak HS, Han YM. Attenuation-based automatic kilovoltage selection and sinogram-affirmed iterative reconstruction: effects on radiation exposure and image quality of portal-phase liver CT. Korean J Radiol 2015;16:69-79 https://doi.org/10.3348/kjr.2015.16.1.69
  44. Deak PD, Smal Y, Kalender WA. Multisection CT protocols: sex- and age-specific conversion factors used to determine effective dose from dose-length product. Radiology 2010;257:158-166 https://doi.org/10.1148/radiol.10100047