AI is certainly a topical issue these days and radiomics refers to techniques which allow the extraction of quantitative features from medical imaging to produce large data sets for clinical problem solving. This paper looks at CT imaging of the paranasal sinus for diagnostics and assessment of treatment outcomes. After excluding irrelevant studies from a huge initial pool, the authors looked at 10 studies in detail and used the radiomics quality assessment (RQS) as a published, validated score to assess the quality of the papers. The mean RQS across all 10 papers was 36.95%, which is considered poor, but the range was 19.44% - 63.89% (very low to good). When using the AUC-ROC (area under the curve receiver operating characteristics) assessment of the studies, looking at diagnostic prognostic studies, results were more impressive showing diagnostic utility ratings of excellent, very good and good. Three studies analysed prognostication in paranasal sinus CT radiomics looking at nasopharyngeal carcinoma distant metastases or looking at chronic rhinosinusitis post endoscopic sinus surgery and the need for further treatment, and the radionics were useful. The take-home message is that, whilst some of these studies show promise, the most beneficial part of radiomics is when it can identify patterns of disease which require large data sets and vast experience or minute and difficult-to-discern features which are challenging for humans. The benefits of radionics are lost in situations such as assessing a CT scan for chronic rhinosinusitis, as this is something which clinicians find easy to recognise and are trained to do. Radiomics has huge potential, however the challenge lies in learning the correct question which needs to be answered and this is where we will find the true benefits in the future.
Not the answer yet, but it may be in the future
Top Of Page
Accreditations
Pinpoint Scotland Ltd (Registered in Scotland No. SC068684) | © 2025 - Website by Gecko Agency