RussianBrain tumors make up quite a large percentage of total cancer incidence; however, a few pathologists have required focused specialization and are able to perform a qualified histological diagnosis.
Decision support system we present is designed for general pathologists. It is intended to be used in cases when an urgent impression on kind of neoplasm is required while getting an urgent consultation from the focused specialization pathologist is impossible. In such a situation, any additional information on tumor may be useful for diagnosis.
Here is a brief description of the system functioning. Detailed instructions are in the corresponding page of user account.
During the work with system user uploads patient data and gets the system response. The following information should be uploaded:
-
Clinical data on patient;
-
Information on biopsy material properties;
-
Images (photos from microscope) of histologic cuts.
After all data is uploaded, user makes request. System gives response – to what kind of tumors does uploaded data belong most likely.
NB! System makes choice from the kinds of tumors it was trained by only. This means that independently of data origin system response will contain one of “known” by system tumors. For example in present system is trained by astrocytic tumors and healthy (not tumor) tissue. If, say, meningioma images will be uploaded, system’s response will point to one of astrocytic tumors or healthy tissue.
All uploaded patients data are being kept on the server as a database. The only exclusion are histological cuts images due to their excess “weight”. Database search is possible.