Liver tumour segmentation from MR-Linac

Lead: Maryam Fallahpoor

active 2026–ongoing NHMRC

Background

Accurate liver tumour delineation is essential for MR-guided radiotherapy, yet it remains challenging due to low soft-tissue contrast variability, respiratory motion artefacts, and anatomical deformation between fractions. The MR-Linac provides on-treatment MRI at every fraction, offering an opportunity to improve contouring accuracy and support adaptive workflows — but the volume and complexity of this data demand automated solutions.

Aims

  • Develop robust deep learning–based segmentation models for liver tumours in MR-Linac imaging data.
  • Support adaptive radiotherapy workflows through reliable tumour localisation and contouring in real-time or near real-time clinical environments.
  • Improve the precision of radiation treatment delivery by enabling accurate fraction-by-fraction tumour tracking.

Methods

Outcomes