Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification

La retinopatía diabética (RD) es una de las complicaciones microvasculares de la diabetes mellitus, que sigue siendo una de las principales causas de ceguera en todo el mundo. Los modelos computacionales basados ​​en redes neuronales convolucionales representan el estado del arte para la detección a...

Full description

Autores:
Gonzalez Osorio, Fabio
Perdomo Charry, Oscar Julian
Toledo Cortes, Santiago
De La Pava, Melissa
Tipo de recurso:
Article of investigation
Fecha de publicación:
2020
Institución:
Escuela Colombiana de Ingeniería Julio Garavito
Repositorio:
Repositorio Institucional ECI
Idioma:
eng
OAI Identifier:
oai:repositorio.escuelaing.edu.co:001/1425
Acceso en línea:
https://repositorio.escuelaing.edu.co/handle/001/1425
Palabra clave:
Retinopatía diabética
Aprendizaje
Método gaussiano
Gaussian method
Deep Learning
Diabetic Retinopathy
Gaussian Process
Uncertainty Quantification
Aprendizaje profundo
Retinopatía diabética
Proceso gaussiano
Cuantificación de la incertidumbre
Rights
openAccess
License
http://purl.org/coar/access_right/c_abf2
id ESCUELAIG2_7aa57b6b4be643e5d983fb53809a3539
oai_identifier_str oai:repositorio.escuelaing.edu.co:001/1425
network_acronym_str ESCUELAIG2
network_name_str Repositorio Institucional ECI
repository_id_str
dc.title.spa.fl_str_mv Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
title Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
spellingShingle Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
Retinopatía diabética
Aprendizaje
Método gaussiano
Gaussian method
Deep Learning
Diabetic Retinopathy
Gaussian Process
Uncertainty Quantification
Aprendizaje profundo
Retinopatía diabética
Proceso gaussiano
Cuantificación de la incertidumbre
title_short Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
title_full Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
title_fullStr Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
title_full_unstemmed Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
title_sort Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty Quantification
dc.creator.fl_str_mv Gonzalez Osorio, Fabio
Perdomo Charry, Oscar Julian
Toledo Cortes, Santiago
De La Pava, Melissa
dc.contributor.author.none.fl_str_mv Gonzalez Osorio, Fabio
Perdomo Charry, Oscar Julian
Toledo Cortes, Santiago
De La Pava, Melissa
dc.contributor.researchgroup.spa.fl_str_mv GiBiome
dc.subject.armarc.none.fl_str_mv Retinopatía diabética
Aprendizaje
topic Retinopatía diabética
Aprendizaje
Método gaussiano
Gaussian method
Deep Learning
Diabetic Retinopathy
Gaussian Process
Uncertainty Quantification
Aprendizaje profundo
Retinopatía diabética
Proceso gaussiano
Cuantificación de la incertidumbre
dc.subject.armarc.spa.fl_str_mv Método gaussiano
dc.subject.armarc.eng.fl_str_mv Gaussian method
dc.subject.proposal.spa.fl_str_mv Deep Learning
Diabetic Retinopathy
Gaussian Process
Uncertainty Quantification
Aprendizaje profundo
Retinopatía diabética
Proceso gaussiano
Cuantificación de la incertidumbre
description La retinopatía diabética (RD) es una de las complicaciones microvasculares de la diabetes mellitus, que sigue siendo una de las principales causas de ceguera en todo el mundo. Los modelos computacionales basados ​​en redes neuronales convolucionales representan el estado del arte para la detección automática de RD utilizando imágenes de fondo de ojo. La mayor parte del trabajo actual aborda este problema como una tarea de clasificación binaria. Sin embargo, incluir la estimación de leyes y la cuantificación de la incertidumbre de las predicciones puede aumentar potencialmente la solidez del modelo. En este artículo, se presenta un método de proceso híbrido de aprendizaje profundo y gaussiano para el diagnóstico de RD y la cuantificación de la incertidumbre. Este método combina el poder de representación del aprendizaje profundo con la capacidad de generalizar a partir de pequeños conjuntos de datos de modelos de procesos gaussianos. Los resultados muestran que la cuantificación de la incertidumbre en las predicciones mejora la interpretabilidad del método como herramienta de apoyo al diagnóstico
publishDate 2020
dc.date.issued.none.fl_str_mv 2020
dc.date.accessioned.none.fl_str_mv 2021-05-12T19:00:05Z
2021-10-01T17:16:54Z
dc.date.available.none.fl_str_mv 2021-05-12
2021-10-01T17:16:54Z
dc.type.spa.fl_str_mv Artículo de revista
dc.type.coarversion.fl_str_mv http://purl.org/coar/version/c_970fb48d4fbd8a85
dc.type.version.spa.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.coar.spa.fl_str_mv http://purl.org/coar/resource_type/c_2df8fbb1
dc.type.content.spa.fl_str_mv Text
dc.type.driver.spa.fl_str_mv info:eu-repo/semantics/article
dc.type.redcol.spa.fl_str_mv http://purl.org/redcol/resource_type/ART
format http://purl.org/coar/resource_type/c_2df8fbb1
status_str publishedVersion
dc.identifier.issn.none.fl_str_mv 0302-9743
dc.identifier.uri.none.fl_str_mv https://repositorio.escuelaing.edu.co/handle/001/1425
dc.identifier.doi.none.fl_str_mv 10.1007/978-3-030-63419-3_21
dc.identifier.url.none.fl_str_mv DOI:10.1007/978-3-030-63419-3_21
identifier_str_mv 0302-9743
10.1007/978-3-030-63419-3_21
DOI:10.1007/978-3-030-63419-3_21
url https://repositorio.escuelaing.edu.co/handle/001/1425
dc.language.iso.spa.fl_str_mv eng
language eng
dc.relation.citationedition.spa.fl_str_mv Lecture Notes in Computer Science (LNCS, volumen 12069)
dc.relation.citationissue.spa.fl_str_mv 206
dc.relation.citationstartpage.spa.fl_str_mv 215
dc.relation.citationvolume.spa.fl_str_mv 12069
dc.relation.indexed.spa.fl_str_mv N/A
dc.relation.ispartofjournal.spa.fl_str_mv Lecture Notes in Computer Science
dc.relation.references.eng.fl_str_mv Abrámoff, M.D., et al.: Automated analysis of retinal images for detection of referable diabetic retinopathy. JAMA Ophthalmol. 131(3), 351–357 (2013).
American Academy of Ophthalmology: International clinical diabetic retinopathy disease severity scale detailed table. International Council of Ophthalmology (2002)
Bradshaw, J., Matthews, A.G.d.G., Ghahramani, Z.: Adversarial examples, uncertainty, and transfer testing robustness in Gaussian process hybrid deep networks, eprint, pp. 1-33 (2017).
Decenciére, E., et al.: Feedback on a publicly distributed image database: the Messidor database. Image Anal. Stereol. 33(3), 231–234 (2014).
Diabetic Retinopathy Detection of Kaggle: Eyepacs challenge.
Ethem, A.: Introduction to Machine Learning, 3rd edn. The MIT Press, Cambridge (2014)
Gargeya, R., Leng, T.: Automated identification of diabetic retinopathy using deep learning. Ophthalmology 124(7), 962–969 (2017)
Gulshan, V., et al.: Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA J. Am. Med. Assoc. 316(22), 2402–2410 (2016)
Kaya, M., Bilge, H.: Deep metric learning: a survey. Symmetry 11, 1066 (2019).
Krause, J., et al.: Grader variability and the importance of reference standards for evaluating machine learning models for diabetic retinopathy. Ophthalmology 125(8), 1264–1272 (2018).
Leibig, C., Allken, V., Ayhan, M.S., Berens, P., Wahl, S.: Leveraging uncertainty information from deep neural networks for disease detection. Sci. Rep. 7(1), 1–14 (2017)
Li, F., Liu, Z., Chen, H., Jiang, M., Zhang, X., Wu, Z.: Automatic detection of diabetic retinopathy in retinal fundus photographs based on deep learning algorithm. Transl. Vis. Sci. Technol. 8(6) (2019).
Lim, Z.W., Lee, M.L., Hsu, W., Wong, T.Y.: Building trust in deep learning system towards automated disease detection. In: The Thirty-First AAAI Conference on Innovative Applications of Artificial Intelligence, pp. 9516–9521 (2018)
Perdomo, O., Gonzalez, F.: A systematic review of deep learning methods applied to ocular images. Ciencia e Ingenieria Neogranadina 30(1), 9–26 (2019)
Raghu, M., Blumer, K., Sayres, R., Obermeyer, Z., Kleinberg, R., Mullainathan, S., Kleinberg, J.: Direct uncertainty prediction for medical second opinions. In: Proceedings of the 36th International Conference on Machine Learning, PMLR 97, Long Beach, California (2019)
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., Wojna, Z.: Rethinking the inception architecture for computer vision. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, December 2016, pp. 2818–2826 (2016).
Voets, M., Møllersen, K., Bongo, L.A.: Reproduction study using public data of: Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. PLoS ONE 14(6), 1–11 (2019)
Wells, J.A., et al.: aflibercept, bevacizumab, or ranibizumab for diabetic macular edema two-year results from a comparative effectiveness randomized clinical trial. Ophthalmology 123(6), 1351–1359 (2016)
Wilkinson, C.P.P., et al.: Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology 110(9), 1677–1682 (2003)
Wilson, A., Nickisch, H.: Kernel interpolation for scalable structured gaussian processes (KISS-GP). In: Proceedings of the 32nd International Conference on Machine Learning. JMLR: W&CP, Lille, France (2015)
Xin, Q., Elliot, M., Miikkulainen, R.: Quantifying point-prediction uncertainty in neural networks via residual estimation with an I/O Kernel. In: ICLR 2020, Addis Ababa, Ethiopia, pp. 1–17 (2019)
Yau, J.W., et al.: Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care 35(3), 556–564 (2012)
Zeng, X., Chen, H., Luo, Y., Ye, W.: Automated diabetic retinopathy detection based on binocular Siamese-like convolutional neural network. IEEE Access 7(c), 30744–30753 (2019)
dc.rights.coar.fl_str_mv http://purl.org/coar/access_right/c_abf2
dc.rights.accessrights.spa.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
rights_invalid_str_mv http://purl.org/coar/access_right/c_abf2
dc.format.extent.spa.fl_str_mv 10 páginas
dc.format.mimetype.spa.fl_str_mv application/pdf
dc.publisher.spa.fl_str_mv Springer Science
dc.source.spa.fl_str_mv https://link.springer.com/chapter/10.1007/978-3-030-63419-3_21
institution Escuela Colombiana de Ingeniería Julio Garavito
bitstream.url.fl_str_mv https://repositorio.escuelaing.edu.co/bitstream/001/1425/3/Hybrid%20Deep%20Learning%20Gaussian%20Process%20for.pdf.txt
https://repositorio.escuelaing.edu.co/bitstream/001/1425/4/Hybrid%20Deep%20Learning%20Gaussian%20Process%20for.pdf.jpg
https://repositorio.escuelaing.edu.co/bitstream/001/1425/1/license.txt
https://repositorio.escuelaing.edu.co/bitstream/001/1425/2/Hybrid%20Deep%20Learning%20Gaussian%20Process%20for.pdf
bitstream.checksum.fl_str_mv a03e60b923aeb9258cab0d1a7df79b16
6def9325bf09d7abcea6e49bb12b43f0
5a7ca94c2e5326ee169f979d71d0f06e
94a9c5ad984ac88fd8c0e1906ccf5e48
bitstream.checksumAlgorithm.fl_str_mv MD5
MD5
MD5
MD5
repository.name.fl_str_mv Repositorio Escuela Colombiana de Ingeniería Julio Garavito
repository.mail.fl_str_mv repositorio.eci@escuelaing.edu.co
_version_ 1814355580761931776
spelling Gonzalez Osorio, Fabioafbb77c7b853278c83659a12e1b8dbe6600Perdomo Charry, Oscar Julianc280ba13fd48e8dbf9cdbc8179aa9c94600Toledo Cortes, Santiago599b76396f9dc557e94e41ec3b2f3331600De La Pava, Melissa176fe4dc82980a29aba79efc4179328a600GiBiome2021-05-12T19:00:05Z2021-10-01T17:16:54Z2021-05-122021-10-01T17:16:54Z20200302-9743https://repositorio.escuelaing.edu.co/handle/001/142510.1007/978-3-030-63419-3_21DOI:10.1007/978-3-030-63419-3_21La retinopatía diabética (RD) es una de las complicaciones microvasculares de la diabetes mellitus, que sigue siendo una de las principales causas de ceguera en todo el mundo. Los modelos computacionales basados ​​en redes neuronales convolucionales representan el estado del arte para la detección automática de RD utilizando imágenes de fondo de ojo. La mayor parte del trabajo actual aborda este problema como una tarea de clasificación binaria. Sin embargo, incluir la estimación de leyes y la cuantificación de la incertidumbre de las predicciones puede aumentar potencialmente la solidez del modelo. En este artículo, se presenta un método de proceso híbrido de aprendizaje profundo y gaussiano para el diagnóstico de RD y la cuantificación de la incertidumbre. Este método combina el poder de representación del aprendizaje profundo con la capacidad de generalizar a partir de pequeños conjuntos de datos de modelos de procesos gaussianos. Los resultados muestran que la cuantificación de la incertidumbre en las predicciones mejora la interpretabilidad del método como herramienta de apoyo al diagnósticoDiabetic retinopathy (DR) is one of the microvascular complications of diabetes mellitus, which remains a leading cause of blindness worldwide. Computational models based on convolutional neural networks represent the state of the art for automatic detection of DR using fundus images. Most of the current work addresses this problem as a binary classification task. However, including law estimation and quantification of prediction uncertainty can potentially increase model robustness. In this paper, a hybrid deep learning and Gaussian process method for DR diagnosis and uncertainty quantification is presented. This method combines the representational power of deep learning with the ability to generalize from small data sets of Gaussian process models. The results show that the quantification of uncertainty in the predictions improves the interpretability of the method as a diagnostic support tool. Translated with www.DeepL.com/Translator (free version)Este trabajo fue parcialmente financiado por un premio de investigación de Google y por el proyecto Colciencias número 1101-807-63563.10 páginasapplication/pdfengSpringer Sciencehttps://link.springer.com/chapter/10.1007/978-3-030-63419-3_21Hybrid Deep Learning Gaussian Process for Diabetic Retinopathy Diagnosis and Uncertainty QuantificationArtículo de revistainfo:eu-repo/semantics/publishedVersionhttp://purl.org/coar/resource_type/c_2df8fbb1Textinfo:eu-repo/semantics/articlehttp://purl.org/redcol/resource_type/ARThttp://purl.org/coar/version/c_970fb48d4fbd8a85Lecture Notes in Computer Science (LNCS, volumen 12069)20621512069N/ALecture Notes in Computer ScienceAbrámoff, M.D., et al.: Automated analysis of retinal images for detection of referable diabetic retinopathy. JAMA Ophthalmol. 131(3), 351–357 (2013).American Academy of Ophthalmology: International clinical diabetic retinopathy disease severity scale detailed table. International Council of Ophthalmology (2002)Bradshaw, J., Matthews, A.G.d.G., Ghahramani, Z.: Adversarial examples, uncertainty, and transfer testing robustness in Gaussian process hybrid deep networks, eprint, pp. 1-33 (2017).Decenciére, E., et al.: Feedback on a publicly distributed image database: the Messidor database. Image Anal. Stereol. 33(3), 231–234 (2014).Diabetic Retinopathy Detection of Kaggle: Eyepacs challenge.Ethem, A.: Introduction to Machine Learning, 3rd edn. The MIT Press, Cambridge (2014)Gargeya, R., Leng, T.: Automated identification of diabetic retinopathy using deep learning. Ophthalmology 124(7), 962–969 (2017)Gulshan, V., et al.: Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA J. Am. Med. Assoc. 316(22), 2402–2410 (2016)Kaya, M., Bilge, H.: Deep metric learning: a survey. Symmetry 11, 1066 (2019).Krause, J., et al.: Grader variability and the importance of reference standards for evaluating machine learning models for diabetic retinopathy. Ophthalmology 125(8), 1264–1272 (2018).Leibig, C., Allken, V., Ayhan, M.S., Berens, P., Wahl, S.: Leveraging uncertainty information from deep neural networks for disease detection. Sci. Rep. 7(1), 1–14 (2017)Li, F., Liu, Z., Chen, H., Jiang, M., Zhang, X., Wu, Z.: Automatic detection of diabetic retinopathy in retinal fundus photographs based on deep learning algorithm. Transl. Vis. Sci. Technol. 8(6) (2019).Lim, Z.W., Lee, M.L., Hsu, W., Wong, T.Y.: Building trust in deep learning system towards automated disease detection. In: The Thirty-First AAAI Conference on Innovative Applications of Artificial Intelligence, pp. 9516–9521 (2018)Perdomo, O., Gonzalez, F.: A systematic review of deep learning methods applied to ocular images. Ciencia e Ingenieria Neogranadina 30(1), 9–26 (2019)Raghu, M., Blumer, K., Sayres, R., Obermeyer, Z., Kleinberg, R., Mullainathan, S., Kleinberg, J.: Direct uncertainty prediction for medical second opinions. In: Proceedings of the 36th International Conference on Machine Learning, PMLR 97, Long Beach, California (2019)Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., Wojna, Z.: Rethinking the inception architecture for computer vision. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, December 2016, pp. 2818–2826 (2016).Voets, M., Møllersen, K., Bongo, L.A.: Reproduction study using public data of: Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. PLoS ONE 14(6), 1–11 (2019)Wells, J.A., et al.: aflibercept, bevacizumab, or ranibizumab for diabetic macular edema two-year results from a comparative effectiveness randomized clinical trial. Ophthalmology 123(6), 1351–1359 (2016)Wilkinson, C.P.P., et al.: Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology 110(9), 1677–1682 (2003)Wilson, A., Nickisch, H.: Kernel interpolation for scalable structured gaussian processes (KISS-GP). In: Proceedings of the 32nd International Conference on Machine Learning. JMLR: W&CP, Lille, France (2015)Xin, Q., Elliot, M., Miikkulainen, R.: Quantifying point-prediction uncertainty in neural networks via residual estimation with an I/O Kernel. In: ICLR 2020, Addis Ababa, Ethiopia, pp. 1–17 (2019)Yau, J.W., et al.: Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care 35(3), 556–564 (2012)Zeng, X., Chen, H., Luo, Y., Ye, W.: Automated diabetic retinopathy detection based on binocular Siamese-like convolutional neural network. IEEE Access 7(c), 30744–30753 (2019)info:eu-repo/semantics/openAccesshttp://purl.org/coar/access_right/c_abf2Retinopatía diabéticaAprendizajeMétodo gaussianoGaussian methodDeep LearningDiabetic RetinopathyGaussian ProcessUncertainty QuantificationAprendizaje profundoRetinopatía diabéticaProceso gaussianoCuantificación de la incertidumbreTEXTHybrid Deep Learning Gaussian Process for.pdf.txtHybrid Deep Learning Gaussian Process for.pdf.txtExtracted texttext/plain26327https://repositorio.escuelaing.edu.co/bitstream/001/1425/3/Hybrid%20Deep%20Learning%20Gaussian%20Process%20for.pdf.txta03e60b923aeb9258cab0d1a7df79b16MD53open accessTHUMBNAILHybrid Deep Learning Gaussian Process for.pdf.jpgHybrid Deep Learning Gaussian Process for.pdf.jpgGenerated Thumbnailimage/jpeg10506https://repositorio.escuelaing.edu.co/bitstream/001/1425/4/Hybrid%20Deep%20Learning%20Gaussian%20Process%20for.pdf.jpg6def9325bf09d7abcea6e49bb12b43f0MD54open accessLICENSElicense.txttext/plain1881https://repositorio.escuelaing.edu.co/bitstream/001/1425/1/license.txt5a7ca94c2e5326ee169f979d71d0f06eMD51open accessORIGINALHybrid Deep Learning Gaussian Process for.pdfapplication/pdf232426https://repositorio.escuelaing.edu.co/bitstream/001/1425/2/Hybrid%20Deep%20Learning%20Gaussian%20Process%20for.pdf94a9c5ad984ac88fd8c0e1906ccf5e48MD52open access001/1425oai:repositorio.escuelaing.edu.co:001/14252022-10-06 18:34:40.808open accessRepositorio Escuela Colombiana de Ingeniería Julio Garavitorepositorio.eci@escuelaing.edu.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