Evaluation of post-mortem lateral cerebral ventricle changes using sequential scans during post-mortem computed tomography

Abstract

In the present study, we evaluated post-mortem lateral cerebral ventricle (LCV) changes using computed tomography (CT). Subsequent periodical CT scans termed "sequential scans" were obtained for three cadavers. The first scan was performed immediately after the body was transferred from the emergency room to the institute of legal medicine. Sequential scans were obtained and evaluated for 24 h at maximum. The time of death had been determined in the emergency room. The sequential scans enabled us to observe periodical post-mortem changes in CT images. The series of continuous LCV images obtained up to 24 h (two cases)/16 h (1 case) after death was evaluated. The average Hounsfield units (HU) within the LCVs progressively increased, and LCV volume progressively decreased over time. The HU in the cerebrospinal fluid (CSF) increased at an individual rate proportional to the post-mortem interval (PMI). Thus, an early longitudinal radiodensity change in the CSF could be potential indicator of post-mortem interval (PMI). Sequential imaging scans reveal post-mortem changes in the CSF space which may reflect post-mortem brain alterations. Further studies are needed to evaluate the proposed CSF change markers in correlation with other validated PMI indicators.

Keywords: Lateral cerebral ventricle; Post mortem interval; Post-mortem change; Post-mortem computed tomography; Sequential scan.

Fig. 1

Case #1. The number under each image denotes post-mortem time. The right-bottom image denotes the sagittal plane from which these slices were obtained. The yellow line indicates the OM line, 3 cm above the highest point of the posterior clinoid process

Fig. 2

Case #1. Three-dimensional image of the lateral cerebral ventricle (LCV). Left : 2 h post-mortem. The volume was 26.8 ml. Middle: 20 h post-mortem. The volume was 21.8 ml. The LCV was slightly thinner at 20 h than at 2 h. Right: Flowchart of semi-automated segmentation algorithm of the LCV. Input: 3D CT images of a cadaver. Step 1: Input of seed regions of the LCV and background by an observer for a 3D CT image. The seed regions are deformed to generate seed regions for the remaining 3D CT images with different acquisition times in the same cadaver. Step 2: Statistical parameter estimation of a probability distribution for each 3D CT image by using an expectation maximization algorithm followed by likelihood computation. Step 3: Graph-cut segmentation for each 3D CT image that optimizes a given energy function. Output: 3D-labeled images of the LCV in the 3D CT images, in which the gray values inside the LCV are one and the outside is zero

Fig. 3

LCV volume decreases as a function of post-mortem time

Fig. 4

HU unit increases within the LCVs as a function of post-mortem time