Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

AI in oral cancer diagnosis

Yoshizawa K, Ando H, Kimura Y et al. Automatic discrimination of Yamamoto-Kohama classification by machine learning approach for invasive pattern of oral squamous cell carcinoma using digital microscopic images: a retrospective study. Oral Surg Oral Med Oral Pathol Oral Radiol 2022; 133: 441-452.

Possible greater consistency.

Oral squamous cell carcinoma (OSCC) is invasive, leading to metastases and poor survival prognosis. The Yamamoto-Kohama (YK) classification differentiates the histopathological appearance of OSCC into grades 1 to 4D, the higher grades being the most invasive.

Here, historic (1989 to 2009) stained specimens of OSCC (n = 101) were graded according to the YK system. Using a two-stage machine learning approach - identifying a region of interest (the invasive front of the lesion) and evaluating the mode of invasion - results showed no significant differences in classification accuracy between clinician and machine (except for YK grade 2 cases, the reasons for which discrepancies are discussed).

It is known that visual examination of specimens leads to significant differences in determinations between examiners and the use of AI may introduce greater consistency. AI may also be useful in countries where there are few pathologists. However, where 'decision making and prediction' are left to the machine, caution is needed.

Author information

Authors and Affiliations

Authors

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Hellyer, P. AI in oral cancer diagnosis. Br Dent J 232, 629 (2022). https://doi.org/10.1038/s41415-022-4275-x

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41415-022-4275-x

Search

Quick links