Computer vision enables the analysis of nanoparticles in industrial and medical applications

Computer vision, operating at a microscopic level, can compute the characteristics of nanoparticles of a given substance. Compared to human observations and calculations, automated computation is significantly faster and less costly. In a dissertation published by LUT University, a novel computational method introduces new possibilities for industrial and biomedical applications. For example, biomedical applications could use the method to detect cancer cells.


Automated particle computation is challenging because the studied images nearly always consist of a large number of particles that vary in shape, size and appearance. The overlapping of the particles makes the task even more difficult. In industrial and biomedical applications, changes in particle size and shape must be detected reliably to ensure for instance the safety of a material or to identify dangerous cancer cells.

Sahar Zafari, a researcher at LUT University, has introduced a computer vision method that can automatically detect overlapping particles of similar shapes. Zafari's dissertation in the field of computational engineering and analytics will undergo its public examination at the University of Texas, in the United States, where Zafari is currently working as a visiting researcher from LUT.

"The scientific value of Zafari's application lies in its ability to detect overlapping particles," explains Professor Heikki Kälviäinen of the LUT School of Engineering Science.

Sahar Zafari's study relied on the assumption that the shape of particles is convex. The reliability of the study was assessed by based on synthetic and real data. The data presented objects of various shapes that were composed of particles of different sizes overlapping fully or partially.

"Zafari's method proved more accurate in detecting and segmenting particles than competing methods," states Professor Kälviäinen.

Hazards of titanium dioxide draw global attention

Zafari's dissertation is closely related to research on the impacts of titanium dioxide, which has drawn significant global attention. Titanium dioxide is used as white colourant in, for example, foods and toothpastes. It is currently being investigated extensively due to a suspected link to the onset of cancer.

"The impact of titanium dioxide on human cancers has yet to be confirmed. If evidence shows that it contributes to the development of cancer, we will be able to employ computer vision to assist diagnostics. In any case, the methods can be applied to cellular analysis," relates Kälviäinen.

The dissertation of Sahar Zafari, M.Sc. (Tech.), entitled Segmentation of Partially Overlapping Convex Objects in Silhouette Images will undergo a public examination at the University of Texas at Austin, in the United States, on 29 November 2018 at 10 a.m. local time. The public examination will be streamed live to the conference room 2411 at Lappeenranta University of Technology, Lappeenranta, Finland on the same day as above, at 6 pm. The opponent will be Professor Matti Pietikäinen from the University of Oulu, Finland. The custos will be Professor Heikki Kälviäinen from LUT University.

More information:
Heikki Kälviäinen, Professor, LUT University, tel. +358 40 586 7552