Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography

Background:Data-driven development of medical biomarkers of bone requires a large amount of image data but physical measurements are generally too restricted in size and quality to perform a robust training. Purpose: This study aims to provide a reliable in silico method for the generation of re...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Iarussi, Emmanuel, Thomsen, Felix S. L., Borggrefe, Jan, Boyd, Steven K., Wang, Yue, Battié, Michele C.
Formato: Artículo publishedVersion
Lenguaje:Inglés
Publicado: Medical Physics 2023
Materias:
Bone microstructure
Gestalt
Progressive generative adversarial network
Structural morphing
XtremeCT
Acceso en línea:https://repositorio.utdt.edu/handle/20.500.13098/11858
Aporte de:
Repositorio Digital Universidad Torcuato Di Tella de Universidad Torcuato Di Tella
id I57-R163-20.500.13098-11858
record_format dspace
spelling I57-R163-20.500.13098-118582023-06-06T07:00:16Z Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography Iarussi, Emmanuel Thomsen, Felix S. L. Borggrefe, Jan Boyd, Steven K. Wang, Yue Battié, Michele C. Bone microstructure Gestalt Progressive generative adversarial network Structural morphing XtremeCT Background:Data-driven development of medical biomarkers of bone requires a large amount of image data but physical measurements are generally too restricted in size and quality to perform a robust training. Purpose: This study aims to provide a reliable in silico method for the generation of realistic bone microstructure with defined microarchitectural properties. Synthetic bone samples may improve training of neural networks and serve for the development of new diagnostic parameters of bone architecture and mineralization. Methods: One hundred-fifty cadaveric lumbar vertebrae from 48 different male human spines were scanned with a high resolution peripheral quantitative CT. After prepocessing the scans, we extracted 10,795 purely spongeous bone patches, each with a side length of 32 voxels (5 mm) and isotropic voxel size of 164 m. We trained a volumetric generative adversarial network (GAN) in a progressive manner to create synthetic microstructural bone samples. We then added a style transfer technique to allow the generation of synthetic samples with defined microstructure and gestalt by simultaneously optimizing two entangled loss functions. Reliability testing was performed by comparing real and synthetic bone samples on 10 well-understood microstructural parameters. Results: The method was able to create synthetic bone samples with visual and quantitative properties that effectively matched with the real samples. The GAN contained a well-formed latent space allowing to smoothly morph bone samples by their microstructural parameters, visual appearance or both. Optimum performance has been obtained for bone samples with voxel size 32 × 32 × 32, but also samples of size 64 × 64 × 64 could be synthesized. Conclusions: Our two-step-approach combines a parameter-agnostic GAN with a parameter-specific style transfer technique. It allows to generate an unlimited anonymous database of microstructural bone samples with sufficient realism to be used for the development of new data-driven methods of bonebiomarkers. Particularly, the style transfer technique can generate datasets of bone samples with specific conditions to simulate certain bone pathologies. 2023-06-05T19:23:35Z 2023-06-05T19:23:35Z 2023 info:eu-repo/semantics/article info:eu-repo/semantics/publishedVersion https://repositorio.utdt.edu/handle/20.500.13098/11858 eng Medical Physics 2023;1-12. info:eu-repo/semantics/openAccess https://creativecommons.org/licenses/by-sa/2.5/ar/ pp. 1-12 application/pdf application/pdf Medical Physics American Association of Physicists in Medicine
institution Universidad Torcuato Di Tella
institution_str I-57
repository_str R-163
collection Repositorio Digital Universidad Torcuato Di Tella
language Inglés
orig_language_str_mv eng
topic Bone microstructure
Gestalt
Progressive generative adversarial network
Structural morphing
XtremeCT
spellingShingle Bone microstructure
Gestalt
Progressive generative adversarial network
Structural morphing
XtremeCT
Iarussi, Emmanuel
Thomsen, Felix S. L.
Borggrefe, Jan
Boyd, Steven K.
Wang, Yue
Battié, Michele C.
Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
topic_facet Bone microstructure
Gestalt
Progressive generative adversarial network
Structural morphing
XtremeCT
description Background:Data-driven development of medical biomarkers of bone requires a large amount of image data but physical measurements are generally too restricted in size and quality to perform a robust training. Purpose: This study aims to provide a reliable in silico method for the generation of realistic bone microstructure with defined microarchitectural properties. Synthetic bone samples may improve training of neural networks and serve for the development of new diagnostic parameters of bone architecture and mineralization. Methods: One hundred-fifty cadaveric lumbar vertebrae from 48 different male human spines were scanned with a high resolution peripheral quantitative CT. After prepocessing the scans, we extracted 10,795 purely spongeous bone patches, each with a side length of 32 voxels (5 mm) and isotropic voxel size of 164 m. We trained a volumetric generative adversarial network (GAN) in a progressive manner to create synthetic microstructural bone samples. We then added a style transfer technique to allow the generation of synthetic samples with defined microstructure and gestalt by simultaneously optimizing two entangled loss functions. Reliability testing was performed by comparing real and synthetic bone samples on 10 well-understood microstructural parameters. Results: The method was able to create synthetic bone samples with visual and quantitative properties that effectively matched with the real samples. The GAN contained a well-formed latent space allowing to smoothly morph bone samples by their microstructural parameters, visual appearance or both. Optimum performance has been obtained for bone samples with voxel size 32 × 32 × 32, but also samples of size 64 × 64 × 64 could be synthesized. Conclusions: Our two-step-approach combines a parameter-agnostic GAN with a parameter-specific style transfer technique. It allows to generate an unlimited anonymous database of microstructural bone samples with sufficient realism to be used for the development of new data-driven methods of bonebiomarkers. Particularly, the style transfer technique can generate datasets of bone samples with specific conditions to simulate certain bone pathologies.
format Artículo
publishedVersion
author Iarussi, Emmanuel
Thomsen, Felix S. L.
Borggrefe, Jan
Boyd, Steven K.
Wang, Yue
Battié, Michele C.
author_facet Iarussi, Emmanuel
Thomsen, Felix S. L.
Borggrefe, Jan
Boyd, Steven K.
Wang, Yue
Battié, Michele C.
author_sort Iarussi, Emmanuel
title Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
title_short Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
title_full Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
title_fullStr Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
title_full_unstemmed Bone-GAN: Generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
title_sort bone-gan: generation of virtual bone microstructure of high resolution peripheral quantitative computed tomography
publisher Medical Physics
publishDate 2023
url https://repositorio.utdt.edu/handle/20.500.13098/11858
work_keys_str_mv AT iarussiemmanuel bonegangenerationofvirtualbonemicrostructureofhighresolutionperipheralquantitativecomputedtomography
AT thomsenfelixsl bonegangenerationofvirtualbonemicrostructureofhighresolutionperipheralquantitativecomputedtomography
AT borggrefejan bonegangenerationofvirtualbonemicrostructureofhighresolutionperipheralquantitativecomputedtomography
AT boydstevenk bonegangenerationofvirtualbonemicrostructureofhighresolutionperipheralquantitativecomputedtomography
AT wangyue bonegangenerationofvirtualbonemicrostructureofhighresolutionperipheralquantitativecomputedtomography
AT battiemichelec bonegangenerationofvirtualbonemicrostructureofhighresolutionperipheralquantitativecomputedtomography
_version_ 1768086693067358208

Ejemplares similares