The Lanit-Tercom team is currently working on a project using fluorographies to identify pathologies. We were handed several thousand anonymous and digitized fluorographies, part of which were marked as featuring pathologies.
We are developing a solution that identifies deviations from the norm using deep learning. For the first stage, we plan to filter out deviations from how the normal ribcage is supposed to look. Once that is done, we’ll analyze the deviations themselves, using that information to come up with software doctors can use to make their jobs much easier. Healthy patients will be set aside; attention will be drawn to cases where there are clear deviations.
Not only that, but we’re also involved in another project that’s similar, only using smart data analysis of ultrasounds to diagnose cerebrovascular diseases.
It’s an international project with a Chinese scientific group managed by Professor Mingyue Ding.
The smart data analysis tools developed for the project could easily be used in a variety of applications, not just for automating ultrasound diagnosis. If testing goes well, a hardware platform will be built on a field-programmable gate array (FPGA).