Análisis de modelos de difusión por imágenes de resonancia magnética nuclear con machine learning

dc.contributor.advisorAgulles Pedros, Luis
dc.contributor.authorPrieto González, Leonar Steven
dc.contributor.researchgroupGrupo de Física Médicaspa
dc.date.accessioned2022-08-25T20:20:41Z
dc.date.available2022-08-25T20:20:41Z
dc.date.issued2022-05
dc.descriptionilustraciones, fotografías, gráficasspa
dc.description.abstractEn este trabajo se presenta la caracterización de las curvas de atenuación por difusión vóxel a vóxel de 4 conjuntos de imágenes (uno de próstata, dos de cerebro humano y uno de cerebro ex vivo de una neoplasia benigna). Esta caracterización incluye la determinación de los valores de difusión (D), pseudo-difusión (D∗), perfusión (f) y curtosis (K) usando los métodos de ADC (mono-exponencial) e IVIM (bi-exponencial) con y sin curtosis. Estos valores son utilizados como referencia para entrenar y probar la validez de varios algoritmos de machine learning (ML) que permiten disminuir el tiempo en la caracterización de la atenuación. Para decidir si en un vóxel existe difusión se implementan algoritmos de clasificación (Extra-Tree Classifier (ETC), Regresión logística (LR), C-Support vector (SVC), Extra-Gradient Boost (XGB) y perceptron multicapa (MLP)), evaluados mediante la precisión y el test AUC. Mientras que para estimar los parámetros característicos se implementan métodos de regresión (Regresión lineal (LinR), regr. polinómica (Poly), XGB, Ridge, Lasso, Random Forest (RF), ElasticNet y support-vector machine (SVM)) que son evaluados mediante diferentes métricas de regresión, particularmente, la raíz del error cuadrático medio de la validación cruzada (RMSE CV). El objetivo de este trabajo es aplicar estas herramientas de ML para el análisis de difusión por imágenes de resonancia magnética. Se obtuvieron como mejores clasificadores el ETC y el MLP con una precisión del 94.1 % y 91.7 % respectivamente. Para la estimación de parámetros el mejor algoritmo fue RF; D posee un RMSECV del 8.39 %, D∗ del 3.57 %, f con 4.52 % y K con 3.53 %. Aunque estos resultados pueden ser considerados satisfactorios, es posible que otros algoritmos que no se tuvieron en cuenta en este trabajo puedan reportar un mejor desempeño. El tiempo promedio que los algoritmos de ML tardan en caracterizar 100.000 vóxeles es 18, 998 ± 0, 135 s mientras que mediante métodos convencionales es de 4408 ± 351 s.spa
dc.description.abstractIn this work we present the characterization of the voxel-by-voxel diffusion attenuation curves of 4 sets of images (one from the prostate, two from the human brain and one from the ex vivo brain of a mini pig). This characterization includes the determination of the values of diffusion (D), pseudo-diffusion (D∗), perfusion (f), and kurtosis (K); using ADC (mono-exponential) and IVIM (bi-exponential) considering kurtosis when convenient. These values are used as a reference to train and test the validity of several machine learning (ML) algorithms that allow to reduce the CPU time to characterize the attenuation curves. To decide if there is diffusion in a voxel, classification algorithms are implemented; (Extra-Tree Classifier (ETC), Logistic Regression (LR), C-Support Vector (SVC), Extra-Gradient Boost (XGB) and Multilayer Perceptron (MLP)), were evaluated by precision and the AUC tests. On the other hand, regression methods; (Linear Regres sion (LinR), Polynomial Regr. (Poly), XGB, Ridge, Lasso, Random Forest (RF), Elastic Net and Support-Vector Machines (SVM)) are implemented to estimate the characteristic parameters and are evaluated using different regression metrics, particularly, root mean square error of cross-validation (RMSECV). The main objective of this work is to apply the se ML tools for diffusion analysis in magnetic resonance images. The ETC and the MLP showed the best classifiers with accuracies of 94.1 % and 91.7 %, respectively. For parame ters estimation, the best algorithm was RF; D has an RMSECV of 8.39 %, D∗ of 3.57 %, f of 4.52 % and K of 3.53 %. Although these results can be considered satisfactory, it is possi ble that other algorithms that were not taken into account in this work may report better performance.The average time that ML algorithms take to characterize 100.000 voxels is 18,998 ± 0,135 s while using conventional methods it is 4408 ± 351 s.eng
dc.description.degreelevelMaestríaspa
dc.description.degreenameMagíster en Física Médicaspa
dc.description.researchareaRadiología e Imágenes Diagnósticasspa
dc.format.extentx, 76 páginasspa
dc.format.mimetypeapplication/pdfspa
dc.identifier.instnameUniversidad Nacional de Colombiaspa
dc.identifier.reponameRepositorio Institucional Universidad Nacional de Colombiaspa
dc.identifier.repourlhttps://repositorio.unal.edu.co/spa
dc.identifier.urihttps://repositorio.unal.edu.co/handle/unal/82118
dc.language.isospaspa
dc.publisherUniversidad Nacional de Colombiaspa
dc.publisher.branchUniversidad Nacional de Colombia - Sede Bogotáspa
dc.publisher.departmentDepartamento de Físicaspa
dc.publisher.facultyFacultad de Cienciasspa
dc.publisher.placeBogotá, Colombiaspa
dc.publisher.programBogotá - Ciencias - Maestría en Física Médicaspa
dc.relation.referencesValerij G. Kiselev. Fundamentals of diffusion mri physics. NMR in biomedicine, 30, 3 2017.spa
dc.relation.referencesDenis Le Bihan. What can we see with ivim mri? NeuroImage, 187:56–67, 2 2019.spa
dc.relation.referencesMara Cercignani and Mark A. Horsfield. The physical basis of diffusion-weighted mri. Journal of the neurological sciences, 186 Suppl 1, 5 2001.spa
dc.relation.referencesRadiopaedia.org. Diffusion-weighted imaging — radiology reference article —, 2021.spa
dc.relation.referencesLaura Andrea Pastor Luque and Luis Agulles Pedros. Modelación de difusión en irm, 2020.spa
dc.relation.referencesDenis Le Bihan. Apparent diffusion coefficient and beyond: What diffusion mr imaging can tell us about tissue structure. https://doi.org/10.1148/radiol.13130420, 268:318–322, 8 2013.spa
dc.relation.referencesCarmelo Messina, Rodolfo Bignone, Alberto Bruno, Antonio Bruno, Federico Bruno, Marco Calandri, Damiano Caruso, Pietro Coppolino, Riccardo De Robertis, Francesco Gentili, Irene Grazzini, Raffaele Natella, Paola Scalise, Antonio Barile, Roberto Grassi, and Domenico Albano. Diffusion-weighted imaging in oncology: An update. Cancers, 12:1–28, 6 2020.spa
dc.relation.referencesSaef Izzy, Daniel B. Rubin, Firas S. Ahmed, Feras Akbik, Simone Renault, Katelyn W. Sylvester, Henrikas Vaitkevicius, Jennifer A. Smallwood, Michael M. Givertz, and Steven K. Feske. Cerebrovascular accidents during mechanical circulatory support: New predictors of ischemic and hemorrhagic strokes and outcome. Stroke, 49:1197– 1203, 2018.spa
dc.relation.referencesAshok Srinivasan, R. Dvorak, K. Perni, S. Rohrer, and S. K. Mukherji. Differentiation of benign and malignant pathology in the head and neck using 3t apparent diffusion coefficient values: Early experience. American Journal of Neuroradiology, 29:40–44, 1 2008.spa
dc.relation.referencesYoung Jin Choi, In Sook Lee, You Seon Song, Jeung Il Kim, Kyung Un Choi, and Jong Woon Song. Diagnostic performance of diffusion-weighted (dwi) and dynamic contrast-enhanced (dce) mri for the differentiation of benign from malignant soft-tissue tumors. Journal of magnetic resonance imaging : JMRI, 50:798–809, 9 2019.spa
dc.relation.referencesLluís Agulles Pedros, R. Acosta, Peter Blümler, and Hans Spiess. Diffusion influence on gas magnetic resonance imaging. Universitas Scientiarum, 13:281–289, 09 2008.spa
dc.relation.referencesE. O. Stejskal and J. E. Tanner. Spin diffusion measurements: Spin echoes in the presence of a time-dependent field gradient. The Journal of Chemical Physics, 42:288, 7 1965.spa
dc.relation.referencesD. Le Bihan, E. Breton, D. Lallemand, M. L. Aubin, J. Vignaud, and M. Laval-Jeantet. Separation of diffusion and perfusion in intravoxel incoherent motion mr imaging. https://doi.org/10.1148/radiology.168.2.3393671, 168:497–505, 8 1986.spa
dc.relation.referencesOlaf Dietrich, Andreas Biffar, Andrea Baur-Melnyk, and Maximilian F. Reiser. Tech nical aspects of mr diffusion imaging of the body. European Journal of Radiology, 76:314–322, 2010.spa
dc.relation.referencesFumiyuki Yamasaki, Kaoru Kurisu, Kenichi Satoh, Kazunori Arita, Kazuhiko Sugi yama, Megu Ohtaki, Junko Takaba, Atushi Tominaga, Ryosuke Hanaya, Hiroyuki Yoshioka, Seiji Hama, Yoko Ito, Yoshinori Kajiwara, Kaita Yahara, Taiichi Saito, and Muhamad Thohar Arifin. Apparent diffusion coefficient of human brain tumors at mr imaging. Radiology, 235:985–991, 6 2005.spa
dc.relation.referencesSigne Swerkersson, Oscar Grundberg, Karl Kolbeck, Andreas Carlberg, Sven Nyr ¨ en, ´ and Mikael Skorpil. Optimizing diffusion-weighted magnetic resonance imaging for evaluation of lung tumors: A comparison of respiratory triggered and free breathing techniques. European Journal of Radiology Open, 5:189, 1 2018.spa
dc.relation.referencesFrancesco Giganti, Elena Orsenigo, Antonio Esposito, Damiano Chiari, Annalaura Salerno, Alessandro Ambrosi, Luca Albarello, Elena Mazza, Carlo Staudacher, Ales sandro Del Maschio, and Francesco De Cobelli. Prognostic role of diffusion-weighted mr imaging for resectable gastric cancer. Radiology, 276:444–452, 8 2015.spa
dc.relation.referencesSungmin Woo, Chong Hyun Suh, Sang Youn Kim, Jeong Yeon Cho, Seung Hyup Kim, and Min Hoan Moon. Head-to-head comparison between biparametric and multiparametric mri for the diagnosis of prostate cancer: A systematic review and meta-analysis. AJR. American journal of roentgenology, 211:W226–W241, 11 2018.spa
dc.relation.referencesWeigen Yao, Jie Liu, Jiaju Zheng, Pengcong Lu, Shufang Zou, and Ying Xu. Study on diagnostic value of quantitative parameters of intravoxel incoherent motion diffusion-weighted imaging (ivim-dwi) in prostate cancer. American Journal of Translational Research, 13:3696, 2021.spa
dc.relation.referencesHarri Merisaari, Pekka Taimen, Rakesh Shiradkar, Otto Ettala, Marko Pesola, Jani Saunavaara, Peter J. Bostrom, Anant Madabhushi, Hannu J. Aronen, and Ivan Jam- ¨ bor. Repeatability of radiomics and machine learning for dwi: Short-term repeatability study of 112 patients with prostate cancer. Magnetic resonance in medicine, 83:2293–2309, 6 2020.spa
dc.relation.referencesGaryfallidis E, Brett M, Amirbekian B, Rokem A, van der Walt S, Descoteaux M, Nimmo-Smith I, and Dipy Contributors. Dipy, a library for the analysis of diffusion mri data. frontiers in neuroinformatics. 8, 2014.spa
dc.relation.referencesRafael Neto Henriques, Ariel Rokem, Eleftherios Garyfallidis, Samuel St-Jean, Eric Thomas Peterson, and Marta Morgado Correia. [re] optimization of a free water elimination two-compartment model for diffusion tensor imaging. bioRxiv, page 108795, 2 2017.spa
dc.relation.referencesBrian Hansen and Sune Nørhøj Jespersen. Data for evaluation of fast kurtosis strategies, b-value optimization and exploration of diffusion mri contrast. Scientific Data 2016 3:1, 3:1–5, 8 2016.spa
dc.relation.referencesEric Peterson. Ivim dataset. figshare. dataset., 2016.spa
dc.relation.referencesTim B. Dyrby, William F.C. Baare, Daniel C. Alexander, Jacob Jelsing, Ellen Garde, ´ and Lise V. Søgaard. An ex vivo imaging pipeline for producing high-quality and high-resolution diffusion-weighted imaging datasets. Human brain mapping, 32:544– 563, 4 2011.spa
dc.relation.referencesAndrés Felipe Giraldo Román and Luis Agulles Pedrós. Análisis en imágenes de resonancia magnética pesadas por difusión. Master’s thesis, Universidad Nacional de Colombia, 2014.spa
dc.relation.referencesNational Institute of Biomedical Imaging and Bioengineering (NIBIB). Imagen por resonancia magnetica (irm).spa
dc.relation.referencesJ. J. (Jun John) Sakurai. Advanced quantum mechanics. Addison-Wesley Pub. Co, 1 edition, 1967.spa
dc.relation.referencesDavid J. Griffiths and Darrell F. Schroeter. Introduction to Quantum Mechanics. Cambridge University Press, 3 edition, 8 2018.spa
dc.relation.referencesE. Mark Haacke, Robert W. Brown, Michael R. Thompson, and Ramesh Venkatesan. Magnetic Resonance Imaging - Physical Principles and Sequence Design. John Wiley & Sons, 1 edition, 1999.spa
dc.relation.referencesJoseph Larmor. Ix. a dynamical theory of the electric and luminiferous medium.— part iii. relations with material media. Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, 190:205–300, 12 1897.spa
dc.relation.referencesRobin A. De Graaf. In vivo NMR spectroscopy: Principles and techniques. John Wiley & Sons, 2 edition, 2007.spa
dc.relation.referencesF Reif. Fundamentals of statistical and thermal physics. McGraw-Hill, Inc, 1 edition, 1965.spa
dc.relation.referencesEvert J Blink. Basic mri physics, 2010.spa
dc.relation.referencesMalcolm H Levitt. Spin Dynamics Basics of Nuclear Magnetic Resonance Second edition. Ltd John Wiley & Sons, second edition edition, 2008.spa
dc.relation.referencesRay Freeman. Spin choreography : basic steps in high resolution nmr. page 391, 1999.spa
dc.relation.referencesF. Bloch. Nuclear induction. Physical Review, 70:460, 10 1946.spa
dc.relation.referencesJean Philibert. One and a half century of diffusion: Fick, einstein, before and beyond. 2005.spa
dc.relation.referencesAdolf Fick. Ueber diffusion. Annalen der Physik, 170:59–86, 1855.spa
dc.relation.referencesJ Crank. The mathematics of diffusion. Clerendon Press, 2 edition, 1975.spa
dc.relation.referencesÁngel Franco García. Difusión. ley de Fick, 2010.spa
dc.relation.referencesJaume Gili and Julio Alonso. Introducción biofísica a la resonancia magnética en neuroimagen, 2004.spa
dc.relation.referencesH. C. Torrey. Bloch equations with diffusion terms. Physical Review, 104:563, 11 1956.spa
dc.relation.referencesV M Kenkre, Eiichi Fukushima, and D Sheltraw. Simple solutions of the torrey-bloch equations in the nmr study of molecular diffusion. JOURNAL OF MAGNETIC RESONANCE, 128:62–69, 1997.spa
dc.relation.referencesE. L. Hahn. Spin echoes. Physical Review, 80:580, 11 1950.spa
dc.relation.referencesH. Y. Carr and E. M. Purcell. Effects of diffusion on free precession in nuclear magnetic resonance experiments. Physical Review, 94:630, 5 1954spa
dc.relation.referencesJonathan H. Burdette, Allen D. Elster, and Peter E. Ricci. Acute cerebral infarction: quantification of spin-density and t2 shine-through phenomena on diffusion weighted mr images. Radiology, 212:333–339, 1999.spa
dc.relation.referencesD. Le Bihan, E. Breton, D. Lallemand, M. L. Aubin, J. Vignaud, and M. Laval-Jeantet. Separation of diffusion and perfusion in intravoxel incoherent motion mr imaging. Radiology, 168:497–505, 1988.spa
dc.relation.referencesD. (Denis) Le Bihan, Mami Lima, Christian Federau, and Eric E. Sigmund. Intravoxel incoherent motion (IVIM) MRI : principles and applications. Jenny Stanford Publishing, 1 edition, 2018.spa
dc.relation.referencesAndrew J. Steven, Jiachen Zhuo, and Elias R. Melhem. Diffusion kurtosis imaging: An emerging technique for evaluating the microstructural environment of the brain. http://dx.doi.org/10.2214/AJR.13.11365, 202, 12 2013.spa
dc.relation.referencesMami Iima and Denis Le Bihan. Clinical intravoxel incoherent motion and diffusion mr imaging: Past, present, and future. Radiology, 278:13–32, 1 2016.spa
dc.relation.referencesSteren Chabert and Paola Scifo. Diffusion signal in magnetic resonance imaging: Origin and interpretation in neurosciences. Biological Research, 40:385–400, 2007.spa
dc.relation.referencesJens H. Jensen, Joseph A. Helpern, Anita Ramani, Hanzhang Lu, and Kyle Kaczynski. Diffusional kurtosis imaging: the quantification of non-gaussian water diffusion by means of magnetic resonance imaging. Magnetic resonance in medicine, 53:1432–1440, 2005.spa
dc.relation.referencesRobert V. Mulkern, Steven J. Haker, and Stephan E. Maier. On high b diffusion imaging in the human brain: ruminations and experimental insights. Magnetic resonance imaging, 27:1151, 10 2009.spa
dc.relation.referencesDmitriy A. Yablonskiy, G. Larry Bretthorst, and Joseph J.H. Ackerman. Statistical model for diffusion attenuated mr signal. Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine, 50:664, 10 2003.spa
dc.relation.referencesKevin M. Bennett, Kathleen M. Schmainda, Raoqiong Bennett, Daniel B. Rowe, Hanbing Lu, and James S. Hyde. Characterization of continuously distributed cortical water diffusion rates with a stretched-exponential model. Magnetic resonance in medicine, 50:727–734, 10 2003spa
dc.relation.referencesQiang Zeng, Feina Shi, Jianmin Zhang, Chenhan Ling, Fei Dong, and Biao Jiang. A modified tri-exponential model for multi-b-value diffusion-weighted imaging: A method to detect the strictly diffusion-limited compartment in brain. Frontiers in Neuroscience, 12:102, 2 2018.spa
dc.relation.referencesMatt G. Hall and Thomas R. Barrick. From diffusion-weighted mri to anomalous diffusion imaging. Magnetic resonance in medicine, 59:447–455, 2008.spa
dc.relation.referencesMehryar Mohri, Afshin Rostamizadeh, and Ameet Talwalkar. Foundations of Machine Learning. The MIT Press, 1 edition, 2012.spa
dc.relation.referencesC. E. Brodley and M. A. Friedl. Identifying mislabeled training data. Journal Of Artificial Intelligence Research, 11:131–167, 6 2011.spa
dc.relation.referencesJoaqu´ın Amat Rodrigo. Machine learning w python & scikit-learn.spa
dc.relation.referencesScikit learn 1.0.2 documentation. Cross-validation: evaluating estimator performance, 2022.spa
dc.relation.referencesMarkus Ojala@tkk Fi and Gemma C Garriga. Permutation tests for studying classifier performance markus ojala. Journal of Machine Learning Research, 11:1833–1863, 2010.spa
dc.relation.referencesDavid Powers and Ailab. Evaluation: From precision, recall and f-measure to roc, informedness, markedness & correlation. J. Mach. Learn. Technol, 2:2229–3981spa
dc.relation.referencesS. Madeh Piryonesi and Tamer E. El-Diraby. Data analytics in asset management: Cost-effective prediction of the pavement condition index. Journal of Infrastructure Systems, 26:04019036, 12 2019.spa
dc.relation.referencesAna Rocío Rocío Del Valle Benavides and Juan Manuel Muñoz Pichardo. Curvas roc (receiver-operating-characteristic) y sus aplicaciones, 2017spa
dc.relation.referencesKelly H. Zou, A. James O’Malley, and Laura Mauri. Receiver-operating characteristic analysis for evaluating diagnostic tests and predictive models. Circulation, 115:654– 657, 2 2007.spa
dc.relation.referencesGoogle Developers. Machine learning crash course.spa
dc.relation.referencesChristopher D Brown and Herbert T Davis. Receiver operating characteristics curves and related decision measures: A tutorial. 2005.spa
dc.relation.referencesMorris H. DeGroot and Mark J. Schervish. Probability and Statistics, volume 1. Fourth edition edition, 2011.spa
dc.relation.referencesA. Mood, F. Graybill, and D Boes. Introduction to the Theory of Statistics. McGraw-Hill, 3rd edition, 1974.spa
dc.relation.referencesF. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12:2825–2830, 2011.spa
dc.relation.referencesNadege Corbin and Martina F Callaghan. Imperfect spoiling in variable flip angle t 1 mapping at 7t: Quantifying and minimizing impact. Magn Reson Med, 00:1–16, 2021.spa
dc.relation.referencesSaurav Basu, Thomas Fletcher, and Ross Whitaker. Rician noise removal in diffusion tensor mri. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 4190 LNCS - I:117–125, 2006spa
dc.rightsDerechos reservados al autor, 2022spa
dc.rights.accessrightsinfo:eu-repo/semantics/openAccessspa
dc.rights.licenseReconocimiento 4.0 Internacionalspa
dc.rights.urihttp://creativecommons.org/licenses/by-nc/4.0/spa
dc.subject.ddc610 - Medicina y saludspa
dc.subject.decsImagen de Difusión por Resonancia Magnéticaspa
dc.subject.decsDiffusion Magnetic Resonance Imagingeng
dc.subject.decsProcesamiento de Imagen Asistido por Computadorspa
dc.subject.decsImage Processing, Computer-Assistedeng
dc.subject.proposalImágenes por resonancia magnéticaspa
dc.subject.proposalIRMspa
dc.subject.proposaldifusión tisularspa
dc.subject.proposalADCeng
dc.subject.proposalDWIeng
dc.subject.proposalmovimiento incoherente intravóxelspa
dc.subject.proposalIVIMeng
dc.subject.proposalaprendizaje automáticospa
dc.subject.proposalMagnetic Resonance Imagingeng
dc.subject.proposalMRIeng
dc.subject.proposaltissue diffusioneng
dc.subject.proposalIntravoxel Incoherent Motioneng
dc.subject.proposalMachine Learningeng
dc.titleAnálisis de modelos de difusión por imágenes de resonancia magnética nuclear con machine learningspa
dc.title.translatedAnalysis of nuclear magnetic resonance imaging diffusion models using machine learningeng
dc.typeTrabajo de grado - Maestríaspa
dc.type.coarhttp://purl.org/coar/resource_type/c_bdccspa
dc.type.coarversionhttp://purl.org/coar/version/c_ab4af688f83e57aaspa
dc.type.contentTextspa
dc.type.driverinfo:eu-repo/semantics/masterThesisspa
dc.type.redcolhttp://purl.org/redcol/resource_type/TMspa
dc.type.versioninfo:eu-repo/semantics/acceptedVersionspa
dcterms.audience.professionaldevelopmentEstudiantesspa
dcterms.audience.professionaldevelopmentInvestigadoresspa
dcterms.audience.professionaldevelopmentPúblico generalspa
oaire.accessrightshttp://purl.org/coar/access_right/c_abf2spa

Archivos

Bloque original

Mostrando 1 - 1 de 1
Cargando...
Miniatura
Nombre:
1018452370.2022.pdf
Tamaño:
2.63 MB
Formato:
Adobe Portable Document Format
Descripción:
Tesis de Maestría en Física Médica

Bloque de licencias

Mostrando 1 - 1 de 1
No hay miniatura disponible
Nombre:
license.txt
Tamaño:
1.71 KB
Formato:
Item-specific license agreed upon to submission
Descripción: