Ƙwararrun Ƙwararrun Ƙwararru na iya Taimakawa a cikin Yaƙin COVID-19

Cutar ta COVID-19 ta mamaye duniya da guguwa a farkon shekarar 2020 kuma tun daga lokacin ta zama sanadin mutuwar mutane a kasashe da dama, ciki har da China, Amurka, Spain, da Ingila. Masu bincike suna aiki tuƙuru kan haɓaka hanyoyin amfani don gano cututtukan COVID-19, kuma da yawa daga cikinsu sun mai da hankali kan yadda za a iya amfani da hankali na wucin gadi (AI) don wannan dalili.       

Yawancin bincike sun ba da rahoton cewa ana iya amfani da tsarin tushen AI don gano COVID-19 a cikin hotunan X-ray na ƙirji saboda cutar tana ƙoƙarin samar da wuraren da ƙwayar ƙwayar cuta da ruwa a cikin huhu, waɗanda ke nunawa a matsayin fararen tabo a cikin sikanin X-ray. . Ko da yake an ba da shawarar samfuran AI daban-daban na bincike bisa wannan ka'ida, haɓaka daidaitonsu, saurin su, da kuma amfani da su ya kasance babban fifiko.

Yanzu, ƙungiyar masana kimiyya karkashin jagorancin Farfesa Gwanggil Jeon na Jami'ar Kasa ta Incheon, Koriya, sun ƙirƙiri tsarin gano cutar COVID-19 ta atomatik wanda ke mayar da al'amura gabaɗaya ta hanyar haɗa dabarun tushen AI guda biyu masu ƙarfi. Ana iya horar da tsarin su don bambanta daidai tsakanin hotunan X-ray na marasa lafiya na COVID-19 daga waɗanda ba COVID-19 ba. An samar da takardar su akan layi a ranar 27 ga Oktoba, 2021, kuma aka buga a ranar 21 ga Nuwamba, 2021, a cikin juzu'i na 8, fitowar 21 na IEEE Intanet na Abubuwan Jarida.

Algorithms biyu da masu binciken suka yi amfani da su sune Saurin R-CNN da ResNet-101. Na farko shine samfurin tushen koyo na inji wanda ke amfani da hanyar sadarwa na yanki, wanda za'a iya horar da shi don gano yankunan da suka dace a cikin hoton shigarwa. Na biyu cibiyar sadarwa ce mai zurfin ilmantarwa wacce ta ƙunshi yadudduka 101, waɗanda aka yi amfani da su azaman kashin baya. ResNet-101, lokacin da aka horar da shi da isassun bayanan shigarwa, samfuri ne mai ƙarfi don gane hoto. "A iyakar saninmu, tsarinmu shine farkon wanda zai hada ResNet-101 da Fast R-CNN don gano COVID-19," in ji Farfesa Jeon, "Bayan horar da samfurin mu da hotuna 8800 na X-ray, mun sami ingantaccen daidaito na 98%."

Ƙungiyar binciken ta yi imanin cewa dabarun su na iya tabbatar da amfani ga farkon gano COVID-19 a asibitoci da cibiyoyin kiwon lafiyar jama'a. Yin amfani da dabarun gano cutar ta atomatik dangane da fasahar AI na iya ɗaukar wasu ayyuka da matsin lamba daga likitocin rediyo da sauran ƙwararrun likitocin, waɗanda ke fuskantar manyan ayyuka tun bayan barkewar cutar. Bugu da ƙari, yayin da ƙarin na'urorin likitanci na zamani ke haɗuwa da Intanet, zai yiwu a ciyar da adadi mai yawa na bayanan horo zuwa samfurin da aka tsara; wannan zai haifar da madaidaicin daidaito, kuma ba ga COVID-19 kawai ba, kamar yadda Farfesa Jeon ya ce: "Tsarin ilmantarwa mai zurfi da aka yi amfani da shi a cikin bincikenmu ya shafi sauran nau'ikan hotunan likita kuma ana iya amfani da su don gano cututtuka daban-daban."