Deep Phenotyping from Photographs
What matters to me is “developing algorithms for good” (or something less cheesy!). Computer science and big data are fascinating disciplines, but I think it’s important that software developers keep in mind their obligation and responsibility towards society.
Shiri on the bottom left, at the computer lab in Oxford
At Oxford, I researched the development of algorithms that diagnose patients with rare genetic disorders via their facial photos. The application of my research is that it can helping obtain faster diagnoses for patients and help discover new syndromes.
For my MSc dissertation in computer science, I was fortunate to be able to collaborate with Dr. Chris Nellaker of the Big Data Institute and Prof. Andrew Zisserman of the Department of Engineering Sciences. My research is motivated by the fact that even today, when parents bring in their child to a doctor due to a health concern, the process of obtaining a correct diagnosis of a rare genetic diagnosis is often prolonged. The difficulty lies with the fact that there are thousands of rare genetic syndromes and many of them are truly rare: doctors may not have come across them before or would not consider them for a possible diagnosis. For example, only 1 in 200,000 infants are born with Kleefstra Syndrome.
One of the possible clues to a diagnosis is if the disease has associated with it some facial changes. Clinical dysmorphology is a specific clinical expertise domain for discerning facial feature changes to help make a diagnosis. One example of this type of analysis is in the picture of Abraham Lincoln below. The suggestion is that he had a marfanoid disorder, or Marfan syndrome. The algorithms I am working on are able to predict likely syndromes given a 2D facial photo.
Picture from https://elifesciences.org/articles/02020
The hope is that clinicians will eventually be able to use such technologies to take a photo of a patient in order to help obtain a faster diagnosis.
There are so many figures who inspire me in my work. One of them is Edsger Dijkstra who was a pioneer in computer science – it is simply incredible to read about all his contributions to my field. I recently read a biography of Marie Curie, and she is a wonderful example of a scientist who cared and worked hard to give back to society. I strongly recommend reading about her development of radiological cars in France during World War I.
I am excited to see what fields computer science revolutionizes next.
