Machine Learning in Health Care: Develop algorithms to detect select items in an image
|Name||Brittany Dugger, PhD|
|Affiliation||University of California Davis|
|Project title||Machine Learning in Health Care: Develop algorithms to detect select items in an image|
|Background||At the University of California Davis we have one of the most diverse collection of human brains from Alzheimer’s disease and related disorders through our Alzheimer’s Disease Center ( https://www.ucdmc.ucdavis.edu/alzheimers/ ).
|Description||A specific challenge in research on dementia is obtaining rapid precise quantification of the pathology within the brain, especially those associated with changes to blood vessels. examples of some techniques and output data types: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4247208/ ( see Figure 2 ) https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5563333/ ( see Figure 3 ) Recent paper describing workflow: https://www.biorxiv.org/content/early/2018/10/30/454793|
|Deliverable||The student team will develop an automated algorithm to detect and measure the diameter of human blood vessels as well as the thickness of the blood vessel wall using post-mortem human brain tissues slides stained with hematoxylin and eosin that have been converted to digital images.
|Skill set desirable||This is a project that is especially challenging and may require learning some neuroanatomy and histopathology. This project should only be for students truly wanting to go into health care machine learning focusing on image recognition. Specific still sets include: knowledge of machine learning workflows, proficient in PyVips, python, and pytorch are desirable.
|Client time availability||30-60 min every two weeks|
|IP requirement||Open source project|