How new image-processing methods can accelerate drug studies

With a research project around the disease cystic fibrosis, medicine and computer science are enriching the field with new insights. The Italian researcher Paola Lecca is developing at the Free University of Bozen-Bolzano statistical processing methods for data from medical images that can be used to identify anomalies in the growth of organoids — small organic structures that simulate an organ.

The physicist at the Faculty of Computer Science thus provides valuable basics to accelerate drug studies. For the Faculty, Bioinformatics is an area of research that is increasingly coming into focus.

Cystic fibrosis is a severe congenital metabolic disease caused by a mutation in the CFTR gene. To develop new drugs for the treatment of this mutation in the cystic fibrosis gene, a team of doctors at the University of Verona is focusing on the use of intestinal organoids. An organoid is a simplified, miniaturized version of a three-dimensional organ produced in vitro, which has realistic microanatomical properties. These organoids are treated with various new drugs in the laboratory, with growth being minutely recorded by video under a microscope.

What effect has anisotropic growth?

The effect and effectiveness of the drugs are determined by the change in shape of the organoid: Since organoid growth is anisotropic, i.e. irregular, the researchers need a large number of video images for the volume calculation as well as image-processing methods capable of evaluating these huge amounts of image and video data.

For this bioinformatics challenge, the Faculty of Computer Science at the Free University of Bozen-Bolzano and the Bruno Kessler Research Institute have initiated the CORVO project, which over a period of two years will provide a mathematical principle for calculating the volume of organoids in medical images based on the case study of cystic fibrosis.

What does CORVO stand for?

CORVO stands for Computing ORganoid’s VOlume. CORVO is also the name of the software that recognizes the organoids in the medical images. The software was developed together with the research group “Technologies for Vision” of the Bruno Kessler Foundation. The software analyses temporal sequences of three-dimensional images of physiological solutions in which some organoids are immersed. Through a process called segmentation, CORVO identifies the regions occupied by the organoids in each image and calculates their volume. This process, developed by FBK researcher Michela Lecca, runs completely automatically but, thanks to a user interface specifically designed for medical staff, allows them to interact with the software to select organoids of clinical interest, for example.

Finally, CORVO integrates a statistical analysis of the temporal variations in the volume of the organoids, which makes it possible to distinguish the different responses of the organoids to the various pharmacological treatments.

“Bioinformatics is strategically very interesting for us as a faculty, as it shows the importance of informatics for medicine,” emphasizes Dean Prof. Claus Pahl.

On the special case of cystic fibrosis

Patients born with the disease cystic fibrosis have inherited a copy of the mutated gene from both their father and mother. Both are almost always unknowingly healthy carriers of a copy of this gene, which normally determines the synthesis of a protein called CFTR (Cystic Fibrosis Transmembrane Conductance Regulator) and is responsible for the regular function of the secretions of many organs. In people born with the double copy of the mutated gene, this protein functions little or not at all.

What do statistics say?

In Italy, there is about one healthy carrier for every 25 people. As a couple, two healthy carriers have a one-in-four chance of having a sick child with each pregnancy, i.e. with two copies of the mutated gene.

“At the Faculty of Computer Science at unibz, I have developed a mathematical method based on advanced statistical techniques that allows us to identify, in the medical setting, which organoids develop while maintaining a spherical shape and which develop in irregular shapes,” explains Paola Lecca.

“Intestinal organoids could be figuratively described as balloons that can be filled with water through many openings: these CFTR channels are defective in cystic fibrosis.

Since water entry depends largely on the opening of the CFTR channels, organoids from healthy individuals have a swollen and rounded appearance due to the activity of well-functioning CFTR channels; in contrast, organoids from diseased patients appear folded in on themselves, reduced in volume and not spherical. Regular growth of the organoid in response to a drug is an indication of the effectiveness of that drug. Failure of the organoid to bulge is an indicator that the drug is not effective,” she says. This differentiation between organoids that respond well to a drug and organoids that do not respond or respond poorly to a drug by calculating volumes would not be possible without the help of software and algorithms based on sound mathematical principles.”

“I participate in the research team led by Professors Bruno Carpentieri in the LACS lab and Diego Calvanese in the Smart Data Factory, the technology transfer lab of the Faculty of Computer Science,” says Lecca. “We observe changes in the three-dimensional form of intestinal organoids derived from healthy individuals as well as patients with cystic fibrosis. The organoids are subjected to different drug treatments and we calculate the volume of these mini-intestines as they change over time in response to different drugs, as well as their rate of growth or decline.

The experimental data for the study were generated by the Laboratory for Applied Research in Cystic Fibrosis of the Faculty of Medicine (General Pathology Section) of the University of Verona, under the direction of Claudio Sorio and in collaboration with doctors from the Cystic Fibrosis Centre of Verona “Azienda Ospedaliera Integrata” (Paola Melotti).

For the Faculty of Computer Science, researcher Paola Lecca has published first results under the title “Computing organoids’ volume in medical images: The case study of cystic fibrosis” in the library IEEE Xplore. This is the world’s largest database of high-quality technical literature in the fields of electrical engineering, computer science and electronics. Previously, Paola Lecca has developed mathematical models for the study of resistance to cancer therapies; she is currently working on the implementation of computational mathematical methods to predict response to experimental pharmacological treatments for chronic myeloid leukaemia, in close collaboration with doctors from the Department of General Pathology at the University of Verona and with the Verona Cystic Fibrosis Integrated Opedical Company.

Paola Lecca, a native of Trento, studied theoretical physics at the University of Trento, where she later completed a doctorate in computer science and telecommunications, since 2019 she has been conducting research at the Faculty of Computer Science in Bolzano. She has already presented her computational model at the IEEE Symposium on Computational Intelligence and Bioinformatics and Computational Biology (CIBCB) conference. Read here the paper.