Researchers say a new AI model could be more generalizable for cancer diagnosis and assessment than existing deep learning methods,Euronews.com reports.
Scientists have designed a new artificial intelligence (AI) model that may be able to diagnose and evaluate several types of cancer.
The new model, called CHIEF (Clinical Histopathology Imaging Evaluation Foundation), was up to 36 percent more effective at detecting cancer, determining the origin of a tumor and predicting patient outcomes than other deep learning models, the researchers said.
The team led by Harvard Medical School researchers wanted the model to be more widely applicable to various diagnostic tasks, as many current deep learning models for cancer are trained to perform specific functions.
“Unlike existing methods, our AI tool provides clinicians with an accurate, real-time second opinion on cancer diagnosis, taking into account a wide spectrum of cancer types and variations,” said Assistant Professor Kun-Hsing Yu of biomedical informatics at Harvard Medical School and lead author of the study, in an email to Euronews Health.
How does CHIEF work?
The model was trained on more than 15 million different pathological images, “which increases its reliability in diagnosing cancers with atypical features,” Yu added.
The researchers tested their model on more than 19,400 images from 24 hospitals and patient cohorts around the world and published the results in the journal Nature.
The team said the model works by reading digital slides of tumor tissues and can predict their molecular profile based on features in the image. It can also identify characteristics of a tumor that relate to how a patient may respond to treatment.
It achieved nearly 94% accuracy in detecting cancer cells in 11 types of cancer, based on a metric for model performance.
“In certain applications, such as identifying colon cancer cells or predicting genetic mutations, the performance of our model has reached up to 99.43 percent,” said Yu.
The researchers hope the AI model will help doctors more accurately assess a patient’s tumor.
The new model “represents a promising advance” in the application of AI in oncology, according to Ajit Goenka, a professor of radiology at the Mayo Clinic in the US, who was not involved in the study.
It told Euronews Health in an email that it could “simplify preliminary diagnostic assessments” and provide pathologists with a tool that analyzes slides to “highlight critical areas for further examination”.
“Despite these capabilities, the robustness of CHIEF in various clinical settings remains to be rigorously tested, and the potential for bias resulting from its training on large, possibly unrepresentative data sets cannot be ignored,” said Goenka, who researchers using AI to improve pancreatic cancer diagnosis.
What’s next for the AI model?
The remaining step before CHIEF can be used in doctors’ offices is to receive regulatory approval.
“To achieve this, we are launching a prospective clinical trial to validate the CHIEF model in real clinical settings,” said Yu, who added that they are also working to expand its ability to detect rare cancers.
Goenka added that to be used clinically, the model will need “extensive validation in real-world settings, encompassing various patient demographics and various clinical conditions.”
“This validation is essential to ensure that the performance of the model is not only theoretically superior but also practically reliable in everyday clinical practice,” he said.
