In our second post of the “Meet the Team” series, we sat down with Combinostics Imaging Scientist Jin Gong, to chat about her degrees, her personal connection with Alzheimer’s disease (AD), and her love of travel, gaming, and great bargains.
What positions have you previously held and at what companies?
For my first job, I worked as a research associate at Tampere University in Tampere, Finland. Here, I studied the hierarchical word embedding algorithm. In my second position, I worked on segmenting white matter (matter from the deepest part of the brain and central nervous system) from MRI images. I held this position while I was working on my thesis.
Are you using your academic degree or previous experience in your current position?
Absolutely! My bachelor’s degree is in software engineering (Sichuan University), so that’s where I first learned to code. My master’s degree is in computational big data analytics, which deals more with statistics than with programming. As the Imaging Scientist at Combinostics, I use the skills I gained from both degrees, but it’s my statistics background that really comes in handy. It’s given me the foundation to design algorithms here with the team.
What intrigues you the most about being at the center of technology and health care?
I first got interested in health care when my grandmother was diagnosed with AD. I always have my eye out for new drug approvals and software tools that can help with diagnostics, predicting disease progression, and ultimately improving the lives of those with AD. I know we may not find a cure during my grandmother’s lifetime, but I am dedicated to helping the mission.
What's the most interesting aspect about your position at Combinostics right now?
The most interesting aspect is the variety of the work that I do. Many of our customers have different wants and needs from our tools. For instance, one customer might want us to segment tumors while another might want to segment specific parts of the brain, like white matter, to find other types of brain abnormalities. As the segmentation area of interest changes, the models need to be adjusted.
What’s the most memorable thing you’ve learned at Combionstics?
Never give up! Once a customer requested that we segment the cerebrospinal fluid (CSF) area from CT images that they were having trouble with. We evaluated the model and couldn’t find anything wrong, and the algorithm was fine. So what was the problem? We continued to dive deep into the data, checking the sources slice by slice. We finally discovered that the wrong training data were used in the first place! As a result, we developed an automation that gives us the statistical ground truth before anything else such as image transmission, erosion, and more.
I ended up asking my manager why, given the success of automation, physicians continue to manually segment their data. He reminded me that behind all of those numbers are real patients with real diseases. All cases are unique and require a human touch to really improve lives.
Tell us a fun or unique fact about yourself.
I’m a deal hunter! All my life, I have been able to hunt out the best deals for products, and I remember the cost of the items for years! I am even flying to Milan to go shopping for Black Friday. A new winter coat is on my list this year.
What vacation destination are you dreaming of?
I can’t wait to visit Iceland. There are around 130 volcanoes in Iceland, some active and some not. The Fagradalsfjall volcano is currently experiencing an active eruption, so tourists can see the magma and lava. I’d also like to visit the Thrihnukagigur volcano, which is the only volcano that allows visitors in the magma chamber. The last time Thrihnukagigur erupted was around 4,000 years ago, so we are safe!