Japanese AI System Differentiates between Cancer Cells and Automatically Detects Their Susceptibility to Radiotherapy – Innovita Research

Japanese AI System Differentiates between Cancer Cells and Automatically Detects Their Susceptibility to Radiotherapy

Reporting in the academic journal Cancer Research, a group of Japanese researchers have devised an artificial intelligence (AI) system (called VGG16) that can identify different types of cancer cells comprising a single tumour, and successfully gauge their susceptibility to radiotherapy.

“This study demonstrates rapid and accurate identification of radioresistant tumour cells in culture using artificial intelligence; this should have applications in future preclinical cancer research,” the group stated in its paper.

Identifying specific types of cancer cells present within a cancerous lesion is of crucial importance when choosing the best approach to treatment. Such identification, however, is often a lengthy process, further complicated by human error and constraints related to human eyesight.

Based on a convolutional neural network (CNN) – a type of deep learning algorithm commonly deployed for analysing visual imagery – the new system managed to outperform humans by accurately identifying different cancer cells present in microscopic images.

AI system can not only differentiate between specific types of cancer cells, but also determine which ones are going to be sensitive to radiotherapy. Image credit: Pavlo Luchkovski via pexels.com.

After training the system on 8,000 images of cells obtained from a phase-contrast microscope, the team assessed its accuracy on a further 2,000 images, demonstrating VGG16’s ability to distinguish between cancer cells from mice and humans, as well as those resistant and sensitive to radiotherapy.

The automation and high accuracy (no less than 96%, as reported in the paper) of the system could be used for identifying the specific cancer cells present within individual tumours or circulating in the blood:

“For example, knowing whether or not radioresistant cells are present is vital when deciding whether radiotherapy would be effective, and the same approach can then be applied after treatment to see whether it has had the desired effect,” said lead researcher Masayasu Toratani.

Given the success of the study, Masayasu and colleagues now plan to continue working on their new algorithm with the view to eventually develop a system capable of correctly identifying all types of cancer cells present in human tumours.

Sources: study abstract, resou.osaka-u.ac.jp, asianscientist.com.