ASCHOPLEX: A generalizable approach for the automatic segmentation of choroid plexus

Valentina Visani; Mattia Veronese; Francesca B. Pizzini; Annalisa Colombi; Valerio Natale; Corina Marjin; Agnese Tamanti; Julia J. Schubert; Noha Althubaity; Inés Bedmar-Gómez; Neil A. Harrison; Edward T. Bullmore; Federico E. Turkheimer; Massimiliano Calabrese; Marco Castellaro

doi:10.1016/j.compbiomed.2024.109164

ASCHOPLEX: A generalizable approach for the automatic segmentation of choroid plexus

Valentina Visani, Mattia Veronese, Francesca B. Pizzini, Annalisa Colombi, Valerio Natale, Corina Marjin, Agnese Tamanti, Julia J. Schubert, Noha Althubaity, Inés Bedmar-Gómez, Neil A. Harrison, Edward T. Bullmore, Federico E. Turkheimer, Massimiliano Calabrese, Marco Castellaro^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

Abstract

Background: The Choroid Plexus (ChP) plays a vital role in brain homeostasis, serving as part of the Blood-Cerebrospinal Fluid Barrier, contributing to brain clearance pathways and being the main source of cerebrospinal fluid. Since the involvement of ChP in neurological and psychiatric disorders is not entirely established and currently under investigation, accurate and reproducible segmentation of this brain structure on large cohorts remains challenging. This paper presents ASCHOPLEX, a deep-learning tool for the automated segmentation of human ChP from structural MRI data that integrates existing software architectures like 3D UNet, UNETR, and DynUNet to deliver accurate ChP volume estimates. Methods: Here we trained ASCHOPLEX on 128 T1-w MRI images comprising both controls and patients with Multiple Sclerosis. ASCHOPLEX's performances were evaluated using traditional segmentation metrics; manual segmentation by experts served as ground truth. To overcome the generalizability problem that affects data-driven approaches, an additional fine-tuning procedure (ASCHOPLEX_tune) was implemented on 77 T1-w PET/MRI images of both controls and depressed patients. Results: ASCHOPLEX showed superior performance compared to commonly used methods like FreeSurfer and Gaussian Mixture Model both in terms of Dice Coefficient (ASCHOPLEX 0.80, ASCHOPLEX_tune 0.78) and estimated ChP volume error (ASCHOPLEX 9.22%, ASCHOPLEX_tune 9.23%). Conclusion: These results highlight the high accuracy, reliability, and reproducibility of ASCHOPLEX ChP segmentations.

Original language	English
Article number	109164
Journal	Computers in Biology and Medicine
Volume	182
Early online date	25 Sept 2024
DOIs	https://doi.org/10.1016/j.compbiomed.2024.109164
Publication status	Published - Nov 2024

Keywords

Choroid plexus
Deep neural networks
Ensemble
Magnetic resonance imaging
Semantic segmentation

Access to Document

10.1016/j.compbiomed.2024.109164Licence: CC BY

Cite this

Visani, V., Veronese, M., Pizzini, F. B., Colombi, A., Natale, V., Marjin, C., Tamanti, A., Schubert, J. J., Althubaity, N., Bedmar-Gómez, I., Harrison, N. A., Bullmore, E. T., Turkheimer, F. E., Calabrese, M., & Castellaro, M. (2024). ASCHOPLEX: A generalizable approach for the automatic segmentation of choroid plexus. Computers in Biology and Medicine, 182, Article 109164. https://doi.org/10.1016/j.compbiomed.2024.109164

@article{c57379d0c5f9466e8db812449b5d4fe1,

title = "ASCHOPLEX: A generalizable approach for the automatic segmentation of choroid plexus",

abstract = "Background: The Choroid Plexus (ChP) plays a vital role in brain homeostasis, serving as part of the Blood-Cerebrospinal Fluid Barrier, contributing to brain clearance pathways and being the main source of cerebrospinal fluid. Since the involvement of ChP in neurological and psychiatric disorders is not entirely established and currently under investigation, accurate and reproducible segmentation of this brain structure on large cohorts remains challenging. This paper presents ASCHOPLEX, a deep-learning tool for the automated segmentation of human ChP from structural MRI data that integrates existing software architectures like 3D UNet, UNETR, and DynUNet to deliver accurate ChP volume estimates. Methods: Here we trained ASCHOPLEX on 128 T1-w MRI images comprising both controls and patients with Multiple Sclerosis. ASCHOPLEX's performances were evaluated using traditional segmentation metrics; manual segmentation by experts served as ground truth. To overcome the generalizability problem that affects data-driven approaches, an additional fine-tuning procedure (ASCHOPLEXtune) was implemented on 77 T1-w PET/MRI images of both controls and depressed patients. Results: ASCHOPLEX showed superior performance compared to commonly used methods like FreeSurfer and Gaussian Mixture Model both in terms of Dice Coefficient (ASCHOPLEX 0.80, ASCHOPLEXtune 0.78) and estimated ChP volume error (ASCHOPLEX 9.22%, ASCHOPLEXtune 9.23%). Conclusion: These results highlight the high accuracy, reliability, and reproducibility of ASCHOPLEX ChP segmentations.",

keywords = "Choroid plexus, Deep neural networks, Ensemble, Magnetic resonance imaging, Semantic segmentation",

author = "Valentina Visani and Mattia Veronese and Pizzini, {Francesca B.} and Annalisa Colombi and Valerio Natale and Corina Marjin and Agnese Tamanti and Schubert, {Julia J.} and Noha Althubaity and In{\'e}s Bedmar-G{\'o}mez and Harrison, {Neil A.} and Bullmore, {Edward T.} and Turkheimer, {Federico E.} and Massimiliano Calabrese and Marco Castellaro",

note = "Publisher Copyright: {\textcopyright} 2024 The Authors",

year = "2024",

month = nov,

doi = "10.1016/j.compbiomed.2024.109164",

language = "English",

volume = "182",

journal = "Computers in Biology and Medicine",

issn = "0010-4825",

publisher = "Elsevier Limited",

}

Visani, V, Veronese, M, Pizzini, FB, Colombi, A, Natale, V, Marjin, C, Tamanti, A, Schubert, JJ , Althubaity, N, Bedmar-Gómez, I, Harrison, NA, Bullmore, ET, Turkheimer, FE, Calabrese, M & Castellaro, M 2024, 'ASCHOPLEX: A generalizable approach for the automatic segmentation of choroid plexus', Computers in Biology and Medicine, vol. 182, 109164. https://doi.org/10.1016/j.compbiomed.2024.109164

TY - JOUR

T1 - ASCHOPLEX

T2 - A generalizable approach for the automatic segmentation of choroid plexus

AU - Visani, Valentina

AU - Veronese, Mattia

AU - Pizzini, Francesca B.

AU - Colombi, Annalisa

AU - Natale, Valerio

AU - Marjin, Corina

AU - Tamanti, Agnese

AU - Schubert, Julia J.

AU - Althubaity, Noha

AU - Bedmar-Gómez, Inés

AU - Harrison, Neil A.

AU - Bullmore, Edward T.

AU - Turkheimer, Federico E.

AU - Calabrese, Massimiliano

AU - Castellaro, Marco

PY - 2024/11

Y1 - 2024/11

N2 - Background: The Choroid Plexus (ChP) plays a vital role in brain homeostasis, serving as part of the Blood-Cerebrospinal Fluid Barrier, contributing to brain clearance pathways and being the main source of cerebrospinal fluid. Since the involvement of ChP in neurological and psychiatric disorders is not entirely established and currently under investigation, accurate and reproducible segmentation of this brain structure on large cohorts remains challenging. This paper presents ASCHOPLEX, a deep-learning tool for the automated segmentation of human ChP from structural MRI data that integrates existing software architectures like 3D UNet, UNETR, and DynUNet to deliver accurate ChP volume estimates. Methods: Here we trained ASCHOPLEX on 128 T1-w MRI images comprising both controls and patients with Multiple Sclerosis. ASCHOPLEX's performances were evaluated using traditional segmentation metrics; manual segmentation by experts served as ground truth. To overcome the generalizability problem that affects data-driven approaches, an additional fine-tuning procedure (ASCHOPLEXtune) was implemented on 77 T1-w PET/MRI images of both controls and depressed patients. Results: ASCHOPLEX showed superior performance compared to commonly used methods like FreeSurfer and Gaussian Mixture Model both in terms of Dice Coefficient (ASCHOPLEX 0.80, ASCHOPLEXtune 0.78) and estimated ChP volume error (ASCHOPLEX 9.22%, ASCHOPLEXtune 9.23%). Conclusion: These results highlight the high accuracy, reliability, and reproducibility of ASCHOPLEX ChP segmentations.

AB - Background: The Choroid Plexus (ChP) plays a vital role in brain homeostasis, serving as part of the Blood-Cerebrospinal Fluid Barrier, contributing to brain clearance pathways and being the main source of cerebrospinal fluid. Since the involvement of ChP in neurological and psychiatric disorders is not entirely established and currently under investigation, accurate and reproducible segmentation of this brain structure on large cohorts remains challenging. This paper presents ASCHOPLEX, a deep-learning tool for the automated segmentation of human ChP from structural MRI data that integrates existing software architectures like 3D UNet, UNETR, and DynUNet to deliver accurate ChP volume estimates. Methods: Here we trained ASCHOPLEX on 128 T1-w MRI images comprising both controls and patients with Multiple Sclerosis. ASCHOPLEX's performances were evaluated using traditional segmentation metrics; manual segmentation by experts served as ground truth. To overcome the generalizability problem that affects data-driven approaches, an additional fine-tuning procedure (ASCHOPLEXtune) was implemented on 77 T1-w PET/MRI images of both controls and depressed patients. Results: ASCHOPLEX showed superior performance compared to commonly used methods like FreeSurfer and Gaussian Mixture Model both in terms of Dice Coefficient (ASCHOPLEX 0.80, ASCHOPLEXtune 0.78) and estimated ChP volume error (ASCHOPLEX 9.22%, ASCHOPLEXtune 9.23%). Conclusion: These results highlight the high accuracy, reliability, and reproducibility of ASCHOPLEX ChP segmentations.

KW - Choroid plexus

KW - Deep neural networks

KW - Ensemble

KW - Magnetic resonance imaging

KW - Semantic segmentation

UR - http://www.scopus.com/inward/record.url?scp=85204771997&partnerID=8YFLogxK

U2 - 10.1016/j.compbiomed.2024.109164

DO - 10.1016/j.compbiomed.2024.109164

M3 - Article

C2 - 39326265

AN - SCOPUS:85204771997

SN - 0010-4825

VL - 182

JO - Computers in Biology and Medicine

JF - Computers in Biology and Medicine

M1 - 109164

ER -

ASCHOPLEX: A generalizable approach for the automatic segmentation of choroid plexus

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this