TUCAN (tuberculosis and lung cancer) Early diagnosis of tuberculosis and lung cancer using neuronets

In April, 2018 as CNews knew, in Russia the project startup under the working name TUCAN, the tuberculosis devoted early diagnostics and lung cancer using neuronets prepares. Project title is formed from the English words tuberculosis and lung cancer that it is translated as "tuberculosis" and "lung cancer" respectively. Further within the project diagnostics will be carried out and other diseases of respiratory organs^[1].

TUCAN provides use of artificial intelligence for the analysis of radiological researches. A neuronet it will be obligatory to participate in decision making the person — the clinical physician or the radiologist. He will make the final diagnosis, being based on the data of the computer. It allows to classify TUCAN in the sphere of artificial intelligence (Artificial Intelligence) as belonging to area of the mixed intelligence (Augmented Intelligence).

Authors of the project focus attention that responsibility for diagnosis will lie on the specialist person. The artificial intelligence in this case is useful first of all that it is capable to reveal and record formal symptoms of a disease to which the radiologist can not pay attention — for example, because of fatigue at the end of the working day.

Project team

As the head of TUCAN the chief non-staff specialist on radiodiagnosis of the Ministry of Health, the doctor of medical sciences, professor Igor Tyurin acts.

Will develop models of machine learning on the basis of deep neural networks education center "the Ideas. Technologies. Examination" (ITEF). The center participates in implementation of social and educational projects in the field of medicine since 2012 of ITEF and acted as the initiator of the project. The technical director of the center Oleg Fateev directed software development in Paragon Software Group and IBS before. Also it is a coordinator of open community of professionals in the field of Big Data and machine learning of Big Big Data Group (BBDG).

Specialists of the Central Research and Development Institute of Tuberculosis (CRDIT) who will be engaged in a marking of digital x-rays and drawing up summaries to them will take part in the project. In process data of various researches, including results of surgical intervention will be used, the director of TsNIIT, the doctor of medical sciences, professor Atadzhan Ergeshov explains.

Rights and financing

As representatives of ITEF explained CNews, equal share financing of the project in a scientific part will be performed by TsNIIT, the Russian oncological scientific center of N.N. Blochin (RONTs) and the Russian Medical Academy for Post-Graduate Education (RMAPGE), in a technical IT part — ITEF.

As for intellectual property rights, "the group of individuals" which will perform joint surveys will have them, the director of ITEF Andrey Yanychev explained CNews. At the moment the issue of right possession is handled, but the algorithm will remain in Russia, he added. Also Yanychev noted that the Civil code of Russia in its current form does not regulate patenting of models of machine learning.

In return, Fateev reported that there are no models on tuberculosis anywhere in the world yet. As for intellectual property, the Civil code accurately defines her owners in case of the conclusion of the government contract, he adds. However receiving a state task with the subsequent conclusion of the government contract is possible only after models prove the efficiency.

"We have all softwares which is in the Register of the Russian software of the Ministry of Telecom and Mass Communications, belongs regarding copyright not to the state, but the companies which are specified there. The main thing that the companies were Russian. If software is developed according to the government contract, then it gets to National fund of algorithms and programs — that is another matter, the state can use it free of charge" — Fateev sums up the result. However TUCAN will be able to get to this fund of development only after the beginning of financing of the project by structures of the Ministry of Health that will occur only after completion of clinical trials.

Libraries of pictures

As Fateev as soon as neuronets reached the sufficient level of accuracy in image understanding explains, they began to be applied in medicine to processing of x-rays. This process became more active in 2016 when in the USA the program for fight against cancer of Cancer Moonshot which budget was calculated by billions was started.

During this period collections of the depersonalized x-rays were actively uploaded publicly. In pictures the marking is made — places of emergence and explicit absence of tumors are mentioned. The most part of similar archives was brought together during national programs for scanning of a thorax regarding lung cancer. These pictures will also be used during implementation of that part of TUCAN which concerns lung cancer.

Materials on tuberculosis will be provided by TsNIIT. According to Fateev, in Russia unique experience of a research of tuberculosis is accumulated.

Notes