Study cases

We retrospectively reviewed 214 cases that underwent PMCT and forensic autopsies from January 2015 to December 2015 at our institution. We excluded cases with severe pneumothorax or haematothorax, a large pleural effusion, severe trauma, infants, pronounced decomposition, and those who were mummified. Ultimately, 170 cases were selected. Their average age at death was 59 (range, 8–91) years, and there were 50 females and 120 males.

CT and autopsy

We used 8- or 64-channel multi-detector row CT scanners (Aquilion; Toshiba Medical Systems, Tokyo, Japan). All cases were scanned in body bags. No contrast material was administered. A high-resolution conventional chest scan with 1.0 mm slices was performed at 25–30 mm intervals through the thorax. The tube voltage was 120 kVp, the tube current was 300 mAs, and the rotation time was 1.5 s per rotation. Subsequently, helical scanning with 2.0 mm collimation was performed from head to pelvis. The tube voltage was 120 kVp and the tube current was arbitrarily changed by the installed mode. The rotation time was 0.75 s per rotation, the table speed was 14 mm per rotation, and the helical beam pitch was 0.875. The lung portion of the helical data was reconstructed into a 2.0 mm slice thickness with the lung kernel.

   Conventional autopsies were performed after forensic CT in all cases. A forensic pathologist with >30 years of experience performed all autopsies with his colleagues.

Image assessment

All CT image data were delivered to a DICOM server (POP-Net Server; ImageONE, Tokyo, Japan) and evaluated on a DICOM viewer (POP-Net Essential; ImageONE). We classified lung findings evident on PMCT and the causes of death as determined by autopsy into several major types. Next, we compared the lung findings with the causes of death and calculated the diagnostic performance (sensitivity, specificity, predictive positive value [PPV], predictive negative value [NPV], and accuracy) of PMCT lung findings in terms of the major causes of death.