Around 3,000 new cases of mesothelioma are diagnosed in the US every year.
Several researchers, including those from King’s College London (KCL), have detailed how a new artificial intelligence (AI)-based tool called MesoGraph could help clinicians to diagnose mesothelioma.
Published in Cell Reports Medicine, the findings from the study could lead to more effective treatments for different forms of mesothelioma.
Responsible for around 3,000 new cases in the US every year, mesothelioma is a rare cancer of the mesothelial tissue, which surrounds the organs in the chest, abdominal cavity and pelvis.
The three types of mesotheliomas that can develop include pleural mesothelioma, the most common form of the disease, which occurs in the chest; peritoneal mesothelioma, which begins in the lining of the abdomen; and pericardial mesothelioma, which grows in the lining of the heart.
When diagnosing the disease, scientists will use histopathological images (very thin slices of tissue) to study the structures or cells under a microscope. Specifically, scientists will focus on changes in cellular morphology or shape to identify the disease.
However, diagnosis of mesothelioma is challenging due to the variation of the phenotype of cancer cells within tumours, which can lead to different diagnoses among healthcare professionals.
Using the MesoGraph, researchers were able to visualise the results of an analysis using a user-friendly interactive graphical user interface.
The tool employs graph neural networks, which consider tumours as a network of interacting cells that capture the surrounding environment alongside the characteristics of individual cells.
Researchers were then able to test the reliability of the tool with morphological analyses of the predicted cell scores.
They found that the MesoGraph successfully identified and recognised the various subtypes of mesothelioma.
Dr Heba Sailem, author of the study and senior lecturer of biomedical AI and data science at KCL, said: “Such technology not only has the potential to improve patient diagnosis but also can be utilised for targeted evaluation of therapy efficacy.”
As the disease can develop into one of three types, this analysis could help clinicians identify the type of mesothelioma and the best course of treatment.