TY - GEN
T1 - Pseudo-healthy image synthesis for white matter lesion segmentation
AU - Bowles, Christopher
AU - Qin, Chen
AU - Ledig, Christian
AU - Guerrero, Ricardo
AU - Gunn, Roger
AU - Hammers, Alexander
AU - Sakka, Eleni
AU - Dickie, David Alexander
AU - Hernández, Maria Valdés
AU - Royle, Natalie
AU - Wardlaw, Joanna
AU - Rhodius-Meester, Hanneke
AU - Tijms, Betty
AU - Lemstra, Afina W.
AU - van Der Flier, Wiesje
AU - Barkhof, Frederik
AU - Scheltens, Philip
AU - Rueckert, Daniel
N1 - Publisher Copyright:
© Springer International Publishing AG 2016.
PY - 2016
Y1 - 2016
N2 - White matter hyperintensities (WMH) seen on FLAIR images are established as a key indicator of Vascular Dementia (VD) and other pathologies.We propose a novel modality transformation technique to generate a subject-specific pathology-free synthetic FLAIR image from a T1 -weighted image. WMH are then accurately segmented by comparing this synthesized FLAIR image to the actually acquired FLAIR image. We term this method Pseudo-Healthy Image Synthesis (PHI-Syn). The method is evaluated on data from 42 stroke patients where we compare its performance to two commonly used methods from the Lesion Segmentation Toolbox. We show that the proposed method achieves superior performance for a number of metrics. Finally, we show that the features extracted from the WMH segmentations can be used to predict a Fazekas lesion score that supports the identification of VD in a dataset of 468 dementia patients. In this application the automatically calculated features perform comparably to clinically derived Fazekas scores.
AB - White matter hyperintensities (WMH) seen on FLAIR images are established as a key indicator of Vascular Dementia (VD) and other pathologies.We propose a novel modality transformation technique to generate a subject-specific pathology-free synthetic FLAIR image from a T1 -weighted image. WMH are then accurately segmented by comparing this synthesized FLAIR image to the actually acquired FLAIR image. We term this method Pseudo-Healthy Image Synthesis (PHI-Syn). The method is evaluated on data from 42 stroke patients where we compare its performance to two commonly used methods from the Lesion Segmentation Toolbox. We show that the proposed method achieves superior performance for a number of metrics. Finally, we show that the features extracted from the WMH segmentations can be used to predict a Fazekas lesion score that supports the identification of VD in a dataset of 468 dementia patients. In this application the automatically calculated features perform comparably to clinically derived Fazekas scores.
UR - http://www.scopus.com/inward/record.url?scp=84994074794&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-46630-9_9
DO - 10.1007/978-3-319-46630-9_9
M3 - Conference contribution
AN - SCOPUS:84994074794
SN - 9783319466293
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 87
EP - 96
BT - Simulation and Synthesis in Medical Imaging - 1st International Workshop, SASHIMI 2016 held in conjunction with MICCAI 2016, Proceedings
A2 - Tsaftaris, Sotirios A.
A2 - Gooya, Ali
A2 - Frangi, Alejandro F.
A2 - Prince, Jerry L.
PB - Springer Verlag
T2 - 1st International Workshop on Simulation and Synthesis in Medical Imaging, SASHIMI 2016 held in conjunction with 19th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016
Y2 - 21 October 2016 through 21 October 2016
ER -