In many cases, modern ophthalmic surgery involves the video recording of surgical proceedings. The video material is either used for training purposes or for the subsequent reconstruction of operation sequences. Klaus Schöffmann has assembled a research team to work on the automatic recognition of relevant sequences within the scope of an FWF-funded project. Natalia Sokolova, a doctoral candidate at the Department of Information Technology, is a member of the project team and her work focuses on improving the search for particularly “relevant” surgical phases.
“The surgical instruments that are in use at any given point in time can provide a valuable which surgery stage is performed”, Natalia Sokolova explains. In preparation for a paper, which she is due to present at the “International Conference on Multimedia Modeling” in South Korea in January, she and her colleagues have analysed models aimed at identifying the instruments. “Related studies have already shown that it is possible to recognise instruments in videos capturing surgery on eye lenses”, Sokolova expands. However, the underlying data set is of high visual quality, unlike the typical quality of videos of this kind. Natalia Sokolova goes on: “It is our aim to analyse whether Deep Learning models trained on one dataset for instrument recognition, will also work with data from another dataset.”
The project labelled “Relevance Detection in Ophthalmic Surgery Videos“ is set to continue for two more years. This is also the time-frame within which Natalia Sokolova hopes to complete her doctoral thesis. She is well on track to achieve this goal, having already completed approximately one third of the research work. Upon completing her Master’s degree, Sokolova worked as a software tester in the industrial sector. She soon realized that the routine activities lacked creativity: “I’m keen to keep my brain active and to work on my personal development.” The doctoral degree programme in Klagenfurt offered her the opportunities she sought, particularly in the future-oriented field of image and video analysis, which is going through a phase of rapid development. In response to the question whether technical studies were always easy in her experience, she tells us: “The one person who is primarily responsible for my education success is my strong grandmother, who is a key figure in our family. She was always a powerful motivator, though she could be strict at times.” In any case, she seems to have inherited the inclination to think in terms of algorithms at an early age: “In order to learn, I always seek the logic that underpins the facts and figures. This type of thinking is useful in computer science.”