One of the most common tasks in small rodents MRI pipelines is the voxel-wise segmentation of the volume in multiple classes. While many segmentation schemes have been developed for the human brain, fewer are available for rodent MRI, often by adaptation from human neuroimaging. Common methods include atlas-based and clustering schemes. The former labels the target volume by registering one or more pre-labeled atlases using a deformable registration method, in which case the result depends on the quality of the reference volumes, the registration algorithm and the label fusion approach, if more than one atlas is employed. The latter is based on an expectation maximization procedure to maximize the variance between voxel categories, and is often combined with Markov Random Fields and the atlas based approach to include spatial information, priors, and improve the classification accuracy. Our primary goal is to critically review the state of the art of rat and mouse segmentation of neuro MRI volumes and compare the available literature on popular, readily and freely available MRI toolsets, including SPM, FSL and ANTs, when applied to this task in the context of common pre-processing steps. Furthermore, we will briefly address the emerging Deep Learning methods for the segmentation of medical imaging, and the perspectives for applications to small rodents.

Towards an efficient segmentation of small rodents brain: a short critical review / Feo, R.; Giove, F.. - In: JOURNAL OF NEUROSCIENCE METHODS. - ISSN 0165-0270. - 323:(2019), pp. 82-89. [10.1016/j.jneumeth.2019.05.003]

Towards an efficient segmentation of small rodents brain: a short critical review

Feo R.;Giove F.
2019

Abstract

One of the most common tasks in small rodents MRI pipelines is the voxel-wise segmentation of the volume in multiple classes. While many segmentation schemes have been developed for the human brain, fewer are available for rodent MRI, often by adaptation from human neuroimaging. Common methods include atlas-based and clustering schemes. The former labels the target volume by registering one or more pre-labeled atlases using a deformable registration method, in which case the result depends on the quality of the reference volumes, the registration algorithm and the label fusion approach, if more than one atlas is employed. The latter is based on an expectation maximization procedure to maximize the variance between voxel categories, and is often combined with Markov Random Fields and the atlas based approach to include spatial information, priors, and improve the classification accuracy. Our primary goal is to critically review the state of the art of rat and mouse segmentation of neuro MRI volumes and compare the available literature on popular, readily and freely available MRI toolsets, including SPM, FSL and ANTs, when applied to this task in the context of common pre-processing steps. Furthermore, we will briefly address the emerging Deep Learning methods for the segmentation of medical imaging, and the perspectives for applications to small rodents.
2019
brain; mouse; rat; segmentation
01 Pubblicazione su rivista::01g Articolo di rassegna (Review)
Towards an efficient segmentation of small rodents brain: a short critical review / Feo, R.; Giove, F.. - In: JOURNAL OF NEUROSCIENCE METHODS. - ISSN 0165-0270. - 323:(2019), pp. 82-89. [10.1016/j.jneumeth.2019.05.003]
File allegati a questo prodotto
File Dimensione Formato  
Feo_Towards_2019.pdf

accesso aperto

Tipologia: Documento in Post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza: Creative commons
Dimensione 460.4 kB
Formato Adobe PDF
460.4 kB Adobe PDF

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1338621
Citazioni
  • ???jsp.display-item.citation.pmc??? 8
  • Scopus 28
  • ???jsp.display-item.citation.isi??? 27
social impact