Brit J Cancer: Diagnosis of head and neck cancer based on artificial intelligence

Recently, a research article was published in the British Journal of Cancer, an authoritative journal in the field of oncology.
This systematic review aims to review the use of artificial intelligence/machine learning (AI/ML) methods to automatically analyze images for diagnosis and evaluation of head and neck cancer (HNC) Research.

The researchers searched electronic databases such as MEDLINE, OVID, EMBASE, and Google Scholar to determine the literature on AI/ML for diagnostic evaluation of HNC (2009-2020).
Researchers have no restrictions on the AI/ML methods or imaging methods used when searching.

The analysis identified a total of 32 documents.
HNC sites include oral cavity (n=16), nasopharynx (n=3), oropharynx (n=3), larynx (n=2), salivary glands (n=2), sinuses (n=1), and multiple sites (N=5).
Imaging methods include histology (n=9), imaging (n=8), hyperspectral (n=6), endoscopy/clinical (n=5), infrared thermal imaging (n=1) and optical imaging ( n=1).

Two studies used clinicopathological/genomic data.
Twenty-two studies (69%) used traditional machine learning methods for evaluation, eight studies (25%) used deep learning (DL) for evaluation, and two studies (6%) used both methods for diagnostic evaluation.

It can be seen that more and more researches have explored the role of AI/ML in the use of multiple imaging methods to assist HNC detection.
These methods can obtain high accuracy, which can surpass the human judgment ability in data prediction .
A large-scale multi-center prospective study is needed to promote its application in clinical practice.
Original source:

Hanya Mahmood.