Dr Balaji Ganeshan, Founder and Chief-Scientist, TexRAD Global Business Development, Director, Feedback Medical
Using AI to analyse tumours and uncover biomarkers Radiomics is an area of medical study that aims to extract large amounts of quantitative features from medical images using algorithms to analyse and characterise data. Using new technologies and artificial intelligence (AI), this field is undergoing rapid change as the potential to uncover disease characteristics (biomarkers) that traditional analysis of scans are often unable to identify, is being developed. When a radiologist looks at the image of a tumour on a CT or MRI scan they can see the shape, outline and position of the cancer, while the computer creating the image sees a large number of individual pixels, each with a different density. New algorithms and AI are being developed which can use this data to analyse tumours on a region by region basis, generating more information for the clinician to help them understand the nature of the disease.
One of these tools, TexRAD is an algorithm used as a research tool by more than 60 prolific university hospitals and imaging centres across the world to compare patterns of pixel density to produce information on the texture of the tissue. The system uncovers potential radiological biomarkers in medical images that are typically not visible to the naked eye, providing significant additional information which may assist discussions about treatment regimes. As tissue structure is altered by disease, especially by cancer, biomarkers which reliably measure this could be extremely valuable.
When an abnormality is seen on a scan, the TexRAD examination of the mass may identify differences between benign and malignant lesions, potentially helping in diagnosis. Published research (100+ in peer-reviewed journals and conferences) has shown that the technology may identify tissue changes such as fibrosis, which can put patients at risk of malignancy. Going one step further, there is also a growing body of data suggesting that cancers with specific patterns may be associated with a better or worse response to different treatments.
To maximise the benefit of AI based analysis, a suitable Picture Archiving and Communication System (PACS) is needed. Over the last 15 years, PACS has become the bedrock of medical imaging storage and management, with the need to capture the ever-growing number of digital images being accessed worldwide. In institutions where research is planned, a number of specific additional functions are necessary, such as the ability to link easily with other systems and, crucially, software able to anonymise data and ensure patient confidentiality.