Ubhubhane lwe-COVID-19 lwahlasela umhlaba ekuqaleni kuka-2020 futhi kusukela lapho seluyimbangela ehamba phambili yokufa emazweni amaningana, okuhlanganisa iChina, i-USA, iSpain kanye ne-United Kingdom. Abacwaningi basebenza kakhulu ekwakheni izindlela ezisebenzayo zokuxilonga izifo ze-COVID-19, futhi abaningi babo bagxile ekutheni ubuhlakani bokwenziwa (AI) bungasetshenziswa kanjani ngale njongo.
Ucwaningo oluningana lubike ukuthi izinhlelo ezisekelwe ku-AI zingasetshenziswa ukuthola i-COVID-19 ezithombeni ze-X-ray esifubeni ngoba lesi sifo sivame ukukhiqiza izindawo ezinobomvu namanzi emaphashini, abonakala njengamachashaza amhlophe kuma-X-ray scan. . Nakuba kuye kwahlongozwa amamodeli ahlukahlukene e-AI okuxilonga asekelwe kulesi simiso, ukuthuthukisa ukunemba kwawo, isivinini, nokusebenza kwawo kuseseqhulwini.
Manje, ithimba lososayensi eliholwa uSolwazi Gwanggil Jeon wase-Incheon National University, eKorea, lenze uhlaka oluzenzakalelayo lokuhlonza i-COVID-19 oluguqula izinto zibe inothi ngokuhlanganisa amasu amabili anamandla asekelwe ku-AI. Uhlelo lwabo lungaqeqeshelwa ukuhlukanisa ngokunembile phakathi kwezithombe ze-X-ray zesifuba zeziguli ze-COVID-19 kwezingezona i-COVID-19. Iphepha labo lenziwa latholakala ku-inthanethi ngo-Okthoba 27, 2021, futhi lashicilelwa ngoNovemba 21, 2021, ku-Volume 8, Issue 21 ye-IEEE Internet of Things Journal.
Ama-algorithms amabili abacwaningi abawasebenzisile bekuyi-Faster R-CNN ne-ResNet-101. Eyokuqala imodeli esekelwe ekufundeni komshini esebenzisa inethiwekhi yesiphakamiso sesifunda, engaqeqeshelwa ukukhomba izifunda ezifanele esithombeni sokufaka. Eyesibili inethiwekhi ye-neural yokufunda ejulile ehlanganisa izendlalelo eziyi-101, eyasetshenziswa njengomgogodla. I-ResNet-101, uma iqeqeshwa ngedatha eyanele yokufaka, iyimodeli enamandla yokuqashelwa kwesithombe. “Ngokolwazi lwethu, indlela yethu ingeyokuqala yokuhlanganisa i-ResNet-101 kanye ne-Faster R-CNN ukuze kutholakale i-COVID-19,” kuphawula uProf. Jeon, “Ngemva kokuqeqesha imodeli yethu ngezithombe ze-X-ray ezingu-8800, sathola ukunemba okumangazayo okungu-98%.
Ithimba labacwaningi likholelwa ukuthi isu labo lingaba usizo ekutholakaleni kusenesikhathi kwe-COVID-19 ezibhedlela nasezikhungweni zezempilo zomphakathi. Ukusebenzisa amasu okuxilonga okuzenzakalelayo okusekelwe kubuchwepheshe be-AI kungase kususe umsebenzi kanye nengcindezi kuma-radiologists nabanye ochwepheshe bezokwelapha, abebebhekene nomsebenzi omkhulu kusukela kwaqala lolu bhubhane. Ngaphezu kwalokho, njengoba izinsiza zezokwelapha zesimanje zixhunywa ku-inthanethi, kuzokwazi ukunikeza idatha yokuqeqeshwa eningi kumodeli ehlongozwayo; lokhu kuzophumela ekunembeni okuphakeme nakakhulu, hhayi nje ku-COVID-19, njengoba uProf. Jeon esho: “Indlela yokufunda ejulile esetshenziswe ocwaningweni lwethu iyasebenza kwezinye izinhlobo zezithombe zezokwelapha futhi ingasetshenziswa ukuxilonga izifo ezihlukahlukene.”
LOKHO ONGAKUTHATHE KULESI SIHLOKO:
- Several studies have reported that AI-based systems can be used to detect COVID-19 in chest X-ray images because the disease tends to produce areas with pus and water in the lungs, which show up as white spots in the X-ray scans.
- The first one is a machine learning-based model that uses a region-proposal network, which can be trained to identify the relevant regions in an input image.
- Moreover, as more modern medical devices become connected to the Internet, it will be possible to feed vast amounts of training data to the proposed model.
