APPLICATION OF ARTIFICIAL INTELLIGENCE IN RADIOLOGY

Umjetna inteligencija je grana računarske znanosti koja se bavi proučavanjem i izgradnjom računalnih sustava koji pokazuju neki oblik inteligencije. Takvi sustavi se dizajniraju da mogu učiti, donositi zaključke te obavljati druge vrste vještina kojima je potrebna i zahtijevaju ljudsku inteligenciju. Cilj primjene umjetne inteligencije u medicini, odnosno u radiologiji je povećanje dijagnostičke i terapijske točnosti. Neke od grana umjetne inteligencije koje imaju primjenu u radiologiji su: ekspertni sustavi, strojno učenje i duboko učenje te neuronske mreže. Metode umjetne inteligencije služe kao potpora radiologu pri dijagnosticiranju i interpretaciji nalaza te olakšavaju radiologu posao tako što smanjuju vrijeme potrebno za dijagnostiku. Podatci dobiveni u radu su iz stručnih članaka i raznih istraživanja provedenim u bolnicama, klinikama i sličnim ustanovama te rezultata dobivenih u ne kliničkim uvjetima. Ustanovljeno je da za dobivanje kvalitetnijih umjetnih neuronskih mreža i njihov bolji rad je potrebno da su uvjeti snimanja i protokoli što sličniji, vanjske karakteristike pacijenata što sličnije, edukacija osoblja da pravilno koriste programe umjetne inteligencije te računalna oprema da je što kvalitetnija i modernija. Također je ustanovljeno da umjetna inteligencija i umjetne neuronske mreže imaju veliki potencijal u njihovoj primjeni u radiologiji, olakšavaju rad radiologa te postaju sve sigurnije i točnije poboljšanjem tehnologije. Njihova ograničenja su u još ne sigurnosti u radu, strah radiologa i pacijenata u njihovu efikasnost i rad te problem kod prikupljanja podataka za treniranje neuronskih mreži zbog moralnih i etičkih razloga.Artificial intelligence is a branch of computer science that deals with the study and construction of computer systems that show some form of intelligence. Such systems are designed to be able to learn, draw conclusions and perform other tasks that require and demand human intelligence. The goal of the application of artificial intelligence in medicine, particulary in radiology is to increase diagnostic and therapeutic accuracy. Some of the branches of artificial intelligence that have application use in radiology are: expert systems, machine learning and deep learning as well as neural networks. Artificial intelligence methods serve to support the radiologist in diagnosing and interpreting the findings as well as making the radiologist´s job easier by reducing the time required for diagnosis. The data obtained in this paper are from scientific articles - various studies conducted in hospitals, clinics and similar institutions as well as results obtained in non-clinical conditions. It has been established that in order to obtain better quality artificial neural networks, it is necessary that the imaging conditions and protocols are as similar as possible, the external characteristics of patients are as similar as possible, staff training to use artificial intelligence programs and that computer equipment is modern and has high quality. It is also established that artificial intelligence and artificial neural networks have great potential in their application in radiology, facilitate the work of radiologists as well as becoming safer and more accurate with the improvement of technology. Limitations of artificial intelligence are that they are not yet fully certain while working, fear of radiologists and patients in their efficiency and operation, and the problem of collecting data for neural network training because of moral and ethical reasons