It is well known that tumors exhibit strong phenotypic differences that can be assessed by imaging. A great advantage of medical imaging is its ability to noninvasively visualize cancer’s appearance, such as intra-tumoral heterogeneity, on a macroscopic level, at baseline and follow-up, from primary tumor to potential metastasis. In current clinical practice, tumors are also monitored by invasive biopsy and molecular profiling, but their spatial and temporal pathologic heterogeneity limits the ability of invasive biopsy techniques to fully capture their continuously changing state.
Currently, a hypothesis driven research is the mainstream methodology incorporating a single or limited number of imaging biomarkers approach either to aid in differential diagnosis, to provide with prognostic information regarding treatment response or to monitor and guide therapy. Apparently, this strategy suffers from selection bias, and therefore there is a shift in cancer imaging biomarkers to a more “holistic” approach by quantifying and extracting myriads of imaging patterns including texture and shape features on a pixel by pixel basis otherwise invisible to the human eye. The latter methodology is summarized under the term Radiomics and it’s the process were an intelligent algorithm is undergoing training with labelled data, then validation and testing is performed to determine the clinical performance answering a specific clinical question.
1 CPD credit
Watch the video and complete the online self-reflection form. Go to "My events" to download your certificate. You must login to watch the video and receive CPD.
Dr Nikolas Papanikolaou
Dr Nikolas Papanikolaou is a Biomedical Engineer with a Ph.D. in Medicine. Working with clinical MRI applications development, optimization, focusing on AI and ML in imaging.