KTU graduate leverages AI for skin cancer detection

“For the past year and a half, I’ve been doing projects related to cancerous issues, and this is like a culmination,” says Ukrainian Yehor Yakymov. He graduated from KTU with a bachelor’s in Robotics this June. Inspired by his family’s health journey, Yehor developed a new skin analysis algorithm to detect early melanoma. He successfully presented his project at the annual innovation exhibition Technorama 2026 during his final year of studies.

“My mother had cancer – it was a melanoma, breast cancer and more. So, it was always something I was aware of since a young age. It’s been a very long journey, and thank God everything is fine now, but it definitely gave me a lot of inspiration,” Yehor shares, explaining the deep personal drive behind his project.

Translating Medical Methodology into AI Code

The project goal was to develop a tool to extract features from medical skin images.

“This is a feature extraction algorithm of malignant skin lesions, mainly melanoma. It is based on medical data on how it is really done. It is based on the ABCD method, which basically measures the asymmetry, border regularity, colour variation, and diameter of the lesion,” Yehor explains. “Based on these four parameters, the score is calculated to see if the lesion is benign, suspicious, or malignant.”

To achieve this, Yehor selected a real-world medical method and implemented it through programming. First, he built a strong preprocessing step to clean all artifacts and noise from the skin images. Next, he trained a special artificial intelligence model called Attention U-Net. He tested multiple setups to find the best architecture for making image masks. Finally, his system uses these masks within a custom ABCD framework to calculate the Total Dermoscopic Score.

The long-term vision of this algorithm lies in its potential for mobile integration and remote screening. “In the future, it’s possible to apply it to telecommunication easily. You could take a picture, and if you knew your dimensions and everything, you could run it through and have a preliminary check. It could tell you: is this something you should be aware of? Should you go and look into it further?”

The Power of Robotics

Reflecting on his educational journey at KTU, Yehor emphasizes that a traditional, pure programming background would not have sufficed to bridge the gap between software and real-world system integration.

“In this project, you have to understand a lot of things like visual systems and how it can all be integrated,” the KTU graduate notes. “This is where the Robotics programme definitely helps because it is partially automation. In fact, this project that I did is called computer-aided diagnosis in medical terms, which is an automation of a diagnosis. It allowed me the expertise to understand why certain things are unable to be implemented simply.”

Looking forward, Yehor is highly motivated to turn his bachelor’s success into an ongoing scientific pursuit: “I would really wish to continue working on it to develop it further, because I think it has potential and it is something that is very relevant to me.”

During their studies in this programme, students gain extensive knowledge of artificial intelligence, computer vision, the Internet of Things (IoT), digital twins of robots and their systems, electromechanical processes occurring in robots, and the control of complex technical systems.

Are you interested in exploring these advanced technologies and building real-world automated systems? Learn more about the Robotics study programme at KTU by visiting admissions.ktu.edu.