Article
Sonohistology - ultrasound and artificial intelligence for the determination of salivary-gland-tumors?
Search Medline for
Authors
Published: | September 22, 2005 |
---|
Outline
Text
Despite B-mode-ultrasound, colour-duplex-sonography and the use of ultrasound-contrast-agents are developing, to determine the diagnosis of salivary gland tumors is still not possible. So far, the diagnosis is based on a regulary histological examination after parotidectomy.
In Sonohistology, ultrasound baseband data is acquired and analysed for characteristical findings. Therefore, a software based classification system was developed, which is able to assign histological tissue parameters to the baseband data as well as to recognice them in further examinations.
In the context of a preliminary inquiry, the diagnosis of salivary gland tumors was predicted by B-mode ultrasound in 17 cases and baseband data were acquired. All patients underwent parotidectomy and specimens of the tumors were histologically examinated. The software was trained with the baseband data and the histological findings of 12 cases. Afterwards the software was used to predict the diagnosis in 5 additional cases and was reviewed by the histological findings as a gold standart. The diagnosis of the classification system matched in 80% (4/5) of the cases with the histological findings.
In our opinion, sonohistology could be a proper method to establish the diagnosis of salivary gland tumors. A follow up with more cases is planned.