404. A SYSTEMATIC REVIEW AND META-ANALYSIS ON THE ROLE OF RADIOMICS IN ESOPHAGEAL CANCER: DIAGNOSIS, TREATMENT RESPONSE AND SURVIVAL PREDICTIONNainika Menon, Nadia Guidozzi, Swathikan Chidambaram, Sheraz Rehan Markar
- General Medicine
The use of artificial intelligence, in the form of radiomics, enables more detailed interpretation of radiological images in less time, when compared to the human eye. Some challenges in managing esophageal cancer, particularly those reliant of human factors, can be addressed by incorporating radiomics into image interpretation, thereby supporting diagnosis, treatment planning and predicting treatment response and survival. This systematic review and meta-analysis aims to summarize the evidence in this expanding field.
The systematic review was carried out using Pubmed, MEDLINE and Ovid EMBASE databases—articles describing the use of radiomics in 18F-FDG PET/CT scans and CT scans in the esophageal cancer management were included. A meta-analysis was also performed.
50 studies were included, most of which were single center cohort studies. For the assessment of treatment response using 18F-FDG PET/CT scans, 7 studies (443 patients) were included in the meta-analysis. The pooled sensitivity and specificity were 86.5% (81.1–90.6) and 87.1% (78.0–92.8). For the assessment of treatment response using CT scans, 5 studies (625 patients) were included in the meta-analysis, with a pooled sensitivity and specificity of 86.7% (81.4–90.7) and 76.1% (69.9–81.4). The remaining 38 studies discussed the benefits of radiomics in diagnosis, radiotherapy planning and survival prediction, often in less time and best when used in conjunction with radiologists.
This review explores the many possibilities of radiomics in esophageal cancer management. The sensitivities of 18F-FDG PET/CT scans and CT scans are comparable but 18F-FDG PET/CT scans have improved specificity for AI-based prediction of treatment response. Integrated clinical and radiomic models were the best at facilitating diagnosis and survival prediction, particularly when combined with radiologists. More multi-center studies are required to compare and externally validate models to build an evidence base.