projects

Using AI & MRIs to diagnose Brain Tumours

January 2018

The period from referral to test result can take up to 10% of the average survival time. 4,753 people were waiting longer than 6 weeks for MRI scan results1 and Glioblastoma, the most common high-grade brain tumor has an average survival time 12-18 months 2

A Radiologist’s view of Brain Tumors

Brain Tumors are uncontrolled, unnatural growth, and division of brain tissue. They are thankfully not very common but they are one of the most lethal cancers. A Radiologist’s job is to diagnose and analyse these tumors to provide the best treatment plan. Currently, the time from scan results to report can take more than four solid hours 3 and, due to the requirement of highly specialized trained individuals, these are done back-to-back reducing accuracy.

The use of Magnetic Resonance Imaging (MRI) scans

The most common diagnostic tool for brain tumors is the MRI due to its non-invasive nature and ability to image diverse tissue types and physiological processes. There is a need for accurate and fully automated methods. Manual segmentation around tumor margins on a slice-by-slice basis can take 12 minutes or more per tumor per sequence, with semi-automatic methods taking 3-5 mins 4. Automated segmentation that is not vulnerable to operator subjectivity may be beneficial 5.

Footnotes

  1. NHS Diagnostic Waiting Times and Activity Data

  2. The Brain Tumour Charity Guide to Glioblastoma

  3. DeepMind Press Release

  4. Odland A, Server A, Saxhaug C, Breivik B, Groote R, Vardal J, Larsson C, Bjørnerud A. Volumetric glioma quantification: comparison of manual and semi-automatic tumor segmentation for the quantification of tumor growth. Acta Radiol. 2015;56:1396–1403. doi: 10.1177/0284185114554822.

  5. Aslian H, Sadeghi M, Mahdavi SR, Babapour Mofrad F, Astarakee M, Khaledi N, Fadavi P. Magnetic resonance imaging-based target volume delineation in radiation therapy treatment planning for brain tumors using localized region-based active contour. Int J Radiat Oncol Biol Phys. 2013;87:195–201. doi: 10.1016/j.ijrobp.2013.04.049.