Algorithms will be very useful for doctors, who work 24/7, says one of the top authorities in AI - Prof. Petia Georgieva from the University of Aveiro and head of the research group "Automated Systems with Artificial Intelligence" at the Technical University of Sofia
Everything I have achieved is based on what I received at this educational institution, my education was at a world-class level, I will never be able to repay what I received, she believes
When a scientist with international recognition in the field of artificial intelligence begins cooperation with Bulgarian colleagues from the Technical University - Sofia, then the effect can be nothing other than an "explosion" of useful solutions in medicine, such as assisting medics in the diagnosis of various diseases or robotics, computer vision, flood prediction, etc.
Prof. Dr. Petia Georgieva is among the indisputable authorities in the field of intelligent systems and AI and has been working for years at the University of Aveiro, Portugal. In addition, she has been a guest lecturer at Carnegie Mellon University (CMU) in Silicon Valley, at the CMU campus in Pittsburgh and at a number of other American universities, as well as at the University of Lancaster in the UK.
Prof. Georgieva is a member of the board of the International Neural Network Society, editor in two of the most prestigious journals (rank Q1) in the field of neural networks and machine learning - Neural Networks and Pattern Recognition published by Elsevier.
She herself has never lost contact with her colleagues from the Technical University - Sofia (TU - Sofia), but recently their cooperation has moved to a new level. Then the educational institution, together with several other universities in the country, were announced as research institutions with funding through the National Recovery and Resilience Plan.
"Thus, cooperation between TU - Sofia and the University of Aveiro expanded," Prof. Petia Georgieva told "24 Chasa".
She is also an alumnus of the Technical University, which helps to launch an important research topic - computer vision with a high degree of reliability in processing medical images.
The team is working on AI algorithms that determine which parts of the brain, captured by nuclear magnetic resonance imaging, show signs of tumors or early signals of their development.
"We wanted to develop an algorithm that would convince doctors that it gives correct answers, that it is not making things up, and that it has indeed extracted important characteristics," explained Prof. Georgieva. Therefore, she, together with PhD student Aykut Ismailov, Assoc. Prof. Vladimir Hristov, and Ina Zheleva, who is currently pursuing a master's degree in Vienna, formulated the task so that the algorithm explains on the basis of what arguments it made a given decision and points to which part of the image
Thanks to tumor markers, as well as with the help of doctors, the team managed to train an AI model to give correct answers with high accuracy and explainability. In this way, regardless of whether a doctor is tired at the end of the day, or their vision has weakened after many hours of staring at X-ray images, or they still have little experience, the technology will be an invaluable advisor and guide.
"We are trying to make the system work with low-budget hardware, with laptops; currently, for the purposes of computer vision, expensive computers and supercomputers are required," Vladimir Hristov from the Faculty of "Automation" explained to "24 Chasa". "For example, now with a PhD student from the team, we are trying to create a system that does not exist on a global scale – capturing a 3D image for inspecting micro-optical details under specific conditions. The other direction in which we work is industrial robots, to which we give intelligent functions. Again, we strive to keep them low-cost, but without losing quality – the goal is to simultaneously lighten the AI models and maintain high accuracy in image recognition."
Another successful project is related to surveillance cameras produced by the company "Bosch," with which TU-Sofia has a long-term collaboration.
"We developed a method for quickly detecting defects in the image sensor," explained Petia Georgieva. "This is the most important part of the video camera; if the sensor does not work well, if it has defects, the camera must be discarded, customers are dissatisfied, complaints begin, funds are lost, etc."
Therefore, the team at TU-Sofia, under her leadership, created a computer vision algorithm, with the goal of "scanning" the sensor for potential defects. As a result of many experiments with real data taken from the company, they created precise detection. The company "Bosch" immediately applied the algorithm in the video camera production process.
"We have always sought solutions to specific problems that the industry needs," Petia Georgieva smiles. According to her, the collaboration between the university and the industry is beneficial for everyone.
"Companies usually approach us, offering internships and thesis topics - Petia Georgieva explains the mechanism. - More prosperous firms provide a small scholarship, which attracts more students' interest, allowing us to select the best interns. If the company is satisfied with the results, there is a high chance they will hire the young engineer."
Another area where scientists work is fifth and sixth-generation networks. Their innovative development assists companies like Altice labs, which focuses on optical communications.
"Their problems are related to improving the quality of symbol decoding - Petia Georgieva explained. - For example, for my voice to travel from point A to point B, there are many signal conversions from voice to electrical, optical, and back. However, during this process, many noises and various defects appear, which are due to the optical communication channel. We trained an algorithm to improve the quality of symbol decoding."
The development proved to be extremely successful, so within the framework of project BG-RRP-2.004-0005, funded under the National Recovery and Resilience Plan, the scientific group "Automated Systems with Artificial Intelligence" at TU - Sofia, led by Prof. Georgieva, produced a number of scientific publications in authoritative Q1-ranked journals.
"Altice labs already offer the innovation as a service to their more affluent clients," Petia Georgieva noted.
According to her, opinions on artificial intelligence today are quite diverse - from "hallelujah" to "crucify it," but there is one area of application where there are no negative opinions about this technology, and that is medicine.
"Doctors were more reserved years ago, but now AI models, developed with the help of the best specialists in each field of medicine, are quite effective - Prof. Georgieva points out. - These algorithms can quickly and accurately filter out concerning images and focus doctors' attention to more thoroughly check a given hypothesis."
The models themselves are trained and improved thanks to the fact that more and more data is accumulated. And especially by the European Union, which is the strongest advocate of the idea of open science and sharing results for faster progress.
"Unlike in Europe, most AI-related research in the US is conducted by private companies primarily seeking profit, while in the EU, databases are becoming increasingly accessible to scientists," she adds.
It is precisely with such public medical data that Petia Georgieva's team, colleagues from TU - Sofia, and scientists from other universities and doctors train AI algorithms to recognize pathological cases among millions of images.
This technology then becomes a reliable advisor for every doctor who wants to perform quick and reliable screening for various diseases and make accurate diagnoses for their patients.
Within the framework of the contract, which has been ongoing for two and a half years and will conclude in mid-2026, the team is intensively working on various projects. Among them is the AI model for predicting floods based on meteorological data on water volumes flowing into estuaries, proposed by the team of Prof. Petia Georgieva and Assoc. Prof. Tsonyo Slavov from the Faculty of "Automation" at TU - Sofia.
"The data were collected from rivers in Portugal, and an algorithm was trained with them to warn of potential floods and to take measures against the disaster," explains Petia Georgieva.
How are all these results achieved?
"Through hard work, perseverance, relying on the achievements of science before us, and, of course, on our talented students," Petia Georgieva recounts. "These models are complex, their training is lengthy, tuning is required, and many times the results are discouraging."
Nevertheless, with the help of European and national funding, as well as scholarships provided by businesses, the team strives to retain the best students. Such partnerships between the university and industry ensure the necessary motivation for talents not to go abroad to study.
"Actually, this is the only way a country can develop its economic potential and grow," notes Petya Georgieva thoughtfully.
She herself claims that she will never forget where she started: "Everything I have achieved is based on what I learned at TU - Sofia (then it was called the Higher Institute of Mechanical and Electrical Engineering - HIME), our education was at a world-class level.
All my colleagues who left Bulgaria for one reason or another have done very successfully, and this is due mainly to the training at TU - Sofia.
I would really like our cooperation to continue after the completion of this project. I work with great pleasure with students from TU - Sofia, they are my way of repaying the university that has given me much more than I will ever be able to return.
Working with students is the most important mission in my profession. To awaken their interest, to see how they begin to become more creative, giving increasingly original ideas and finally becoming better than me, this is great inspiration."
She admits that she is extremely happy to contribute to their formation as engineers.
"I have received cards from students that make me cry, especially when they write to me that I am the teacher with a capital letter in their lives - smiles Prof. Petya Georgieva. - Recently, TU - Sofia celebrated its 80th anniversary, I wish all colleagues and students health, success and scientific prosperity."
From this point of view it may seem incredible, but thanks to the accumulated capacity and joint efforts today the Bulgarian university seems to have accelerated in the field of artificial intelligence and it is only a matter of time before it reaches the best in this field.
24chasa.bg
Everything I have achieved is based on what I received at this educational institution, my education was at a world-class level, I will never be able to repay what I received, she believes
When a scientist with international recognition in the field of artificial intelligence begins cooperation with Bulgarian colleagues from the Technical University - Sofia, then the effect can be nothing other than an "explosion" of useful solutions in medicine, such as assisting medics in the diagnosis of various diseases or robotics, computer vision, flood prediction, etc.
Prof. Dr. Petia Georgieva is among the indisputable authorities in the field of intelligent systems and AI and has been working for years at the University of Aveiro, Portugal. In addition, she has been a guest lecturer at Carnegie Mellon University (CMU) in Silicon Valley, at the CMU campus in Pittsburgh and at a number of other American universities, as well as at the University of Lancaster in the UK.
Prof. Georgieva is a member of the board of the International Neural Network Society, editor in two of the most prestigious journals (rank Q1) in the field of neural networks and machine learning - Neural Networks and Pattern Recognition published by Elsevier.
She herself has never lost contact with her colleagues from the Technical University - Sofia (TU - Sofia), but recently their cooperation has moved to a new level. Then the educational institution, together with several other universities in the country, were announced as research institutions with funding through the National Recovery and Resilience Plan.
"Thus, cooperation between TU - Sofia and the University of Aveiro expanded," Prof. Petia Georgieva told "24 Chasa".
She is also an alumnus of the Technical University, which helps to launch an important research topic - computer vision with a high degree of reliability in processing medical images.
The team is working on AI algorithms that determine which parts of the brain, captured by nuclear magnetic resonance imaging, show signs of tumors or early signals of their development.
"We wanted to develop an algorithm that would convince doctors that it gives correct answers, that it is not making things up, and that it has indeed extracted important characteristics," explained Prof. Georgieva. Therefore, she, together with PhD student Aykut Ismailov, Assoc. Prof. Vladimir Hristov, and Ina Zheleva, who is currently pursuing a master's degree in Vienna, formulated the task so that the algorithm explains on the basis of what arguments it made a given decision and points to which part of the image
Thanks to tumor markers, as well as with the help of doctors, the team managed to train an AI model to give correct answers with high accuracy and explainability. In this way, regardless of whether a doctor is tired at the end of the day, or their vision has weakened after many hours of staring at X-ray images, or they still have little experience, the technology will be an invaluable advisor and guide.
"We are trying to make the system work with low-budget hardware, with laptops; currently, for the purposes of computer vision, expensive computers and supercomputers are required," Vladimir Hristov from the Faculty of "Automation" explained to "24 Chasa". "For example, now with a PhD student from the team, we are trying to create a system that does not exist on a global scale – capturing a 3D image for inspecting micro-optical details under specific conditions. The other direction in which we work is industrial robots, to which we give intelligent functions. Again, we strive to keep them low-cost, but without losing quality – the goal is to simultaneously lighten the AI models and maintain high accuracy in image recognition."
Another successful project is related to surveillance cameras produced by the company "Bosch," with which TU-Sofia has a long-term collaboration.
"We developed a method for quickly detecting defects in the image sensor," explained Petia Georgieva. "This is the most important part of the video camera; if the sensor does not work well, if it has defects, the camera must be discarded, customers are dissatisfied, complaints begin, funds are lost, etc."
Therefore, the team at TU-Sofia, under her leadership, created a computer vision algorithm, with the goal of "scanning" the sensor for potential defects. As a result of many experiments with real data taken from the company, they created precise detection. The company "Bosch" immediately applied the algorithm in the video camera production process.
"We have always sought solutions to specific problems that the industry needs," Petia Georgieva smiles. According to her, the collaboration between the university and the industry is beneficial for everyone.
"Companies usually approach us, offering internships and thesis topics - Petia Georgieva explains the mechanism. - More prosperous firms provide a small scholarship, which attracts more students' interest, allowing us to select the best interns. If the company is satisfied with the results, there is a high chance they will hire the young engineer."
Another area where scientists work is fifth and sixth-generation networks. Their innovative development assists companies like Altice labs, which focuses on optical communications.
"Their problems are related to improving the quality of symbol decoding - Petia Georgieva explained. - For example, for my voice to travel from point A to point B, there are many signal conversions from voice to electrical, optical, and back. However, during this process, many noises and various defects appear, which are due to the optical communication channel. We trained an algorithm to improve the quality of symbol decoding."
The development proved to be extremely successful, so within the framework of project BG-RRP-2.004-0005, funded under the National Recovery and Resilience Plan, the scientific group "Automated Systems with Artificial Intelligence" at TU - Sofia, led by Prof. Georgieva, produced a number of scientific publications in authoritative Q1-ranked journals.
"Altice labs already offer the innovation as a service to their more affluent clients," Petia Georgieva noted.
According to her, opinions on artificial intelligence today are quite diverse - from "hallelujah" to "crucify it," but there is one area of application where there are no negative opinions about this technology, and that is medicine.
"Doctors were more reserved years ago, but now AI models, developed with the help of the best specialists in each field of medicine, are quite effective - Prof. Georgieva points out. - These algorithms can quickly and accurately filter out concerning images and focus doctors' attention to more thoroughly check a given hypothesis."
The models themselves are trained and improved thanks to the fact that more and more data is accumulated. And especially by the European Union, which is the strongest advocate of the idea of open science and sharing results for faster progress.
"Unlike in Europe, most AI-related research in the US is conducted by private companies primarily seeking profit, while in the EU, databases are becoming increasingly accessible to scientists," she adds.
It is precisely with such public medical data that Petia Georgieva's team, colleagues from TU - Sofia, and scientists from other universities and doctors train AI algorithms to recognize pathological cases among millions of images.
This technology then becomes a reliable advisor for every doctor who wants to perform quick and reliable screening for various diseases and make accurate diagnoses for their patients.
Within the framework of the contract, which has been ongoing for two and a half years and will conclude in mid-2026, the team is intensively working on various projects. Among them is the AI model for predicting floods based on meteorological data on water volumes flowing into estuaries, proposed by the team of Prof. Petia Georgieva and Assoc. Prof. Tsonyo Slavov from the Faculty of "Automation" at TU - Sofia.
"The data were collected from rivers in Portugal, and an algorithm was trained with them to warn of potential floods and to take measures against the disaster," explains Petia Georgieva.
How are all these results achieved?
"Through hard work, perseverance, relying on the achievements of science before us, and, of course, on our talented students," Petia Georgieva recounts. "These models are complex, their training is lengthy, tuning is required, and many times the results are discouraging."
Nevertheless, with the help of European and national funding, as well as scholarships provided by businesses, the team strives to retain the best students. Such partnerships between the university and industry ensure the necessary motivation for talents not to go abroad to study.
"Actually, this is the only way a country can develop its economic potential and grow," notes Petya Georgieva thoughtfully.
She herself claims that she will never forget where she started: "Everything I have achieved is based on what I learned at TU - Sofia (then it was called the Higher Institute of Mechanical and Electrical Engineering - HIME), our education was at a world-class level.
All my colleagues who left Bulgaria for one reason or another have done very successfully, and this is due mainly to the training at TU - Sofia.
I would really like our cooperation to continue after the completion of this project. I work with great pleasure with students from TU - Sofia, they are my way of repaying the university that has given me much more than I will ever be able to return.
Working with students is the most important mission in my profession. To awaken their interest, to see how they begin to become more creative, giving increasingly original ideas and finally becoming better than me, this is great inspiration."
She admits that she is extremely happy to contribute to their formation as engineers.
"I have received cards from students that make me cry, especially when they write to me that I am the teacher with a capital letter in their lives - smiles Prof. Petya Georgieva. - Recently, TU - Sofia celebrated its 80th anniversary, I wish all colleagues and students health, success and scientific prosperity."
From this point of view it may seem incredible, but thanks to the accumulated capacity and joint efforts today the Bulgarian university seems to have accelerated in the field of artificial intelligence and it is only a matter of time before it reaches the best in this field.
24chasa.bg
