Abaphenyi baseJapan basungule uhlelo lobuciko bokufakelwa olunweba i-endoscope ngokukhulisa izikhathi ezingama-500 emathunjini esiguli. Uhlelo lobuchwepheshe bokufakelwa lungakhomba ukuthi ngabe lukhona yini ushintsho olubi ku-polyp enkulu yamathumbu ku-endoscope kungakapheli imizuzwana engu-0.3, ngokuya ngemiphumela yesikhathi sangempela sokwahlulela Udokotela anganquma ukuthi uzosebenza ngesikhathi sangempela yini.
Compared with the past, it takes a week to make a diagnosis, and now the system can immediately determine whether to remove it, which greatly improves the efficiency of diagnosis and treatment. During the development of this system, more than 60,000 tumor cell pictures were used to build a database. These pictures came from more than 3,000 patients with colorectal cancer diagnosed in 5 hospitals in Japan. By analyzing and deep learning the tumor images in the image database, the system has learned the automatic recognition function of cancer. Not only improve the diagnosis efficiency, but also improve the accuracy.
EJapan, umdlavuza colorectal iyithumba lesibili elibi kakhulu ngemva kokufa ngenxa yomdlavuza wamaphaphu. Ukutholwa kusenesikhathi kuyisihluthulelo sokuthuthukisa izinga lokwelashwa. Le mpumelelo yobuhlakani bokwenziwa e-Japan ingathola ubukhona bomdlavuza kuma-polyp wamathumbu amakhulu esikhathini esingaphansi kwesekhondi. Njengamanje, lolu hlelo lokuhlonza umdlavuza wokufakelwa komdlavuza ohloliwe seluhlolwe emtholampilo ezibhedlela eziyi-6 eJapan, futhi kulindeleke ukuthi luthole ilayisense kuziphathimandla ezifanele ezilawula ezemithi zaseJapan ngo-2018.