Ukulandelela Isikhumba Ngemishini: Indlela i-AI Eshintsha Ukuhlola Kwesikhathi Esifushane

Ubuhlakani bokwenziwa obusetshenziselwa ukuhlola isikhumba busekelwe emkhakheni wezifundo zemishini obizwa ngokuthi ukufunda okujulile, ikakhulukazi amanethiwekhi e-neural convolutional aqeqeshwe ukuze aqaphele izimo ezithile ezitholakala ezithombeni. Lezi zinhlelo zifunda ngokucubungula izinkulungwane zezithombe ezibhalwe phansi, zikhulisa ikhono lokuhlonza izici ezibukwayo ezihlobene nezimo ezahlukene zesikhumba. Inqubo yokufundisa ibandakanya ukuhlinzeka i-algorithm ngezithombe ezihlonziwe ngodokotela bezikhumba abanolwazi futhi ziqinisekiswe nge-biopsy lapho kufanele.

Ngemva kwezikhathi eziningi, inethiwekhi ifunda ukuhlole izimo ezincane zokwabiwa kombala, ubukhulu, izici zomngcele, izici zokwakheka, kanye nezinye izakhiwo ezibukwayo ezihlobene nezibalo ezithile. Umphumela uyi-system engahlola isithombe esisha futhi inikeze ukuhlola kweprobability kwezimo ezahlukene ezingenzeka. Izinhlelo zokuhlola isikhumba ze-AI zanamuhla azisebenzisi kuphela uhlelo lwe-ABCDE rule ngokuqondile.

Zihlonza izimo eziyinkimbinkimbi kakhulu nezincane kunezindlela ezisekelwe ohlwini. Ezinye izici i-AI ezibheka kakhulu zingase zingabi namagama ezobuchwepheshe ngoba zimelela ubudlelwano bezibalo phakathi kwezinga le-pixel elinentshisekelo kwezibalo kodwa ezingabonakali kumuntu. Izinhlelo ze-AI zamanje zokuhlola izikhumba zifeze amazinga okuphumelela afana nodokotela bezikhumba abaqinisekisiwe ezimeni zocwaningo ezilawulwayo, nakuba ukusebenza kwangempela kuhlanganisa nezinye izinto ezithinta ukwethembeka.!!

Lezi zinhlelo zisebenza kahle ekucubunguleni inani elikhulu lezithombe ngokushesha nangokuphindaphindiwe, ngaphandle kokukhathala noma izinkolelo eziphambene ezithinta abahloli abantu. Nokho, ukuqonda ukuthi ukuhlola kwe-AI kusebenza kanjani kusho futhi ukuqonda imikhawulo yakhona. Lezi zinhlelo zikh produce izilinganiso ze-probabilistic, hhayi izinqumo eziphelele.

Ziqeqeshwe kumadathasethi athile angase angamehluko wonke amazinyo esikhumba, izinhlobo ze-lesion, noma izimo zezithombe ngokulinganayo. Ikhwalithi yesithombe esingena ithonya kakhulu ikhwalithi yokuhlola ephumayo, okwenza proper photography technique kubaluleke kakhulu.

Phakathi kwezikhungo ze-dermatology, i-AI ifakwa njengensiza yokusekela izinqumo ethuthukisa hhayi esikhundleni sokuhlola kwezokwelapha. Lezi zinhlelo zoclinical ngokuvamile zisebenza ngezithombe ze-dermoscopic ezithathwe kusetshenziswa imishini yezokwelapha ejwayelekile, inikeza izinga lekhwalithi yesithombe nokuhambisana okusekela ukunemba okuphezulu kokuhlola. Izinhlelo eziningi ze-AI zoclinical zithole imvume yokulawula yokusetshenziswa njengamathuluzi okuxilonga.

Lezi zinhlelo zihlola izithombe ze-dermoscopic ngesikhathi sangempela ngesikhathi sokuhlola iziguli futhi ziqhakambisa izilonda ezikhombisa izici ezihlobene ne-malignancy. Udokotela wezikhumba ubona ukuhlola kwe-AI eduze kokuhlola kwakhe, kudala ukuhlola okunezindlela ezimbili okungase kubambe izithombe ezithile ezitholakala ukuthi kungase kube khona okuthile okungatholakala kuphela ngumuntu noma umshini. Izinkundla zokuthwebula umzimba wonke ezithuthukiswe nge-AI zimelela enye indlela yokusebenza kwezokwelapha.

Lezi zinhlelo zithwebula ubuso besikhumba somguli wonke zisebenzisa isethi ye-multi-camera ejwayelekile, bese zisebenzisa ama-algorithms ukuze zihlukanise, zifake, futhi zilandela zonke izilonda ezibonakalayo. Ezivakashi ezilandelayo, uhlelo luhlola ngokuzenzakalelayo izithombe ezintsha uma kuqhathaniswa ne-baseline, kukhombisa izilonda ezishintshile ngosayizi, isimo, noma umbala futhi kuqhakambisa izilonda ezintsha ezingazange zibe khona ngesikhathi sokuvakasha kwangaphambilini. Le miphumela yokushintsha okuzenzakalelayo ibhekana nenkinga enkulu kakhulu ku-dermatology: ukuthola ngokwethembeka izinguquko ezincane phakathi kwezinduku ezingaba yizinkulungwane kumguli oyedwa.

Ngisho nodokotela wezikhumba onolwazi oluningi akakwazi ukukhumbula ukubukeka okukhululekile kwezinduku zonke phakathi kwezivakashi zonyaka, okwenza ukuqapha okuhlelekile okwenziwa nge-AI kube ukuthuthuka okungokoqobo emandleni ezokwelapha. Ucwaningo luqhubeka nokwandisa izinhlelo ze-AI ku-dermatology ngaphezu kokuthola i-melanoma. Ama-algorithms athuthukiswa futhi ahlaziywa ukuze ahlonze i-basal cell carcinoma, i-squamous cell carcinoma, ama-actinic keratoses, kanye nezimo eziningi ezinhle. Ezinye izinhlelo zihlola izikhalazo zesikhumba ezindaweni zokunakekelwa kwempilo, zisiza odokotela abajwayelekile ukuba benze izinqumo mayelana neziguli ezidinga ukuhanjiswa kwe-dermatology ngokuphuthumayo nokuthi yiziphi ezingaqhubeka ngokuphepha.

Ukuhlaziywa kwesikhumba kwe-AI okuhlinzekwa ngezinhlelo zokusebenza ze-smartphone kwenze ukuthi ukuhlolwa okusekelwe kubuchwepheshe kube lula ukufinyelela kumuntu onefoni kanye nokuxhumeka kwe-inthanethi. Lezi zinhlelo zokusebenza zivumela abasebenzisi ukuba bathathe isithombe se-lesion yesikhumba bese bethola ukuhlola okuqhamukayo kwe-AI phakathi kwemizuzwana, okuvumela ukufinyelela kuhlobo lokuhlola okuqalayo. Ukubukeka kulula.

Akukho muntu onokufinyelela okulula kumdokotela wezikhumba, kungaba ngenxa yokuhlukaniswa kwendawo, izindleko, isikhathi sokulinda, noma ukuqwashisa. Uhlelo lokusebenza lwe-smartphone olungakwazi ukukhomba ama-lesions angase abe yinkinga futhi lukhuthaze abasebenzisi ukuba baphendukele ekuhloleni okuchwepheshile lugcwalisa ibhodi elibalulekile, ikakhulukazi ezindaweni ezingaphansi lapho odokotela bezikhumba bengatholakali. Nokho, ukuhlolwa okusekelwe ku-smartphone kuza nezixwayiso ezibalulekile.

Ikhwalithi yesithombe evela kumakhamera efoni yokuthenga iyahluka kakhulu kuye ngedivayisi, izimo zokukhanya, ukujula kwesandla, kanye nokuphuma kulesion. Izinhlelo ze-AI eziqeqeshwe ngezithombe ezisezingeni eliphezulu ze-dermoscopic zingase zisebenze ngendlela ehlukile uma zihlola izithombe ezithathwe ngama-smartphones okuthenga ngaphansi kwezimo zangempela ezihlukahlukene.!! Abasebenzisi kumele baqonde ukuthi lezi zinsiza zinikeza ukuhlola kokuqala, hhayi iziguli.

Izifundo zokunembile eziphathelene nezinhlelo zokusebenza zokuhlaziya isikhumba zikhombisile imiphumela ehlukahlukene. Ezinye izinhlelo zokusebenza zisebenza kahle ekuhloleni okulawulwayo, kuyilapho ezinye zikhombisile amazinga akhathazayo okungamanga lapho ama-lesions angase abe yingozi ahlukaniswe njengokuphansi. Ukuhlolwa kokulawula lezi zinhlelo zokusebenza kuhluka ngokwezwe, futhi hhayi zonke izinhlelo zokusebenza ezithengiswayo zidlule ekuqinisekisweni okuqinile kwezokwelapha.

Izinhlelo zokusebenza zokuhlaziya isikhumba eziphusile kakhulu ze-smartphone zikhomba njengezinsiza zokuhlola ezisiza abasebenzisi ukuthi baphendule ukuthi kufanele baphendukele ekuhloleni okuchwepheshile, hhayi njengezikh replacements zeziguli. Zikhuthaza abasebenzisi abathola ukuhlola okukhathazayo ukuba bahlele izikhathi zokuhlola ezithile, futhi eziningi zifaka okuqukethwe kwemfundo mayelana nokuhlola ngokwakho nezimpawu zokuxwayisa.

Umqondo obaluleke kakhulu ekuqondeni indima ye-AI empilweni yesikhumba ukuthi lezi zinhlelo zenzelwe ukuhlanganisa ubuchwepheshe bomuntu hhayi ukufaka esikhundleni salo. Le mihluko ayiyona nje eyokuxhumana kodwa ibonisa imikhawulo yezehlakalo nezobuchwepheshe ezidinga ukuhlinzekwa kwabantu. Izinhlelo ze-AI azinakho ukukwazi ukuthatha umlando wesiguli, ubuze ngokuqhubeka kwezimpawu, ubuze ngembali yomndeni, uthinte i-lesion ukuze uhlole ubuchwepheshe nobukhulu, noma uhlanganise umongo ophelele wezokwelapha ohlola ukuhlola komdokotela wezikhumba.

Isithombe sithatha ulwazi lwezithombe kuphela, ngenkathi ukuhlola okuphelele kwezokwelapha kuhlanganisa ulwazi lwezitho, umlando, nolwazi lwezinhlelo okungatholakali kunoma iyiphi i-AI esekelwe ezithombeni. Odokotela bezikhumba baphatha ukucabanga okujwayelekile okungakwazi ukukopishwa yizinhlelo ze-AI zamanje. Baqonda ukuthi i-lesion esolweni lesiguli esidala inezincazelo ezihlukile kune-lesion efanayo ebhajwe emzimbeni wentsha.

Bazi ukuthi uma umlando wemithi wesiguli, isimo somzimba, noma imvelaphi ye-genetic ishintsha amathuba eziguli ezihlukahlukene. Babona izimpawu zokulimala kwelanga, izimo zokuguga, nezinye izinkomba ezitholakala phezu kwesikhumba esibanzi. Imodeli efanele ibeka i-AI njengomsebenzi onamakhono aphezulu ophucula ukunakekelwa kwezokwelapha ezingeni eliningi.

Kubantu, amathuluzi okuzihlola asekelwe ku-AI akhuphula ukuqwashisa futhi anikeze ukuhlolwa kokuqala phakathi nezivakashi ezichwepheshile. Kubadokotela bezinhlaka eziphakathi, izinhlelo zokuhlola ze-AI zisiza ukuthola iziguli ezidinga ukuhanjiswa okuphuthumayo kumachwephesha. Kubodokotela bezikhumba, izinhlelo zokwesekwa kwezinqumo ze-AI zinikeza umbono wesibili ongase ubambe izithombe ezincane futhi uthuthukise ukuvumelana kokuhlola.

Le modeli yokubambisana ibonise imiphumela ekhuthazayo ezifundweni lapho ukuhlanganiswa kwedokotela bezikhumba ne-AI kudlule kokubili odokotela bezikhumba bodwa noma i-AI yodwa. Amandla ahlangene okucabanga kwezokwelapha kwabantu kanye nokuhlonza kwezinhlaka kudala uhlelo lokuthola oluqinile kunoma iyiphi indlela esebenza ngokuzimela.

Umkhondo we-AI empilweni yesikhumba ukhomba ezinhlelweni ezihlanganisiwe, ezitholakalayo, nezichanekile ezizoshintsha indlela izimo zesikhumba ezitholwa futhi zihlolwe eminyakeni ezayo. Ukuthuthukiswa okukhulayo kukhombisa indlela le teknoloji ethuthuka ngayo. Izinhlelo ze-AI eziningi ezihlanganisayo ezihlola hhayi kuphela izithombe kodwa futhi nezimpawu ezibikiwe ngababhalisi, umlando wezokwelapha, izici zokuphila, kanye nedatha yemvelo zisebenza.

Lezi zinhlelo eziphelele zihlose ukuhlinzeka ngokuqonda okuphakeme okuhamba phambili kokuhlola okwenziwa udokotela wezikhumba, kuhlanganisa imithombo eminingi yedatha esikhundleni sokuncika kuphela ekuhloleni kwezithombe. Izindlela zokufunda ezihlanganisiwe zingase zixazulule enye yezinkinga zamanje zokuhlaziya isikhumba se-AI: ubias bedatha. Ngokuqeqesha ama-algorithms phakathi kwezithameli ezihlukahlukene ngaphandle kokuhlanganisa izithombe ezibucayi zezokwelapha, ukufunda okuhlanganisiwe kungakhiqiza amamodeli anembile kakhulu asebenza ngokwethembeka kuwo wonke ama-tones nesigaba sesikhumba.

Izinhlelo zamanje ngokuvamile zihluleka ekusebenzeni kahle kumatoni esikhumba esinsundu ngenxa yedatha yokuqeqesha ehluka kakhulu kumaphuzu esikhumba esikhanyayo, futhi ukuxazulula le nkinga kubalulekile ocwaningweni. Ubuchwepheshe obugqokekayo bungase bukhuthaze ukuqapha okungapheli noma okungapheli kwesikhumba esikhathini esizayo. Amadivayisi anekhono lokuthwebula izithombe angase athathe futhi ahlaziye izikhumba ngezikhathi ezithile, athole ngokuzenzakalelayo izinguquko futhi axwayise abasebenzisi ngezinto ezidinga ukunakwa.

Le shintsho ukusuka ekuhloleni ngokwakho okwenziwa ngezikhathi ezithile kuya ekuhloleni okuqhubekayo okungapheli kungase kuthole izinguquko ezikhathazayo ezikhathini eziphakeme. Ukuhlanganiswa nezincwadi zezempilo ezikwi-inthanethi kuzovumela imiphumela yokuhlaziya isikhumba se-AI ukuba ibe yingxenye yomlando wezokwelapha wesiguli, etholakala kubo bonke abahlinzeki bezempilo. Le nxhaxheba izovumela ukunakekelwa okuhlelekile nokulandelela okuhle kwezimo zempilo yesikhumba phakathi kokuphila kwesiguli.

Izinhlelo zokulawula ziyaqhubeka nokuthuthuka ukuze zihambisane nobuchwepheshe. Njengoba amathuluzi okuhlaziya isikhumba se-AI ethuthuka futhi esetshenziswa kabanzi, izindinganiso ezicacile zokuvumelanisa, ukuhlinzeka ngokuqonda, nokubekwa kwezokwelapha kuzokwenza kube lula kokubili abathengi nabahlinzeki bezempilo ukuqonda indima efanele yalezi zinsiza ngaphakathi kwesistimu enkulu yezempilo.