I-PathaI iya kubonisa inkqubela phambili yamva nje ekusebenziseni i-AI-based pathology

I-PathaI, inkokeli yehlabathi kwitekhnoloji esekweubukrelekrele bokwenziwa (AI) ngenxa ye-pathology, namhlanje ubhengeze ukuba uphando lwakutsha nje lombutho luya kuboniswa kwiNtlanganiso yesibindi ye-2022 AASLD 2022 *, eyenzeka eWashington, DC ngoNovemba 4-8, 2022.

Kumsitho walo nyaka, i-PathAI iza kwabelana ngeenkcazo-ntetho ezintlanu zizonke, iinkcazo-ntetho ezine zepowusta kunye nentetho yomlomo enye, ezine kuzo zaphuhliswa kunye namahlakani ezonyango. Ngokukodwa, iziphumo ezitsha ze-nonalcoholic steatohepatitis (NASH) zibalaselisa indlela i-PathAI iqhubeka ngayo nokujongana nesidingo sesixhobo sokufumana amanqaku e-Clinical Research Network (CRN) esikwaziyo ukuphinda-phinda kwaye sichane ukuhlangabezana nemigangatho yeemvavanyo zeklinikhi zanamhlanje ze-NASH, kunye nemfuno engaphezulu. imilinganiselo yobungakanani bedatha eyomeleleyo.

“Uphando lwamva nje lweqela lethu lujolise ekuphononongeni umahluko okhoyo kuphononongo lwe-pathological ye-biopsies yesibindi. Iziphumo zethu zibonisa ukuba ukudibanisa izisombululo ze-AI kule nkqubo kunokuphucula kakhulu ukuhambelana kunye nokuchaneka kolu vavanyo, "kusho uGqr. UMike Montalto, iGosa eliyiNtloko lezeNzululwazi lePathaI. "Le ndlela iya kwandiswa ngokwemvelo ekuphuculeni uphuhliso lwamachiza e-NASH, ebaluleke kakhulu ekujonganeni nesidingo esikhulu sonyango esingafezekiswanga kwizigulana."

Kwintetho yomlomo ephuhliswe ngentsebenziswano neSayensi yaseGiliyadi,  "Uphononongo lokuphonononga i-histology ye-NASH kusetyenziswa amanqaku e-CRN athathwe kwindlela yokufunda yomatshini enemibala emininzi", I-PathAI iya kugxininisa umatshini omtsha wokufunda (ML) -imodeli yokulinganisa amanqaku, esebenzisa ulwazi oludibeneyo oluvela kwi-H & E kunye ne-Masson imifanekiso ye-tri-color ukuqikelela amabanga / izigaba ze-NASH CRN. Inkqubo yangoku yokuvavanya i-NASH biopsies ixhomekeke ekuhlukeni okuphezulu; Olu phando lubonisa indlela i-PathAI ML enokudibanisa ngayo ulwazi lwe-histological ukusuka kwimifanekiso emininzi yesilayidi ukuqikelela amabanga / izigaba ze-NASH CRN kwisikali esiqhubekayo, okunokunciphisa ukuhluka phakathi kwesampuli yezicubu ezivavanyiweyo kwibala ngalinye. Ngaphezu koko, amanqaku aqhubekayo e-CRN akhutshwe kwezi ziqikelelo athathwe kwigalelo elifanayo; ngolo hlobo, imodeli yePathAI ivumela uthelekiso oluthe ngqo lweempawu ezinxulumene nazo zonke iimpawu ezine zembali ye-NASH,

Ukubonisa ngakumbi ukusebenza kwe-PathAI eqhubekayo yokufumana amanqaku e-ML, i-PathAI iya kubonisa idatha ephuhliswe ngokubambisana ne-Novo Nordisk ebonisa ukuba uvavanyo lwe-histology yesibindi lwaluhlala luhambelana ngokubanzi kwii-pathologists kunye novavanyo lwe-ML kwizigulane ezine-NASH cirrhosis. "Ukuthelekiswa kweempembelelo zeSemaglutide kwi-Histology yesibindi kwizigulane ezine-Nonalcoholic Steatohepatitis Cirrhosis phakathi koVavanyo lweModeli yokuFunda kunye noVavanyo lwePathologist,"  ibonisa ukuba uhlalutyo lwe-ML lwe-PathAI lufumene abaphenduli be-placebo abambalwa kakhulu kunamanqaku e-pathologists, ukuxhasa uqwalaselo oluxelwe ngaphambili kunye nokubonelela ngobungqina obungakumbi bokuba iindlela ze-ML zinokubamba ngokuchanekileyo impendulo yonyango kwizilingo zeklinikhi kwi-NASH.

I-PathaI iyaqhubeka nokuqhubela phambili uphando lwe-NASH ukuxhasa imfuno yexesha elide yovavanyo olungakumbi lwe-granular ye-biopsies yesibindi. Ukubonisa indlela entsha kwi ” I-multimodal anisotropy imaging yenza uqikelelo lokufunda koomatshini lwenqanaba le-CRN NASH fibrosis ngaphandle kwenkcazo yesandla. ", I-PathaI ijongana nokwahluka kwi-pathological staging ye-fibrosis kwi-NASH isebenzisa i-quantitative multi-model anisotropy imaging (QMAI) ye-fibrosis ukubonelela ngeenkcazo ezingakhethi cala ukuqeqesha iimodeli ukuqikelela inqanaba le-NASH CRN fibrosis kwamanye amacandelo ezicubu. Ngokuthelekisa imodeli yokusebenza kunye nenqaku le-pathologist, le mizekelo iqikelele inqanaba le-CRN fibrosis ngokuchaneka ngokuthelekiswa neemodeli eziqeqeshwe yi-pathologist annotation yamacandelo e-MT-stained tissue.

BlogInnovazione.it

​