Cerebral vein analysis provides a chance to study, from an unusual viewpoint, an entire class of brain diseases, including neurodegenerative disorders and traumatic brain injuries. Manual segmentation approaches can be used to assess vascular anatomy, but they are observer-dependent and time-consuming; therefore, automated approaches are desirable, as they also improve reproducibility. In this paper, a new, fully automated algorithm, based on structural, morphological, and relaxometric information, is proposed to segment the entire cerebral venous system from MR images. The algorithm for multi-parametric automated segmentation of brain VEiNs (MAVEN) is based on a combined investigation of multi-parametric information that allows for rejection of false positives and detection of thin vessels. The method is tested on gradient echo brain data sets acquired at 1.5, 3, and 7 T. It is compared to previous methods against manual segmentation, and its inter-scan reproducibility is assessed. The achieved accuracy and reproducibility are good, meaning that MAVEN outperforms previous methods on both quantitative and qualitative analyses. It is usable at all the field strengths explored, showing comparable accuracy scores, with no need for algorithm parameter adjustments, and thus, it is a promising candidate for the characterization of the venous tree topology.
MAVEN: An Algorithm for Multi-Parametric Automated Segmentation of Brain Veins From Gradient Echo Acquisitions / Monti, Serena; Cocozza, Sirio; Borrelli, Pasquale; Straub, Sina; Ladd, Mark E; Salvatore, Marco; Tedeschi, Enrico; Palma, Giuseppe. - In: IEEE TRANSACTIONS ON MEDICAL IMAGING. - ISSN 0278-0062. - 36:5(2017), pp. 1054-1065. [10.1109/TMI.2016.2645286]
MAVEN: An Algorithm for Multi-Parametric Automated Segmentation of Brain Veins From Gradient Echo Acquisitions
COCOZZA, SIRIO;BORRELLI, PASQUALE;SALVATORE, MARCO;TEDESCHI, ENRICO;
2017
Abstract
Cerebral vein analysis provides a chance to study, from an unusual viewpoint, an entire class of brain diseases, including neurodegenerative disorders and traumatic brain injuries. Manual segmentation approaches can be used to assess vascular anatomy, but they are observer-dependent and time-consuming; therefore, automated approaches are desirable, as they also improve reproducibility. In this paper, a new, fully automated algorithm, based on structural, morphological, and relaxometric information, is proposed to segment the entire cerebral venous system from MR images. The algorithm for multi-parametric automated segmentation of brain VEiNs (MAVEN) is based on a combined investigation of multi-parametric information that allows for rejection of false positives and detection of thin vessels. The method is tested on gradient echo brain data sets acquired at 1.5, 3, and 7 T. It is compared to previous methods against manual segmentation, and its inter-scan reproducibility is assessed. The achieved accuracy and reproducibility are good, meaning that MAVEN outperforms previous methods on both quantitative and qualitative analyses. It is usable at all the field strengths explored, showing comparable accuracy scores, with no need for algorithm parameter adjustments, and thus, it is a promising candidate for the characterization of the venous tree topology.File | Dimensione | Formato | |
---|---|---|---|
2017 Monti IEEE-TMI (MAVEN).pdf
solo utenti autorizzati
Descrizione: articolo principale
Tipologia:
Documento in Post-print
Licenza:
Accesso privato/ristretto
Dimensione
3.22 MB
Formato
Adobe PDF
|
3.22 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.