| Developers: | |
| Date of the premiere of the system: | October, 2018 |
| Branches: | Pharmaceutics, medicine, health care |
2018: Announcement and beginning of implementation
On October 12, 2018 the Google company provided perspective solution AI for detection of a breast cancer at assessment of a biopsy of lymph nodes. It began to be used already in the Naval medical center of San Diego which researchers also took part in development of technology.
Accuracy of detection of cancer metastases a system under the name Lymph Node Assistant (LYNA) reached 99% that exceeds indicators of specialists histologists which in the conditions of limited time reveal a tumor on the colored cut only in 62% of cases. LYNA is still incapable to distinguish extent of development of a metastasis and a type of cancer that is extremely important for treatment selection, however its capabilities are enough to bring huge benefit to specialists.
LYNA is founded on Stanford model of deep learning Inception-v3 for image understanding open source. Specialists of Google provided training of a neuronet in two selections of the colored cuts of a biopsy of lymph nodes - 399 images were provided by the Medical center of the Radbud university (Netherlands), and 108 more unique images – the Medical center of Utrecht University (Utrecht, the Netherlands). For direct training of a neuronet were used 270 of these cuts (160 normal, 110 with tumors), the others were applied to assessment.
In the LYNA tests reached the accuracy of detection of tumors of 99.3%, having precisely identified all 40 metastases without false-negative results. At the same time assessment of LYNA was not influenced by such artifacts as vials of air, bad coloring and hemorrhages.
The LYNA system is not ideal — sometimes it mistakenly took for a tumor huge cages and lymphocytes which got into fabric from an arterial bed, a gistiotsita, - however AI a system managed to estimate the same slides better, than the practicing histologists, and much quicker.
In the future researchers hope to improve specificity of this diagnostic method.[1]
Notes
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |