Towards Patient-Individual PI-RADS v2 Sector Map: CNN for Automatic Segmentation of Prostatic Zones from T2-Weighted MRI

Meyer, Anneke; Rak, Marko; Schindele, Daniel; Blaschke, Simon; Schostak, Martin; Fedorov, Andrey; Hansen, Christian

by Anneke Meyer, Marko Rak, Daniel Schindele, Simon Blaschke, Martin Schostak, Andrey Fedorov, Christian Hansen

Abstract:

Automatic segmentation of the prostate, its inner and surrounding structures is highly desired for various applications. Several works have been presented for segmentation of anatomical zones of the prostate that are limited to the transition and peripheral zone. Following the spatial division according to the PI-RADS v2 sector map, we present a multi-class segmentation method that additionally targets the anterior fibromuscular stroma and distal prostatic urethra to improve computer-aided detection methods and enable a more precise therapy planning. We propose a multi-class segmentation with an anisotropic convolutional neural network that generates a topo-logically correct division of the prostate into these four structures. We evaluated our method on a dataset of T2-weighted axial MRI scans (n=98 subjects) and obtained results in the range of inter-rater variability for the majority of the zones.

Reference:

Towards Patient-Individual PI-RADS v2 Sector Map: CNN for Automatic Segmentation of Prostatic Zones from T2-Weighted MRI (Anneke Meyer, Marko Rak, Daniel Schindele, Simon Blaschke, Martin Schostak, Andrey Fedorov, Christian Hansen), In IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), 2019.

Bibtex Entry:

@inproceedings{meyer_towards_2019,
	address = {Venice, Italy},
	title = {Towards {Patient}-{Individual} {PI}-{RADS} v2 {Sector} {Map}: {CNN} for {Automatic} {Segmentation} of {Prostatic} {Zones} from {T}2-{Weighted} {MRI}},
	abstract = {Automatic segmentation of the prostate, its inner and surrounding structures is highly desired for various applications. Several works have been presented for segmentation of anatomical zones of the prostate that are limited to the transition and peripheral zone. Following the spatial division according to the PI-RADS v2 sector map, we present a multi-class segmentation method that additionally targets the anterior fibromuscular stroma and distal prostatic urethra to improve computer-aided detection methods and enable a more precise therapy planning. We propose a multi-class segmentation with an anisotropic convolutional neural network that generates a topo-logically correct division of the prostate into these four structures. We evaluated our method on a dataset of T2-weighted axial MRI scans (n=98 subjects) and obtained results in the range of inter-rater variability for the majority of the zones.},
	booktitle = {{IEEE} 16th {International} {Symposium} on {Biomedical} {Imaging} ({ISBI} 2019)},
	author = {Meyer, Anneke and Rak, Marko and Schindele, Daniel and Blaschke, Simon and Schostak, Martin and Fedorov, Andrey and Hansen, Christian},
	month = jan,
	year = {2019},
	pages = {696--700}
}