Multi-task Fully Convolutional Network for Brain Tumour Segmentation

Haocheng Shen (Lead / Corresponding author), Ruixuan Wang, Jianguo Zhang, Stephen McKenna

Research output: Chapter in Book/Report/Conference proceedingConference contribution

20 Citations (Scopus)


In this paper, a novel, multi-task fully convolutional network (FCN) architecture is proposed for automatic segmentation of brain tumour. The proposed network builds on the hierarchical relationship between tumour substructures with branch and leaf losses imposed and optimised simultaneously. The network takes multimodal MR images along with their symmetric-difference images as input and extracts multi-level contextual information, firstly by the branch losses which are then fed to the leaf loss in a combination stage. The model was evaluated on BRATS13 and BRATS15 datasets and results show that the proposed multi-task FCN outperforms single-task FCN on all sub-tasks. The method is among the most accurate available and its computational cost is relatively low at test time.
Original languageEnglish
Title of host publicationMedical Image Understanding and Analysis
Subtitle of host publication21st Annual Conference, MIUA 2017, Edinburgh, UK, July 11–13, 2017, Proceedings
EditorsMaria Valdes Hernandez, Victor Gonzalez-Castro
Place of PublicationSwitzerland
Number of pages10
ISBN (Electronic)9783319609645
ISBN (Print)9783319609638
Publication statusPublished - 2017
EventMedical Image Understanding and Analysis (MIUA) 2017 - John McIntyre Centre, Pollock Halls, Edinburgh, United Kingdom
Duration: 11 Jul 201713 Jul 2017

Publication series

NameCommunications in Computer and Information Science
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937


ConferenceMedical Image Understanding and Analysis (MIUA) 2017
Abbreviated titleMIUA 2017
Country/TerritoryUnited Kingdom
Internet address


  • Deep learning
  • Tumour segmentation
  • Multi-task learning


Dive into the research topics of 'Multi-task Fully Convolutional Network for Brain Tumour Segmentation'. Together they form a unique fingerprint.

Cite this