A study run by the Royal Marsden NHS Foundation Trust, the Institute of Cancer Research, and Imperial College London has developed an AI algorithm based off of 500 patients with large lung nodules; potentially cancerous abnormal growths. The problem with classifying these growths as high or low risk of cancer is that being too cautious results in unnecessary tests, burdening both the patient and NHS, yet at the other end of the spectrum a doctor risks delayed diagnosis and poorer prognosis for the patient.
The AI in question involves radiomics, a technique that uses advanced mathematical analysis to enhance medical images beyond what is visible to the naked eye.
“Any new technology that helps gives more clarity over whether something on a CT scan is or isn’t cancer would be great. As a patient, you want to know whether you have the disease as soon as possible”
Dr Richard Lee. Consultant Physician in Respiratory Medicine and Champion for Early Cancer Diagnosis at the Royal Marsden Hospital
Current guidelines for categorising potential cancers involve using the Brock and Herder scores. The effectiveness of this AI technology was compared to these using the AUC (area under the curve) measure. With a result of 1 indicating perfect diagnosis, and 0.5 indicating the model was randomly guessing, the AUC of 0.87 with AI looks promising against 0.67 and 0.83 for the Brock and Herder tests respectively. These results suggest that the use of AI technology could help improve and streamline decision-making when it comes to diagnosing lung cancers and allowing high-risk patients to be fast-tracked to treatment. Dr Richard Lee, chief investigator of the study, said “Through this work, we hope to… speed up the detection of the disease”. In the study, the AI technology would have enabled earlier intervention in 82% of the malignant nodules.
Technologies like these are more important now than ever with the increasing pressure on the NHS, so a bit of virtual assistance would certainly be appreciated. It also helps reduce the agonising wait patients undergo to find out if their suspected nodules are in fact cancer. Keith Hewett, diagnosed with lung cancer back in 2018, said “Any new technology that helps gives more clarity over whether something on a CT scan is or isn’t cancer would be great. As a patient, you want to know whether you have the disease as soon as possible”.
Additional testing with larger datasets is required before the technology can be introduced to clinical settings, but the future looks bright.