Session V: Molecular Screening; Imaging; Artificial Intelligence



Short Oral-13: Can we screen for pancreatic cancer? Identifying a sub-population of patients at high risk of subsequent diagnosis using machine learning techniques applied to primary care data

The authors conclude that they have managed to reduce the number of people they need to screen by more than a half. Now a full case-control study is needed (with a longer, more up-to-date prediagnostic period, use of "population-based controls" and a stratified analysis based on clinical status such as smoking & diabetes). The model should be applied "in real-time" to test its performance and there should also be an economic evaluation of cost-effectiveness.


Short Oral-14: Baseline diffusion weighted magnetic resonance imaging features to predict recurrence of anal squamous cell carcinoma (ASCC) following chemoradiotherapy

The authors identified baseline MRI features most notably ADC CoV, with the potential to improve prognostic accuracy of standard imaging evaluation, in addition to clinical variables such as T and N stage. 


Short Oral-15: Radiomic signature for prediction of peritoneal disseminations in gastric cancer which were not detected by routine CT examinations

The authors conclude that they successfully generated a proto prediction radiomic model for P-dis of GC. They would further refine the analytic process and generalize its accuracy by adopting the external validation cohort of GC with P-dis which could not be detected by the imaging examinations routinely used in daily practice.


Oral-15: The dark age of single organ screening is over: CD24 is a novel universal simple blood test for early detection of cancer

The authors conclude that CD24 expression in PBLs may serve as a universal blood test for the early detection of numerous cancers. It is the first-ever blood test to detect adenomas. CD24 can identify family members that are at increased risk for cancer. The test may serve as a novel predictive marker in cancer therapy. The authors conclude that obviously the test must be validated in a prospective multi-center study.