Cuantificación del calcio aórtico en imágenes de tomografía usando redes neuronales convolucionales
Resumen
La detección de calcificaciones aórticas con imágenes de tomografía predice enfermedades cardio vasculares pero su cuantificación manual es tediosa. En este trabajo se entrenaron redes neuronales convolucionales (CNNs) para clasificarlas automáticamente. Se analizaron 1415 pacientes de los cuales se conocía la posición de las calcificaciones junto a la geometría aórtica. Se reconstruyeron digitalmente parches axiales, coronales y sagitales centrados en cada candidato a lesión. Las lesiones candidatas consistieron en agrupaciones de píxeles con atenuación superior a 130 HU en torno a la aorta. La arquitectura de las CNNs fueron dos bloques de convolución y max-pooling seguidos de dos capas fully-connected. Como métricas de evaluación se utilizaron el F1 score, la exactitud, la sensibilidad y la especificidad. La red axial obtuvo los mejores resultados de detección en las porciones ascendente y arco, mientras que la red sagital se destacó en la porción descendente, llegando a detectar correctamente el ≈95% de las lesiones.
Citas
Demer, L.L.; Tintut, Y. (2008). Vascular Calcification: Pathobiology of a Multifaceted Disease. Circulation, 117(22), 2938–48.
Allison, M.A.; Hsi, S.; Wassel, C.L.; Morgan, C.; Ix, J.H.; Wright, C.M.; et al. (2012). Calcified Athero sclerosis in Different Vascular Beds and the Risk of Mortality. Arteriosclerosis, Thrombosis, and Vascular Biology, 32(1), 140–6.
Agatston, A.S.; Janowitz, W.R.; Hildner, F.J.; Zusmer, N.R.; Viamonte, M.; Detrano, R. (1990). Quantifica tion of coronary artery calcium using ultrafast com puted tomography. Journal of the American College of Cardiology, 15(4), 827–32.
Alluri, K.; Joshi, P.H.; Henry, T.S.; Blumenthal, R.S.; Nasir, K.; Blaha, M.J. (2015). Scoring of coronary artery calcium scans: History, assumptions, current limitations, and future directions. Atherosclerosis, 239(1), 109–17.
Budoff, M.J.; Nasir, K.; Katz, R.; Takasu, J.; Carr, J.J.; Wong, N.D.; et al. (2011). Thoracic aortic calcification and coronary heart disease events: The multiethnic study of atherosclerosis (MESA). Ath erosclerosis, 215(1), 196–202.
Craiem, D.; Chironi, G.; Casciaro, M.E.; Graf, S.; Si mon, A. (2014). Calcifications of the Thoracic Aorta on Extended Non Contrast Enhanced Cardiac CT. Hendrikse J, editor. PLoS ONE, 9(10), e109584. [7] Amarenco, P.; Cohen, A.; Tzourio, C.; Bertrand, B.; Hommel, M.; Besson, G.; et al. (1994). Atherosclerotic Disease of the Aortic Arch and the Risk of Ischemic Stroke. New England Journal of Medicine, 331(22), 1474–9.
Craiem, D.; Casciaro, M.; Pascaner, A.; Soulat, G.; Guilenea, F.; Sirieix, M.-E.; et al. (2020). Associa tion of calcium density in the thoracic aorta with risk factors and clinical events. European Radiology, 30(7), 3960–7. [9] Craiem, D.; Chironi, G.; Casciaro, M.E.; Sirieix, M.-E.; Mousseaux, E.; Simon, A. (2016). Association of thoracic aorta calcium and non cardiac vascular events in cardiac disease-free individuals. Athero sclerosis, 245, 22–7.
Hecht, H.S.; Cronin, P.; Blaha, M.J.; Budoff, M.J.; Kazerooni, E.A.; Narula, J.; et al. (2017). 2016 SCCT/ STR guidelines for coronary artery calcium scoring of noncontrast noncardiac chest CT scans: A report of the Society of Cardiovascular Computed Tomog raphy and Society of Thoracic Radiology. Journal of Cardiovascular Computed Tomography, 11(1), 74–84.
Weininger, M.; Ritz, K.S.; Schoepf, U.J.; Flohr, T.G.; Vliegenthart, R.; Costello, P.; et al. (2012). Interplat form Reproducibility of CT Coronary Calcium Scor ing Software. Radiology, 265(1), 70–7.
Abukhaled, N.F.; Alskaini, A.M.; Alziadey, H.; Sulei man, I.; Alsaileek, A.; Al-Mallah, M.H. (2013). The reproducibility of coronary calcium scoring on mul tiple software platforms. Journal of the Saudi Heart Association, 25(2), 151–2.
Santini, G.; Latta, D.D.; Martini, N.; Valvano, G.; Gori, A.; Ripoli, A.; et al. (2018). An automatic deep learning approach for coronary artery calcium seg mentation. In: Eskola H, Väisänen O, Viik J, Hyt tinen J, editors. EMBEC & NBC 2017, Singapore: Springer Singapore; p. 374–7. (IFMBE Proceed ings; vol. 65). Recuperado de: http://link.springer. com/10.1007/978-981-10-5122-7_94
Shadmi, R.; Mazo, V.; Bregman-Amitai, O.; El nekave, E. (2018). Fully-convolutional deep-learn ing based system for coronary calcium score pre diction from non-contrast chest CT. In: 2018 IEEE 15th International Symposium on Biomedical Im aging (ISBI 2018), Washington, DC: IEEE; p. 24–8. Recuperado de: https://ieeexplore.ieee.org/docu ment/8363515/
Išgum, I.; Rutten, A.; Prokop, M.; Staring, M.; Klein, S.; Pluim, J.P.W.; et al. (2010). Automated aortic cal cium scoring on low-dose chest computed tomog raphy: Automated aortic calcium scoring on chest computed tomography. Medical Physics, 37(2), 714–23. 108
Lessmann, N.; Išgum, I.; Setio, A.A.A.; de Vos, B.D.; Ciompi, F.; de Jong, P.A.; et al. (2016). Deep convo lutional neural networks for automatic coronary calcium scoring in a screening study with low-dose chest CT. In: Tourassi GD, Armato SG, editors. San Diego, California, United States; p. 978511. Recu perado de: http://proceedings.spiedigitallibrary. org/proceeding.aspx?doi=10.1117/12.2216978
Lessmann, N.; van Ginneken, B.; Zreik, M.; de Jong, P.A.; de Vos, B.D.; Viergever, M.A.; et al. (2018). Auto matic Calcium Scoring in Low-Dose Chest CT Using Deep Neural Networks With Dilated Convolutions. IEEE Transactions on Medical Imaging, 37(2), 615–25.
González, G.; Washko, G.R.; Estépar, R.S.J.; Cazor la, M.; Cano Espinosa, C. (2018). Automated Agat ston score computation in non-ECG gated CT scans using deep learning. In: Angelini ED, Landman BA, editors. Medical Imaging 2018: Image Processing, Houston, United States: SPIE; p. 91. Recuperado de: https://www.spiedigitallibrary.org/conference-pro ceedings-of-spie/10574/2293681/Automat ed-Agatston-score-computation-in-non-ECG-gat ed-CT-scans/10.1117/12.2293681.full
Chellamuthu, K.; Liu, J.; Yao, J.; Bagheri, M.; Lu, L.; Sandfort, V.; et al. (2017). Atherosclerotic vas cular calcification detection and segmentation on low dose computed tomography scans using convolutional neural networks. In: 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), Melbourne, Australia: IEEE; p. 388–91. Recuperado de: http://ieeexplore.ieee.org/docu ment/7950544/
van Velzen, S.G.M.; Lessmann, N.; Velthuis, B.K.; Bank, I.E.M.; van den Bongard, D.H.J.G.; Leiner, T.; et al. (2020). Deep Learning for Automatic Calcium Scoring in CT: Validation Using Multiple Cardiac CT and Chest CT Protocols. Radiology, 295(1), 66–79.
(No date). PyCharm: the Python IDE for Profession al Developers by JetBrains. Recuperado de: https:// www.jetbrains.com/pycharm/
James, G.; Witten, D.; Hastie, T.; Tibshirani, R. (2021). An Introduction to Statistical Learning: with Appli cations in R. New York, NY: Springer US; (Springer Texts in Statistics). Recuperado de: https://link. springer.com/10.1007/978-1-0716-1418-1
Zou, H.; Hastie, T. (2005). Regularization and vari able selection via the elastic net. Journal of the Roy al Statistical Society: Series B (Statistical Methodolo gy), 67(2), 301–20.
Srivastava, N.; Hinton, G.; Krizhevsky, A.; Sutskever, I.; Salakhutdinov, R. (2014). Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research, 15(1), 1929–58.
Graffy, P.M.; Liu, J.; O’Connor, S.; Summers, R.M.; Pickhardt, P.J. (2019). Automated segmentation and quantification of aortic calcification at abdom inal CT: application of a deep learning-based algo rithm to a longitudinal screening cohort. Abdominal Radiology, 44(8), 2921–8.
Gernaat, S.A.M.; van Velzen, S.G.M.; Koh, V.; Emaus, M.J.; Išgum, I.; Lessmann, N.; et al. (2018). Automatic quantification of calcifications in the coronary arteries and thoracic aorta on radiothera py planning CT scans of Western and Asian breast cancer patients. Radiotherapy and Oncology, 127(3), 487–92.
Kurugol, S.; Come, C.E.; Diaz, A.A.; Ross, J.C.; Kin ney, G.L.; Black-Shinn, J.L.; et al. (2015). Automated quantitative 3D analysis of aorta size, morphology, and mural calcification distributions: Automated quantitative analysis of aorta morphology and cal cifications. Medical Physics, 42(9), 5467–78.
Noothout, J.; de Vos, B.; Wolterink, J.; Isgum, I. (2018). Automatic segmentation of thoracic aor ta segments in low-dose chest CT. In: Angelini ED, Landman BA, editors. Medical Imaging 2018: Image Processing, Houston, United States: SPIE; p. 63. Recuperado de: https://www.spiedigitallibrary.org/ conference-proceedings-of-spie/10574/2293114/ Automatic-segmentation-of-thoracic-aorta-segmen ts-in-low-dose-chest/10.1117/12.2293114.full
Descargas
Publicado
Cómo citar
Número
Sección
Licencia
Esta obra está bajo una licencia internacional Creative Commons Atribución-NoComercial-CompartirIgual 4.0.
