notesum.ai

Published at November 13

MLV$^2$-Net: Rater-Based Majority-Label Voting for Consistent Meningeal Lymphatic Vessel Segmentation

cs.CV

cs.AI

cs.LG

Released Date: November 13, 2024

Authors: Fabian Bongratz¹, Markus Karmann¹, Adrian Holz¹, Moritz Bonhoeffer¹, Viktor Neumaier¹, Sarah Deli, Benita Schmitz-Koep¹, Claus Zimmer¹, Christian Sorg, Melissa Thalhammer¹, Dennis M Hedderich, Christian Wachinger¹

Aff.: ¹Technical University of Munich, Munich, Germany

Arxiv: http://arxiv.org/abs/2411.08537v1

		Cross-validation			Consensus set
Method	Variant	Anterior	Middle	Posterior	Foreground
MLV²-Net	$w_{\text{fg}}=3$	$\underline{0.626}\pm 0.108$	$\underline{0.709}\pm 0.092$	$\underline{0.687}\pm 0.120$	$\textbf{0.806}\pm 0.030$
MLV²-Net	Oracle	$\textbf{0.638}\pm 0.151$	$\textbf{0.712}\pm 0.094$	$\textbf{0.688}\pm 0.122$	-
nnU-Net	3D Fullres	$0.593\pm 0.117$	$0.689\pm 0.098$	$0.682\pm 0.116$	$\underline{0.787}\pm 0.046$
nnU-Net	2D	$0.618\pm 0.106$	$0.707\pm 0.099$	$0.683\pm 0.109$	$0.760\pm 0.024$
UniverSeg	All planes	$0.341\pm 0.106$	$0.459\pm 0.159$	$0.490\pm 0.114$	$0.529\pm 0.177$
UniverSeg	Coronal	$0.262\pm 0.090$	$0.437\pm 0.135$	$0.401\pm 0.100$	$0.498\pm 0.130$
UniverSeg	Sagittal	$0.312\pm 0.113$	$0.411\pm 0.148$	$0.413\pm 0.124$	$0.427\pm 0.187$
UniverSeg	Transverse	$0.275\pm 0.102$	$0.318\pm 0.128$	$0.366\pm 0.107$	$0.417\pm 0.113$
SegProp	Optimized	$0.350\pm 0.137$	$0.445\pm 0.104$	$0.396\pm 0.110$	$0.493\pm 0.079$
SegProp	Standard	$0.200\pm 0.078$	$0.291\pm 0.072$	$0.228\pm 0.080$	$0.294\pm 0.065$