Q&A with Dr Hardeep Singh Kalsi

Date:
04 February 2022

Dr Hardeep Singh Kalsi is a Clinical Research Fellow, whose role is funded by The Royal Marsden Cancer Charity, and member of the NIHR BRC’s Early Diagnosis & Detection team.

Image
Dr Hardeep Singh Kalsi, a clinical research fellow at The Royal Marsden and a member of the NIHR BRC's Early Diagnosis & Detection team
Pictured: Dr Hardeep Singh Kalsi

How might artificial intelligence (AI) improve the early diagnosis of cancer?

AI uses computer systems to perform complex decision-making processes that are normally carried out by humans. For example, it can be used to learn important features in imaging through machine learning, or to read medical text through natural language processing. This means we can develop new tools that navigate large amounts of data and support or even improve decision-making during cancer diagnosis.

AI does this by finding patterns in clinical scans and records that identify patients who might be at increased risk of cancer, or by automating triaging of test results to alert clinicians sooner. It can also be applied to identify pre-cancer states and cancer subtypes, and to predict clinical behaviour, including risk of relapse. Imaging and AI is one of the key themes of The Royal Marsden’s new Early Diagnosis and Detection Centre, which – in partnership with the ICR, and working closely with RM Partners Cancer Alliance – aims to speed up diagnosis and improve outcomes for patients.

What AI projects are you currently working on?

Research shows that patients who survive cancer are more likely to develop another in their lifetime compared with those without any cancer history. Lung cancer is the leading cause of cancer-related deaths worldwide, and the most common type to subsequently develop in cancer survivors. We hope to open the AI-SONAR study soon to explore AI approaches to lung nodules on CT scans in patients previously cured of cancer. Our aim is to see whether this can improve the early diagnosis of both cancer relapse in the lung and new lung cancers, to improve the treatment prospects and long-term outcomes for patients.

What other benefits might AI offer?

Sometimes, scan findings are labelled as ‘indeterminate’ and need to be repeated to watch for changes or concerning behaviour that might indicate cancer or another disease, which can be stressful for patients. AI may enable earlier diagnosis and so reduce such anxieties. It may also help reduce the number of scans and appointments needed, freeing up valuable healthcare resources to support other clinical or research needs.

What are your hopes for the role and use of AI?

We hope that patients will help us to find the most ethical and valuable approaches whereby AI can lead to ‘learning’ healthcare systems. It would be fantastic if earlier diagnosis of cancer remains one of our core strengths in this area. We hope our pioneering studies will provide a springboard to share such research on a wider scale to benefit all patients