EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

ViTMARE – A Vision Transformer Pipeline for Anomaly Detection in 3D Brain MRI

Lorenzo Peracchio, Lorenzo Corso, Gabriele Santangelo, Sithin Thulasi Seetha, Chandra Bortolotto, Arianna Dagliati, Riccardo Bellazzi, Giovanna Nicora

Studies in Health Technology and Informatics · 21.05.2026

DOI: 10.3233/shti260128 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Studies in Health Technology and Informatics, erschienen am 21.05.2026. Verfasst von Lorenzo Peracchio und 7 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer, Electrical and Biomedical Engineering, University of Pavia, Pavia, Italy. Hauptthemen laut MEDLINE: Imaging, Three-Dimensional, Magnetic Resonance Imaging, Brain, Image Interpretation, Computer-Assisted. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

AI models for medical imaging often fail under dataset shifts and on underrepresented patient subgroups. Detecting out-of-distribution scans—arising from rare pathologies, atypical anatomy, or acquisition artifacts—is therefore essential for robust deployment. We introduce ViTMARE (Vision Transformer Masked Autoencoder Reconstruction Error), a volumetric anomaly-detection pipeline for 3D brain MRI that leverages Vision Transformer Masked AutoEncoders (ViTMAEs) adapted to volumetric data by treating axial slices as input channels. The model is fine-tuned on normal brain volumes and evaluated using a synthetic-lesion generator that produces anatomically plausible abnormalities. During inference, ViTMARE performs multiple reconstructions (N=100) and aggregates binary anomaly masks via majority voting, followed by morphological closing and opening to suppress spurious noise. On a test set of real images with added synthetic anomalies, ViTMARE achieves a median Dice score of 0.793, a median precision of 0.912, and a median recall of 0.748. We present a reproducible pipeline and demonstrate that combining voting-based fusion with morphological postprocessing yields robust voxel-level anomaly detection.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Imaging, Three-DimensionalMagnetic Resonance ImagingBrainImage Interpretation, Computer-AssistedHumansAutoencoderAlgorithms

Stichwörter der Autor:innen

Medical ImagingSynthetic LesionsVolumetric Segmentation

Autor:innen und Institutionen

  • Department of Computer, Electrical and Biomedical Engineering, University of Pavia, Pavia, Italy.
  • Research Unit of Computer Systems and Bioinformatics, Department of Engineering, Università Campus Bio-Medico di Roma, Rome, Italy.
  • Clinical Department, National Center for Oncological Hadrontherapy (CNAO), Pavia, Italy.
  • Radiology Department, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy.

Zitieren

APA

Peracchio, L., Corso, L., Santangelo, G., Seetha, S. T., Bortolotto, C., Dagliati, A., Bellazzi, R., & Nicora, G. (2026). ViTMARE – A Vision Transformer Pipeline for Anomaly Detection in 3D Brain MRI. Studies in Health Technology and Informatics. https://doi.org/10.3233/shti260128

Vancouver

Peracchio L, Corso L, Santangelo G, Seetha ST, Bortolotto C, Dagliati A, et al. ViTMARE – A Vision Transformer Pipeline for Anomaly Detection in 3D Brain MRI. Studies in Health Technology and Informatics. 2026. doi:10.3233/shti260128

BibTeX
@article{Peracchio2026,
  author = {Peracchio, Lorenzo and Corso, Lorenzo and Santangelo, Gabriele and Seetha, Sithin Thulasi and Bortolotto, Chandra and Dagliati, Arianna and Bellazzi, Riccardo and Nicora, Giovanna},
  title = {ViTMARE – A Vision Transformer Pipeline for Anomaly Detection in 3D Brain MRI},
  journal = {Studies in Health Technology and Informatics},
  year = {2026},
  doi = {10.3233/shti260128}
}
Kostenlos starten →