Cabrera, Facundo M. (2021) Discriminación de la etiología de hipertrofias cardíacas utilizando el enfoque radiomics / Discrimination of the etiology of cardiac hypertrophies using the radiomics approach. Maestría en Ciencias Físicas, Universidad Nacional de Cuyo, Instituto Balseiro.
![[img]](/style/images/fileicons/application_pdf.png)
| PDF (Tesis) Disponible bajo licencia Creative Commons: Reconocimiento - No comercial - Compartir igual. Español 5Mb |
Resumen en español
La resonancia magnética cardiaca (CMR) es uno de los métodos de diagnostico mas utilizados para caracterizar el corazón de forma no invasiva. Existen diversas secuencias dentro de la CMR, siendo la mas utilizada la secuencia de tipo MR-Cine. Sin embargo, con esta secuencia, no es posible identificar a simple vista fibrosis en el tejido miocárdico. Es por ello que se utiliza la secuencia de Late Gadolinium Enhancement (LGE). Dichas imágenes, utilizan gadolinio como agente de contraste, el cual es administrado de forma intravenosa. Los distintos patrones de retención tarda de gadolinio permiten sospechar distintas etiologías. Las desventajas en este caso vienen dadas tanto porque es una técnica invasiva, como por las contraindicaciones del uso de gadolinio en pacientes con deficiencias renales, as como la duración del estudio y tiempo de diagnostico. En este trabajo, se persigue el objetivo de identificar aquellas zonas donde esta presente una lesión, únicamente a partir de la información provista por las imágenes MR-Cine, mediante el uso de técnicas conocidas como Radiomics utilizando Machine Learning, en particular, redes neuronales. La hipótesis de este trabajo consiste en que la información del tejido miocárdico dañado se encuentra presente en la textura de este tipo de imágenes. Esto, junto al uso de Radiomics, permite una alternativa frente a la modalidad LGE para identificar tejido miocárdico infartado, con la ventaja de reducir los tiempos del estudio de forma notable y no necesitar el uso de gadolinio, con lo que ello conlleva. Los resultados obtenidos para el método propuesto alcanzan un 89% de precisión para los datos de validación y un 70% para los datos de test. Estos resultados, muestran el potencial de la técnica propuesta y la necesidad de incrementar el conjunto de datos para obtener una mejor precisión. A su vez, se presenta una predicción local a nivel de paciente en imágenes de tipo SAX MR-Cine indicando zonas del tejido miocárdico dañado, obteniendo resultados prometedores y acercándose a un resultado visual que puede ser de utilidad para profesionales del ámbito clínico como un método alternativo a las modalidades de LGE para cuantificar viabilidad miocárdica.
Resumen en inglés
Cardiac magnetic resonance (CMR) is one of the most widely used diagnostic methods to characterize the heart noninvasively. There are several sequences within CMR, the most commonly used being the MR-Cine type sequence. However, with this sequence, it is not possible to identify brosis in myocardial tissue with the naked eye. For this reason, the Late Gadolinium Enhancement (LGE) sequence is used. These images use gadolinium as contrast agent, which is administered intravenously. The dierent patterns of late gadolinium retention allow dierent etiologies to be suspected. The disadvantages in this case are given both because it is an invasive technique and because of the contraindications for the use of gadolinium in patients with renal deciencies, as well as the duration of the study and diagnostic time. In this work, the aim is to identify those areas where a lesion is present, solely from the information provided by MR-Cine images, by using techniques known as Radiomics using Machine Learning, in particular, neural networks. The hypothesis of this work is that the information of damaged myocardial tissue is present in the texture of this type of images. This, together with the use of Radiomics, allows an alternative to the LGE modality for identifying infarcted myocardial tissue, with the advantage of signicantly reducing study times and not requiring the use of gadolinium, with all that this entails. The results obtained for the proposed method reach 89% accuracy for the validation data and 70% accuracy for the test data. These results show the potential of the proposed technique and the need to increase the data set to obtain better accuracy. At the same time, we present a local prediction at the patient level in SAX MR-Cine type images indicating areas of damaged myocardial tissue, obtaining promising results and approaching a visual result that may be useful for clinical professionals as an alternative method to LGE modalities to quantify myocardial viability.
| Tipo de objeto: | Tesis (Maestría en Ciencias Físicas) |
|---|---|
| Palabras Clave: | Neural networks; Redes neuronales; Magnetic resonance; Resonancia magnética; [Myocardial fibrosis; Fibrosis miocárdica; Tissue caracterization; Caracterización tisular; Texture analysis; Análisis de textura; Radiomics] |
| Referencias: | [1] WHO. Cardiovascular diseases (cvds), 2021. URL https://www.who.int/en/ news-room/fact-sheets/detail/cardiovascular-diseases-(cvds). 1 [2] Maceira, A. M., Joshi, J., Prasad, S. K., Moon, J. C., Perugini, E., Harding, I., et al. Cardiovascular magnetic resonance in cardiac amyloidosis. Circulation, 111 (2), 186-193, 2005. 1 [3] Moon, J. C., Reed, E., Sheppard, M. N., Elkington, A. G., Ho, S., Burke, M., et al. The histologic basis of late gadolinium enhancement cardiovascular magnetic resonance in hypertrophic cardiomyopathy. Journal of the American College of Cardiology, 43 (12), 2260-2264, 2004. 1 [4] Rodrguez Jornet, A., Andreu Navarro, F. J., Orellana Fernandez, R., Ibeas Lopez, J., Fortuño Andres, J. R. Gadolinium-induced systemic brosis in severe renal failure. Nefrologa (English Edition), 29 (4), 358-363, 2009. 1 [5] Hassani, C., Saremi, F., Varghese, B. A., Duddalwar, V. Myocardial radiomics in cardiac mri. American Journal of Roentgenology, 214 (3), 536-545, 2020. 2, 13 [6] Lambin, P., Rios-Velazquez, E., Leijenaar, R., Carvalho, S., Van Stiphout, R. G., Granton, P., et al. Radiomics: extracting more information from medical images using advanced feature analysis. European journal of cancer, 48 (4), 441-446, 2012. 2, 13 [7] Yip, S. S., Liu, Y., Parmar, C., Li, Q., Liu, S., Qu, F., et al. Associations between radiologist-dened semantic and automatically computed radiomic features in nonsmall cell lung cancer. Scientic reports, 7 (1), 1-11, 2017. 2, 12, 13 [8] Cetin, I., Petersen, S. E., Napel, S., Camara, O., Ballester, M. A. G., Lekadir, K. A radiomics approach to analyze cardiac alterations in hypertension. En: 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), pags. 640-643. IEEE, 2019. 2, 13 [9] Cerqueira, M. D., Weissman, N. J., Dilsizian, V., Jacobs, A. K., Kaul, S., Laskey, W. K., et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart. tomo 105, pags. 539-542. 2002. 2 [10] Le, M., Zarinabad, N., D'Angelo, T., Mia, I., Heinke, R., Vogl, T., et al. Subsegmental quantication of single (stress)-pass perfusion cmr improves the diagnostic accuracy for detection of obstructive coronary artery disease. Journal of Cardiovascular Magnetic Resonance, 22, 14, 02 2020. 2 [11] Python software foundation. URL http://www.python.org. 4 [12] Harris, C. R., Millman, K. J., van der Walt et al, S. Array programming with numpy. Nature, 585, 357-362, 2020. 4 [13] Virtanen, P., Gommers, R., Oliphant, T. E., Haberland, M., Reddy, T., Cournapeau, D., et al. SciPy 1.0: Fundamental Algorithms for Scientic Computing in Python. Nature Methods, 17, 261-272, 2020. 4 [14] Hunter, J. D. Matplotlib: A 2d graphics environment. Computing in Science & Engineering, 9 (3), 90-95, 2007. 4 [15] Lowekamp, B. C., Chen, D. T., Ibañez, L., Blezek, D. The design of simpleitk. Frontiers in neuroinformatics, 7, 45, 2013. 4 [16] Maier, O. Medpy. URL https://pypi.python.org/pypi/MedPy. 4 [17] Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., et al. TensorFlow: Large-scale machine learning on heterogeneous systems, 2015. URL http://tensorflow.org/, software available from tensorow.org. 4 [18] Chollet, F., et al. Keras, 2015. URL https://github.com/fchollet/keras. 4 [19] Hornak, J. Basics of MRI. Rochester Institute of Technology, 1997. 9 [20] Halse, M. E. Terranova-mri efnmr student guide. cap. 1,3,4,8,9. Magritek, 2016.10 [21] Carr, H. Steady-state free precession in nuclear magnetic resonance. Physical Review, 112 (5), 1693, 1958. 11 [22] Markl, M., Leupold, J. Gradient echo imaging. Journal of Magnetic Resonance Imaging, 35 (6), 1274-1289, 2012. 11 [23] Doltra, A., Hoyem Amundsen, B., Gebker, R., Fleck, E., Kelle, S. Emerging concepts for myocardial late gadolinium enhancement mri. Current cardiology reviews, 9 (3), 185-190, 2013. 11 [24] Bydder, G. M., Young, I. R. Mr imaging: clinical use of the inversion recovery sequence. J Comput Assist Tomogr, 9 (4), 659{675, 1985. 12 [25] Kim, R. J., Wu, E., Rafael, A., Chen, E.-L., Parker, M. A., Simonetti, O., et al. The use of contrast-enhanced magnetic resonance imaging to identify reversible myocardial dysfunction. New England Journal of Medicine, 343 (20), 1445-1453, 2000. 12 [26] Stefano, A., Comelli, A., Bravata, V., Barone, S., Daskalovski, I., Savoca, G., et al. A preliminary pet radiomics study of brain metastases using a fully automatic segmentation method. BMC bioinformatics, 21 (8), 1-14, 2020. 13 [27] Chaddad, A., Desrosiers, C., Niazi, T. Deep radiomic analysis of mri related to alzheimer's disease. IEEE Access, 6, 58213-58221, 2018. 13 [28] Geron, A. Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow: Concepts, tools, and techniques to build intelligent systems. O'Reilly Media, 2019. 14, 15, 16, 17, 28 [29] Chollet, F., et al. Deep learning with Python, tomo 361. Manning New York, 2018. 14, 15, 16, 17, 28 [30] Janocha, K., Czarnecki, W. M. On loss functions for deep neural networks in classication. arXiv preprint arXiv:1702.05659, 2017. 15 [31] Bottou, L. Stochastic gradient learning in neural networks. Proceedings of NeuroNmes, 91 (8), 12, 1991. 15 [32] Hecht-Nielsen, R. Theory of the backpropagation neural network. En: Neural networks for perception, pags. 65-93. Elsevier, 1992. 15 [33] Schwing, A. G., Urtasun, R. Fully connected deep structured networks. arXiv preprint arXiv:1503.02351, 2015. 16 [34] Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R. Dropout: a simple way to prevent neural networks from overtting. The journal of machine learning research, 15 (1), 1929-1958, 2014. 16, 18 [35] Gu, J., Wang, Z., Kuen, J., Ma, L., Shahroudy, A., Shuai, B., et al. Recent advances in convolutional neural networks. Pattern Recognition, 77, 354-377, 2018. URL https://www.sciencedirect.com/science/article/pii/S0031320317304120. 17 [36] Ioe, S., Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. En: International conference on machine learning, pags. 448-456. PMLR, 2015. 18 [37] Cunningham, P., Delany, S. J. k-nearest neighbour classiers-a tutorial. ACM Computing Surveys (CSUR), 54 (6), 1-25, 2021. 22 [38] Beauchemin, M., Thomson, K. P., Edwards, G. On the hausdor distance used for the evaluation of segmentation results. Canadian journal of remote sensing, 24 (1), 3-8, 1998. 23, 31 [39] Li, Y., Yuan, Y. Convergence analysis of two-layer neural networks with relu activation. arXiv preprint arXiv:1705.09886, 2017. 28 [40] Larroza, A., Lopez-Lereu, M. P., Monmeneu, J. V., Gavara, J., Chorro, F. J., Bod, V., et al. Texture analysis of cardiac cine magnetic resonance imaging to detect nonviable segments in patients with chronic myocardial infarction. Medical physics, 45 (4), 1471-1480, 2018. 41 [41] Fedorov, A., Beichel, R., Kalpathy-Cramer, J., Finet, J., Fillion-Robin, J.-C., Pujol, S., et al. 3d slicer as an image computing platform for the quantitative imaging network, 2012. URL http://www.sciencedirect.com/science/article/pii/S0730725X12001816, quantitative Imaging in Cancer. 41 |
| Materias: | Medicina > Procesamiento de imágenes |
| Divisiones: | Gcia. de área de Investigación y aplicaciones no nucleares > Gcia. de Física > Física médica |
| Código ID: | 1060 |
| Depositado Por: | Tamara Cárcamo |
| Depositado En: | 15 Jun 2022 15:58 |
| Última Modificación: | 15 Jun 2022 15:58 |
Personal del repositorio solamente: página de control del documento
