EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DeepComBat : A statistically motivated, hyperparameter‐robust, deep learning approach to harmonization of neuroimaging data

Fengling Hu iD, Alfredo Lucas iD, Andrew A Chen, Kyle Coleman, Hannah Horng, Raymond W S Ng, Nicholas J Tustison iD, Kathryn A Davis, Haochang Shou, Mingyao Li, Russell T Shinohara, Alzheimer's Disease Neuroimaging Initiative

Human Brain Mapping · 2024-08 · 45 (11)

DOI: 10.1002/hbm.26708 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Human Brain Mapping, online erschienen am 26.07.2024. Verfasst von Fengling Hu und 11 weiteren Autor:innen. Beteiligte Einrichtung: Penn Statistics in Imaging and Visualization Endeavor (PennSIVE), Department of Biostatistics, Epidemiology, and Informatics, Perelman School of Medi…. Hauptthemen laut MEDLINE: Deep Learning, Neuroimaging, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 23-mal zitiert (Crossref) und nennt 60 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NIA NIH HHS, NINDS NIH HHS, University of Pennsylvania.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Neuroimaging data acquired using multiple scanners or protocols are increasingly available. However, such data exhibit technical artifacts across batches which introduce confounding and decrease reproducibility. This is especially true when multi‐batch data are analyzed using complex downstream models which are more likely to pick up on and implicitly incorporate batch‐related information. Previously proposed image harmonization methods have sought to remove these batch effects; however, batch effects remain detectable in the data after applying these methods. We present DeepComBat, a deep learning harmonization method based on a conditional variational autoencoder and the ComBat method. DeepComBat combines the strengths of statistical and deep learning methods in order to account for the multivariate relationships between features while simultaneously relaxing strong assumptions made by previous deep learning harmonization methods. As a result, DeepComBat can perform multivariate harmonization while preserving data structure and avoiding the introduction of synthetic artifacts. We apply this method to cortical thickness measurements from a cognitive‐aging cohort and show DeepComBat qualitatively and quantitatively outperforms existing methods in removing batch effects while preserving biological heterogeneity. Additionally, DeepComBat provides a new perspective for statistically motivated deep learning harmonization methods.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningNeuroimagingImage Processing, Computer-AssistedHumansMagnetic Resonance ImagingCerebral CortexAgedMaleFemale

Stichwörter der Autor:innen

deep learningimage harmonizationneuroimagingreproducibility

Autor:innen und Institutionen

  • Penn Statistics in Imaging and Visualization Endeavor (PennSIVE), Department of Biostatistics, Epidemiology, and Informatics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
  • Center for Neuroengineering and Therapeutics, Department of Engineering, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
  • Statistical Center for Single-Cell and Spatial Genomics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
  • Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
  • Department of Radiology and Medical Imaging, University of Virginia, Charlottesville, Virginia, USA.
  • Department of Neurology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

ORCID-Profile: Fengling Hu, Alfredo Lucas, Nicholas J Tustison

Förderung

  • NIA NIH HHS
  • NINDS NIH HHS
  • University of Pennsylvania
  • National Multiple Sclerosis Society
  • NIBIB NIH HHS
  • NIH HHS
  • U.S. Department of Defense
  • National Institute of Biomedical Imaging and Bioengineering

Zitieren

APA

Hu, F., Lucas, A., Chen, A. A., Coleman, K., Horng, H., Ng, R. W. S., Tustison, N. J., Davis, K. A., Shou, H., Li, M., Shinohara, R. T., & Alzheimer's Disease Neuroimaging Initiative (2024). DeepComBat : A statistically motivated, hyperparameter‐robust, deep learning approach to harmonization of neuroimaging data. Human Brain Mapping, 45(11). https://doi.org/10.1002/hbm.26708

Vancouver

Hu F, Lucas A, Chen AA, Coleman K, Horng H, Ng RWS, et al. DeepComBat : A statistically motivated, hyperparameter‐robust, deep learning approach to harmonization of neuroimaging data. Human Brain Mapping. 2024;45(11). doi:10.1002/hbm.26708

BibTeX
@article{Hu2024,
  author = {Hu, Fengling and Lucas, Alfredo and Chen, Andrew A and Coleman, Kyle and Horng, Hannah and Ng, Raymond W S and Tustison, Nicholas J and Davis, Kathryn A and Shou, Haochang and Li, Mingyao and Shinohara, Russell T and Alzheimer's Disease Neuroimaging Initiative, Alzheimer's Disease Neuroimaging Initiative},
  title = {DeepComBat : A statistically motivated, hyperparameter‐robust, deep learning approach to harmonization of neuroimaging data},
  journal = {Human Brain Mapping},
  year = {2024},
  volume = {45},
  number = {11},
  doi = {10.1002/hbm.26708}
}
Kostenlos starten →