From Machine Learning to Accurate Cancer Diagnosis

Activity: Talk or presentation typesPublic lecture/debate/seminar


The classification of brain tumours is often achieved by tumour cells’ visual assessment using the microscope-based analysis of tumour samples on glass slides, termed histology. This type of assessment used for the diagnosis of a tumour sample can be very depending on the observer. Machine learning based approaches could improve cancer diagnosis without the need for such subjective diagnosis. In this talk, I would describe the development of MIMIC (minimal methylation classifier), a novel method for the routine detection of four molecular subgroups of medulloblastoma (the most common malignant childhood brain tumour) currently recognised by the World Health Organisation. I would also demonstrate its widespread potential in both immediate diagnostics application and research.
For more details:
Now, clinically available:
Period02 Oct 2019
Held atSchool of Electronics, Electrical Engineering and Computer Science
Degree of RecognitionLocal


  • minimal methylation classifier
  • medulloblastoma
  • diagnosis
  • classification of brain tumours
  • World Health Organisation
  • Web-based software
  • improve cancer diagnosis without the need for such subjective diagnosis