Multimodal cardiac segmentation using disentangled representations

Agisilaos Chartsias, Giorgos Papanastasiou, Chengjia Wang, Colin Stirrat, Scott Semple, David Newby, Rohan Dharmakumar, Sotirios Tsaftaris

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


Magnetic Resonance (MR) protocols use several sequences to evaluate pathology and organ status. Yet, despite recent advances, the analysis of each sequence’s images (modality hereafter) is treated in isolation. We propose a method suitable for multimodal and multi-input learning and analysis, that disentangles anatomical and imaging factors, and combines anatomical content across the modalities to extract more accurate segmentation masks. Mis-registrations between the inputs are handled with a Spatial Transformer Network, which non-linearly aligns the (now intensity-invariant) anatomical factors. We demonstrate ap-plications in Late Gadolinium Enhanced (LGE) and cine MRI segmen-tation. We show that multi-input outperforms single-input models, and that we can train a (semi-supervised) model with few (or no) annotations for one of the modalities. Code will be released upon acceptance.
Original languageEnglish
Title of host publicationSTACOM: Statistical Atlases and Computational Models of the Heart.
Publication statusAccepted/In press - 9 Aug 2019
EventMICCAI 2019 (STACOM Workshop) - Shenzen, China
Duration: 13 Oct 2019 → …


ConferenceMICCAI 2019 (STACOM Workshop)
Period13/10/19 → …
Internet address


  • Multimodal segmentation
  • factorised
  • cardiac


Dive into the research topics of 'Multimodal cardiac segmentation using disentangled representations'. Together they form a unique fingerprint.

Cite this