Combining machine learning and artery characterization to identify the main bifurcations in 3D vascular trees

Automatic Classification of the Cerebral Vascular Bifurcations Using Dimensionality Reduction and Machine Learning

“…In this work, we present a new automated method for identifying major bifurcations of interest in the CoW within 3D human brain acquisitions. As previously mentioned, this study extends our previous work [11], where we labeled major bifurcations in both artificial images (vascusynth [16]) as well as in mouse brain acquisitions. Using a set of 30 vascuSynth synthetic images, we achieved a 98% recognition rate for the 14 main bifurcations using a Linear Discriminant Analysis (LDA).…”

“…In this paper, we present a new automatic method to identify the main BoI on 3D TOF-MRA human brain acquisitions. This study follows a previous work which was focused on the automatic labeling of the main bifurcations constituting the CoW in synthetic and mouse vasculatures [11]. Our suggested method is designed to satisfy the following main expectations: 1) experiment the method on complex images; 2) propose a more robust labeling method, and 3) demonstrate how a dimensionality reduction method can replace the registration step.…”