Web service and software to support diagnosis in pulmonology and oncology
09.12.2024
The development was told by the junior researchers of the laboratory for the analysis of biomedical imaging of the Institute Alexandra Kosareva and Ahmedkhan Rajabov.
— Our software complex allows you to examine in detail the radiological images of the lungs, - explains A. Kosarev. — With the help of artificial intelligence (AI) methods, it recognizes, classifies or distinguishes areas of images associated with different types of pathologies in this organ. — Such web services in the world already exist, — complements colleague A. Rajabov. — The uniqueness of our system is that it is “sharpened” primarily for the population of Belarus. Representatives of different peoples differ from each other anatomically — depending on the conditions of the environment, climate, genetics and other factors. Variability of the shape of the lungs is also present. All this determines the degree of stability or, conversely, the predisposition of the body to the development of certain diseases. Then, taking into account these factors, a statistical analysis is performed.
Therefore, IPI scientists used a closed array of images of light from domestic medical institutions to create a software complex - about 10 million images. This made it possible to adapt the model of artificial intelligence to the peculiarities of the population of Belarus. It already works in test mode on the server of the institute and is available via the Internet. Any user can visit the site lungs.org.by and download a chest X-ray image. In a few minutes, the system will give an answer on the screen about the results of the study. It provides the so-called second opinion to minimize the gaps and repeat results. The final conclusion in any case is formed by a radiologist.
— The nodules in the lungs can be benign and malignant. Therefore, the advantage of our software complex is also that it assesses the likelihood of malignancy found in a light formation. Kosarev. - And the final diagnosis will be given additional tests of the patient, for example, the results of the biopsy. On X-ray images, the IPI web service allows you to detect manifestations of tuberculosis, bronchitis, pneumonia, localization of focal changes (such as “matte glass”). In the images of computed tomography, AI finds signs of nodular clusters of infiltrates, fibrosis, telectase, pleurisy, pneumothorax and caverns.
— How will the introduction of your development in the practice of domestic health care affect the quality of medical care in our country?
— The exact diagnosis of lung diseases strongly depends on the time that the doctor can devote to the study of the picture of a particular patient, — explains A. Rajabov. — His practical experience also plays an important role. But it is unlikely that Belarus has a doctor who analyzed 10 million pulmonary images. But the application on such an extensive database of national data can very quickly compare them with a picture of a particular patient and quickly identify the pathology. Therefore, our AI models built into applications are designed to simplify the work of doctors.
According to A. Kosareva, the advantage of the web service is also the streaming of data research. The specialist can immediately send to the server all the pictures that he needs to study, and after a certain time to pick out the results. According to the head of the laboratory for the analysis of biomedical imaging of the IIPI of the NAS of Belarus, Candidate of Technical Sciences Eduard Snezhko, elements of this software complex began to be created at the institute about 15 years ago. Now its processing is taking place, timed to a large update of the entire information system in the republic. Modernization began with a basic data storage and transmission system, security at a modern level. Further, automated workplaces of doctors of different profiles, including radiologists, will be created. Medical facilities will connect to the IPI NAS server. Doctors will be able to transmit images directly from their office to it, and then get the results of the study of images by artificial intelligence on the computer. — In the development of the web service we are moving in two directions at once, — emphasize E. Snezhko. The first is for the people of Belarus. The second is for mutually beneficial cooperation with representatives of other countries. Therefore, we expand the database of international data on chest images. Across the covers in the lungs received a set of images from Georgia. It is one of the few states where caverns are extracted from the lungs through surgery. And we try to match the manifestations of such caverns in the images of the CT with what their surgeons discover. Also a few years ago, a lot of testing of our software complex was carried out by colleagues from Tatarstan — technical specialists and radiologists. From this region for about six months, we were sent chest images, and then evaluated the results obtained by our system. Now we have joined the international community to combat tuberculosis and are receiving data from computed tomography from them.
— Such cooperation allows you to establish links with clinics and universities of different countries, — explains A. Kosarev. — It brings projects with financing of the Belarusian Republican Foundation for Basic Research, the State Committee for Science and Technology and other large organizations.
E. Snezhko notes that while the interface of the web service in English. On the IPI server, the software complex is available on the Internet to everyone from anywhere in the world in academic mode. It is more convenient to open it from a computer. A person comes to the site under a personal account and sees the results of only his pictures. It can pick up the results of the analysis, and then remove them from the institution server. Responsibility for the security of personal data when downloading images is still assigned to the user. He has the ability to independently clean up all his personal information from the data. In a large automated system, the security of personal data will be resolved at a professional level. Reason
For his development A. Kosareva and A. Rajabov was awarded the title of laureates of the exhibition-competition "100 innovations of young scientists" 2024.
Source: agency «Minsk-News»
— Our software complex allows you to examine in detail the radiological images of the lungs, - explains A. Kosarev. — With the help of artificial intelligence (AI) methods, it recognizes, classifies or distinguishes areas of images associated with different types of pathologies in this organ. — Such web services in the world already exist, — complements colleague A. Rajabov. — The uniqueness of our system is that it is “sharpened” primarily for the population of Belarus. Representatives of different peoples differ from each other anatomically — depending on the conditions of the environment, climate, genetics and other factors. Variability of the shape of the lungs is also present. All this determines the degree of stability or, conversely, the predisposition of the body to the development of certain diseases. Then, taking into account these factors, a statistical analysis is performed.
Therefore, IPI scientists used a closed array of images of light from domestic medical institutions to create a software complex - about 10 million images. This made it possible to adapt the model of artificial intelligence to the peculiarities of the population of Belarus. It already works in test mode on the server of the institute and is available via the Internet. Any user can visit the site lungs.org.by and download a chest X-ray image. In a few minutes, the system will give an answer on the screen about the results of the study. It provides the so-called second opinion to minimize the gaps and repeat results. The final conclusion in any case is formed by a radiologist.
— The nodules in the lungs can be benign and malignant. Therefore, the advantage of our software complex is also that it assesses the likelihood of malignancy found in a light formation. Kosarev. - And the final diagnosis will be given additional tests of the patient, for example, the results of the biopsy. On X-ray images, the IPI web service allows you to detect manifestations of tuberculosis, bronchitis, pneumonia, localization of focal changes (such as “matte glass”). In the images of computed tomography, AI finds signs of nodular clusters of infiltrates, fibrosis, telectase, pleurisy, pneumothorax and caverns.
— How will the introduction of your development in the practice of domestic health care affect the quality of medical care in our country?
— The exact diagnosis of lung diseases strongly depends on the time that the doctor can devote to the study of the picture of a particular patient, — explains A. Rajabov. — His practical experience also plays an important role. But it is unlikely that Belarus has a doctor who analyzed 10 million pulmonary images. But the application on such an extensive database of national data can very quickly compare them with a picture of a particular patient and quickly identify the pathology. Therefore, our AI models built into applications are designed to simplify the work of doctors.
According to A. Kosareva, the advantage of the web service is also the streaming of data research. The specialist can immediately send to the server all the pictures that he needs to study, and after a certain time to pick out the results. According to the head of the laboratory for the analysis of biomedical imaging of the IIPI of the NAS of Belarus, Candidate of Technical Sciences Eduard Snezhko, elements of this software complex began to be created at the institute about 15 years ago. Now its processing is taking place, timed to a large update of the entire information system in the republic. Modernization began with a basic data storage and transmission system, security at a modern level. Further, automated workplaces of doctors of different profiles, including radiologists, will be created. Medical facilities will connect to the IPI NAS server. Doctors will be able to transmit images directly from their office to it, and then get the results of the study of images by artificial intelligence on the computer. — In the development of the web service we are moving in two directions at once, — emphasize E. Snezhko. The first is for the people of Belarus. The second is for mutually beneficial cooperation with representatives of other countries. Therefore, we expand the database of international data on chest images. Across the covers in the lungs received a set of images from Georgia. It is one of the few states where caverns are extracted from the lungs through surgery. And we try to match the manifestations of such caverns in the images of the CT with what their surgeons discover. Also a few years ago, a lot of testing of our software complex was carried out by colleagues from Tatarstan — technical specialists and radiologists. From this region for about six months, we were sent chest images, and then evaluated the results obtained by our system. Now we have joined the international community to combat tuberculosis and are receiving data from computed tomography from them.
— Such cooperation allows you to establish links with clinics and universities of different countries, — explains A. Kosarev. — It brings projects with financing of the Belarusian Republican Foundation for Basic Research, the State Committee for Science and Technology and other large organizations.
E. Snezhko notes that while the interface of the web service in English. On the IPI server, the software complex is available on the Internet to everyone from anywhere in the world in academic mode. It is more convenient to open it from a computer. A person comes to the site under a personal account and sees the results of only his pictures. It can pick up the results of the analysis, and then remove them from the institution server. Responsibility for the security of personal data when downloading images is still assigned to the user. He has the ability to independently clean up all his personal information from the data. In a large automated system, the security of personal data will be resolved at a professional level. Reason
For his development A. Kosareva and A. Rajabov was awarded the title of laureates of the exhibition-competition "100 innovations of young scientists" 2024.
Source: agency «Minsk-News»