Non-invasive assessment of coronary inflammation improves the diagnostic accuracy of coronary death in post-mortem CT angiography

Background Post-mortem computed tomography (PMCT) is becoming the preferred method to diagnose the cause of death over autopsy. There are areas improvements for the technique. This study investigates the use of a minimally invasive procedure under imaging guidance to perform PMCT Angiography (PMCTA), and coronary inflammation assessment to identify coronary death to aid the radiologists' diagnosis against autopsy (ground truth). Methods Adults with unknown cause of death referred for Coroner's autopsy within 48 ​h from death were included. Three independent radiologists performed the analysis - cardiac radiologist performed the Fat Attenuation Index (FAI) score and AI-Risk, two radiologists with 2 years' and >10 years' PMCT experience blinded to autopsy results reported the cause of death without and with coronary inflammatory findings. Results PMCTA included 37 participants (15 by radiologist using imaging guided minimally invasive technique; 22 carotid cut down technique). FAI score and AI-Risk produced high sensitivity (84%) and specificity (94%) for the diagnosis of coronary death. The technique could be applied in both the contrast and non-contrast studies to differentiate coronary death versus non-coronary causes. Both radiologists showed high sensitivity (94%), but low specificity (72% and 56% respectively) in the diagnoses of coronary deaths. With the aid of the FAI score and AI-Risk, radiologists were able to identify all the true coronary deaths, and no false negative. Conclusions PMCTA can be performed by radiologists using imaging guided minimally invasive technique. The FAI Score and AI-Risk can aid in the diagnosis of coronary death in routine PMCT.