GP-Unet: Lesion detection from weak labels with a 3D regression network

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskningfagfællebedømt

  • Florian Dubost
  • Gerda Bortsova
  • Hieab Adams
  • Arfan Ikram
  • Wiro J. Niessen
  • Meike Vernooij
  • de Bruijne, Marleen

We propose a novel convolutional neural network for lesion detection from weak labels. Only a single, global label per image - the lesion count - is needed for training. We train a regression network with a fully convolutional architecture combined with a global pooling layer to aggregate the 3D output into a scalar indicating the lesion count. When testing on unseen images, we first run the network to estimate the number of lesions. Then we remove the global pooling layer to compute localization maps of the size of the input image. We evaluate the proposed network on the detection of enlarged perivascular spaces in the basal ganglia in MRI. Our method achieves a sensitivity of 62% with on average 1.5 false positives per image. Compared with four other approaches based on intensity thresholding, saliency and class maps, our method has a 20% higher sensitivity.

OriginalsprogEngelsk
TitelMedical Image Computing and Computer-Assisted Intervention − MICCAI 2017 : 20th International Conference, Quebec City, QC, Canada, September 11-13, 2017, Proceedings, Part III
RedaktørerMaxime Descoteaux, Lena Maier-Hein, Alfred Franz, Pierre Jannin, D. Louis Collins, Simon Duschesne
Antal sider8
ForlagSpringer
Publikationsdato2017
Sider214-221
ISBN (Trykt)978-3-319-66178-0
ISBN (Elektronisk)978-3-319-66179-7
DOI
StatusUdgivet - 2017
Begivenhed20th International Conference on Medical Image Computing and Computer-Assisted Intervention - Quebec City, Canada
Varighed: 11 sep. 201713 sep. 2017
Konferencens nummer: 20

Konference

Konference20th International Conference on Medical Image Computing and Computer-Assisted Intervention
Nummer20
LandCanada
ByQuebec City
Periode11/09/201713/09/2017
NavnLecture notes in computer science
Vol/bind10435
ISSN0302-9743

Links

ID: 184141962