Beyond human vision: How AI can help in smarter tissue analysis

From complex data to meaningful insights

When light passes or reflects off biological tissues, it changes its direction, intensity and polarization (the orientation of its waves) state. Cancer tissue types reflect light differently as compared to healthy ones. By visualizing the properties of polarized light, we may indicate the presence of abnormalities.

My research uses light polarization properties to detect abnormalities in tissues. Analysis of this polarimetry data is a very time consuming and difficult task to perform manually. Machine learning models excel at processing complex data and extracting useful information at much higher speed without compromising the accuracy of the system.

My research focuses on use of machine learning models to analyze data collected from polarized light imaging. By training AI models on this type of data, we aim to develop a model that can predict cancer earlier, more accurately and efficiently in real time as well. It will help doctors in early diagnosis of disease so that they can make treatment plans beforehand and mortality rate will be reduced.

Transforming cancer detection through physics and AI

Cancer disease has become a major health concern worldwide. It’s affecting many people and spreads to other parts of the body. It is the leading cause of death worldwide and this number is continuing to increase with each passing year. Diseases like cancer often remain hidden until they progress. Early detection is the key in fighting this disease. Many cancers are treatable when detected on time, but the current methods are slow, invasive and limited in accuracy.

In my research, I am using formalin-fixed, paraffin-embedded (FFPE) blocks of human breast tissue obtained from Biobank Borealis (Oulu, Finland) that were scanned using a custom-built polarimetric imaging system. We have used machine learning models such as support vector machine (SVM) and random forest (RF) to analyze the obtained datasets and classify them into two tissue types: adipose and fibrotic. To evaluate model performance, we have analyzed accuracy, sensitivity, and specificity of the model. The results are very promising and in the next step, we are planning to apply deep learning models for three different types of tissues. By combining physics-based measurements (polarized based light interaction with tissue samples) with advanced AI models, we can move towards smart healthcare technology having real-world impact.

Reflecting on my research journey, I also began my secondments in Aston University in United Kingdom in March 2025 and during this time I had a pleasure of getting to know Professor Igor Meglinski and his group. Their warm welcome made me feel at ease and helped me to get started my work in new environment. As I continue to work, I am excited by the interdisciplinary nature of my research field. From engineering to AI to biology, every part of this project contributes to a shared vision: helping people live healthier lives through early, smart and reliable diagnostics. I am hoping that my research not only contributes to the academic progress but also to the real-world impact – where AI truly helps us see what we can’t.

Acknowledgement: I would like to thank my supervisor Dr. Aliaksander Bykau for their support. I am also thankful to I4WORLD Doctoral Programme and European Union for co-funding this project.