Advanced Deep Learning Methods for Automatic Segmentation of Brain Tumours and other Intracranial Structures

Preoperative manual segmentation of Brain tumours is an extremely time-consuming task and adds an enormous amount of Neurosurgical workload. Recent enhancement in computer-vision technology based on advanced deep-learning strategies shows possibilities to precisely segment intracranial structures, including Brain tumours, in MRI images. An accurate auto-segmentation tool can help to save neurosurgeon’s time and automatically evaluate the volumetric features of a tumour, along with its exact location relative to its adjacent tissues, in seconds.

This work aims to use archived data from Auckland Hospital to develop highly-advanced machine-learning algorithms (i.e., deep-learning-based models) to precisely and automatically segment patient-specific brain tumours (e.g., meningiomas), along with their adjacent intracranial structures (including brain tissues, blood vessels, cranial nerves, etc) with an acceptable accuracy compared to human annotators. Results will provide a unique opportunity to build an intelligent computational auto-segmentation platform for pre-surgical assessments, planning and navigation, and finding the most optimal accessible corridors for resection. 

• An enthusiastic team-player with interests in the application of technology and advanced machine learning in healthcare (neurosurgical imaging).
  
  
• Confident proficiency in coding, machine-learning, signal and image processing.
  
  
• Evidence of interest in the topic, self-motivation, and creative and critical thinking.
  
  
• Previous experience and background in one or more of computer science, biomedical engineering, coding and machine learning (deep-learning), and neurophysiological signal processing would be advantageous.
  

Funding source (if applicable)