Virtual autopsies guide postmortem investigation

Physician Rudolf Virchow introduced microscopic examination to classic pathology about midway through the 19th century. This helped to establish modern pathology. Although autopsies are now recognized as valuable medical procedures, the core methodology has not changed for many years. Now a number of digital radiology techniques, including multislice CT, MRI, and photogrammetry with 3D surface scanning, have the potential to change current practices. They could become forensic pathologists' main diagnostic tools in the future.

Autopsies are known for their importance in establishing cause of death. The procedures can provide key forensic information and guide criminal investigations. The number of autopsies carried out, however, is decreasing, and this downward trend in postmortem investigation is a serious issue.

Religious prohibitions are one reason. Orthodox Judaism, for example, prohibits disturbing dead bodies, except when such action may save others and urges avoidance of organ removal. Followers of Islam are opposed to any desecration or exposure of the body of a deceased believer.

Jurors can have difficulty understanding conventional autopsy protocols and photographs that are used as evidence in criminal cases. Autopsy records, such as tissue sections, are also difficult to store indefinitely.

It is now possible to conduct postmortem imaging, or 3D virtual autopsy, as part of the autopsy workflow. This involves cadaver scanning, to acquire MSCT or MRI data sets, and the use of 3D direct volume rendering techniques.1 Several studies have shown that virtual autopsy has great potential to aid forensic investigations.2-5

An imaging-led autopsy may be more compatible with religious beliefs held by the deceased's family. Virtual autopsies may also be seen as a safer option for coroners, pathologists, and medical examiners, given the growing threat of global pandemics such as avian flu.

Virtual autopsies have several other advantages over traditional methods of investigation. The body cannot be reassembled once an invasive autopsy has taken place, preventing other forensic pathologists from beginning an entirely separate analysis. Virtual autopsies, by contrast, can be carried out as many times as required. The use of whole-body imaging prior to conventional autopsy, for broad systemic examinations, can also reduce the total procedure time. Assessing the entire bone structure or searching for air, for instance, is otherwise difficult and time-consuming.

Virtual autopsies also have a number of disadvantages. Postmortem imaging with MSCT and MRI provides no color documentation of the body. This issue may be resolved with emerging 3D volume rendering methods and body surface scanning. Soft-tissue discrimination is poor with MSCT, though energy-resolved CT may improve this in the near future. Visualizing the circulation and possible bleeding is difficult as well, but promising results have been achieved with postmortem angiography.6

Postmortem imaging provides no information on macromorphology (i.e., histology, chemistry). This can be solved to a certain extent with MSCT-guided biopsies or MR spectroscopy. Postmortal gas can be difficult to distinguish from bowel gas or gas in wound channels. It is thus important to perform postmortal imaging as soon as possible after death.

IMAGING STRATEGIES

Both CT and MRI can be used for an imaging postmortem. Practitioners need to appreciate the principals of each technique and also its limitations to make the most of each modality's capabilities. It is relatively easy to visualize bone, gas, and metal with MSCT, for instance, but discrimination of soft tissue is limited.

Tools available to facilitate the visual presentation of postmortem imaging data include software packages that make it easier and quicker to view and interpret large volume data sets. High-quality data should first be acquired with isotropic volume elements (voxels). Image quality is maintained by selection of an appropriate rendering technique. Rendering effectively transfers 3D data in a PACS workstation into the 2D images viewed on monitors or film. Properly rendered images should give viewers a good understanding of the underlying 3D postmortal anatomy and the cause of death.

A striking feature of virtual autopsy data sets is the overwhelming amount of information. Up to 10,000 images can be reconstructed in a typical full-body MSCT virtual autopsy, leading to data sets several gigabytes in size. Lighting models essential for high-quality images also require precise gradient data, increasing the total amount of data still further.

The chosen system for direct volume rendering should enable fast handling of large data sets, seamless navigation and zooming, interactive adjustment of transfer function settings, and support for shading and lighting models. It should run on commercial desktop systems and graphics hardware, but existing systems for personal computers do not yet meet all these requirements.7 Transfer function makes volume data visible by mapping data values (Hounsfield units, signal intensity values) to optical properties, and transfer function settings map data values to color and opacity.

The bottleneck for virtual autopsy is no longer data acquisition. Future progress will depend on the development of appropriate data handling and analysis methods to make information comprehensible to users.

VIRTUAL AUTOPSY IN PRACTICE

The Center for Medical Image Science and Visualization (CMIV) at Linkoping University Hospital in Sweden, in collaboration with the Swedish National Board of Forensic Medicine, has developed a procedure for virtual autopsy that is now used routinely for forensic work. This method has been applied to over 160 cases so far, mostly homicides. Our experience has shown that full-body, high-resolution, direct volume rendered visualization is of great value in crime investigations.

Efforts have focused on total workflow for postmortem MSCT (Figure 1). Researchers have also developed software for interactive visualization of full-body data sets. We initially used a 16-slice CT scanner (Somatom Sensation 16, Siemens Medical Solutions) to acquire postmortem imaging data. This was replaced in September 2006 by a state-of-the-art dual-source CT scanner (Somatom Definition, Siemens). We now use both dual-source and an early beta version of dual-energy scanners for virtual autopsy cases. The first case that we examined with the new CT system was a virtual autopsy.

The forensic pathologist will, in most cases, visit the crime scene and oversee handling of the human cadaver. The cadaver is placed in a sealed body bag for transfer to the forensic department and then put in cold storage. A full-body MSCT scan is performed the next morning. The cadaver will remain in its body bag throughout the virtual autopsy procedure to ensure security of technical evidence of forensic value, such as fibers and body fluids.

The pathologist and radiologist carry out the direct volume rendering together in preparation for the physical autopsy. They can obtain a clear survey of the entire body quickly and localize fractures or air pockets (Figure 2). The procedure permits fast localization of foreign objects such as metal fragments or bullets (Figure 3). These can provide essential information in the early part of the police investigation. The high resolution allows extraction of dental information for identification purposes (Figure 4).

All captured MSCT data are stored, enabling repeat procedures. Findings during the physical autopsy often prompt additional questions that virtual autopsy can answer. The pathologist can also reexamine the cadaver at any point during the investigation and search for additional information. New findings in crime scene investigations may require that other hypotheses be assessed with imaging.

Virtual autopsy is currently used in conjunction with the standard autopsy procedure, but the added workflow is minimal. The MSCT scan and visualization session take much less time than a physical autopsy. The virtual examination can also decrease the time required for a physical investigation, because the pathologist is armed with prior knowledge.

KEY APPLICATIONS

Virtual autopsy can provide information for many possible causes of death.

• Polytrauma. Fractures can be visualized using 3D reconstruction. The rendered image should demonstrate underlying 3D postmortem anatomy, allowing viewers to appreciate the cause of death (Figure 2A). The use of MSCT data (small voxel size) together with direct volume rendering can provide sufficient dental information for identification (Figure 4). This is not always the case with conventional postmortem 2D radiography, which produces shadows from 3D objects. This means that fillings on the cheek side of a tooth cannot be distinguished from those on the tongue side.6

• Drowning. Traditional autopsies provide no unambiguous indications of drowning. Diagnosis relies primarily on the subject's individual characteristics and circumstances, macro- and micropathological postmortem changes, and studies of the lungs, including lung volume. MSCT offers an additional way to determine the cause of death in these cases. Practitioners can study the postmortem lung volume, mean attenuation, anterioposterior attenuation difference, lung density profile, and the amount of water within the lungs.8 This area of postmortem imaging holds great promise for the future.

• Shooting. Knowing the bullet's location before commencing physical autopsy will save the pathologist time, especially if the bullet has actually left the body. Bullets frequently lodge far from their entry points, particularly if they are deflected by curved bones such as the ribs or skull. We find that 3D MSCT data can minimize uncertainty about the location of bullets, entry and exit wounds, and wound channels. This is in line with published research from other institutions.9,10 We use a dual-transfer function rendering tool to further facilitate detection of submillimeter metal fragments. Such fragments can become buried in streaking artifacts around high-intensity materials and are otherwise hard to discover (Figure 3C).6 These tiny fragments can indicate a bullet's path and reveal information about the angle and direction of fire. All streaking artifacts from metal are a significant problem with MSCT. Direct volume rendering reduces the problem.11

• Burn injuries. Virtual autopsy has made a significant contribution to several of our cases involving severely burned bodies, providing information on fractures, gas distribution, and other areas. Routine police investigations became murder investigations in two cases; in both cases, the virtual autopsy revealed important findings that had been overlooked during the physical autopsy. Virtual autopsy findings were overlooked at the physical autopsy in one case.

• Hanging. Three-D reconstruction of MSCT data can verify vital signs that are caused by the hanging procedure to exclude postmortem simulated hanging. Three-D images can additionally visualize ligature marks that can help in reproducing the strangulation process.12-15

FUTURE SCENARIOS

A number of other areas may also benefit from virtual autopsy. These are currently under development.

• Body surface scanning. The body surface technique with 3D photogrammetry-based optical surface scanning can document and compare wounds and objects on the body. A surface scanner with a "light stripes" projector and two combined cameras, which measure the surface-dependent light-stripes deformation, can be used. Then the software can calculate a 3D model based on the deformation of the stripes.16-18

• Image-guided biopsy. Tissue specimens for histological or microradiological investigation can be acquired with minimum invasion under CT fluoroscopy guidance. Gas samples from lungs for chemical analysis can be obtained in the same way.

• Cardiac applications. The leading cause of natural death is cardiac insufficiency. Postmortem imaging must be able to separate these natural deaths from unnatural deaths. Calcification in the coronary arteries and cardiac valves can be quantified easily from MSCT data. Cardiac hypertrophy and acute dilatation can be quantified from MSCT and MRI data using measurements from long- and short-axis views. MSCT is unable to verify the early stages of myocardial infraction and lethal ventricular tachycardia. T2-weighted MRI can visualize the area of infarction if the deceased survived at least 30 to 60 minutes after the onset of infarction. The area of infarction will appear as a hypointensive area with a hyperintensive border in these cases. Pericardial tamponade caused by injuries of major vessels of the pericardium or the heart can be seen with both MSCT and MRI.19-25 Acute bleeds tend to have higher Hounsfield unit values than older bleeds on MSCT.

• Postmortem coronary angiography. Minimally invasive coronary angiography with MSCT can be used for postmortem cardiac diagnoses. Volume-rendered 3D reconstructions of coronary lumina with intraluminar contrast can be used to calculate soft and hard plaque volumes (Figure 5). MRI and MSCT can both be used to perform perfusion studies of the myocardium.6

• MR spectroscopy. Assessing the time since death is a fundamental problem. MR spectroscopy can provide an estimation of this time from metabolic information collected in a predefined region of the brain.26,27

• Data fusion. Fusing data from multiple modalities will enable a total, minimally invasive autopsy. Data fusion may also be an important tool in the courtroom of tomorrow.

Several forces must work in unison if a new era of digital autopsies is to emerge. Medical professionals and legal authorities must determine standard protocols for scanning and storing data. Legal systems around the world must accept the admissibility of imaging evidence in determining the cause and manner of death.

Invasive autopsies will remain the norm for the next few years, at least. Virtual autopsies have the potential to enhance the classic physical autopsy and improve the reliability of results. Noninvasive virtual autopsying could, in some cases, replace the invasive examination altogether. Minimal tissue sampling could then be performed under imaging guidance.

Research on the unique aspects of postmortem radiology is first needed to identify cases and validate procedures. Radiologists will also need special training in postmortem imaging. The dead often look quite different from living patients. Severe trauma and the effects of decomposition, for instance, can displace organs. Understanding these differences will require additional knowledge and expertise.

DR. PERSSON is director of the center for medical image science and visualization (CMIV), in the radiology department at Linkoping University Hospital in Sweden.

References

• Donchin Y, Rivkind AI, Bar-Ziv J, et al. Utility of postmortem computed tomography in trauma victims. J Trauma 1994;37(4):552-555.

• Ros PR, Li KC, Vo P, et al. Preautopsy magnetic resonance imaging: initial experience. Magn Reson Imag 1990;8:303-308.

• Thali M, Taubenreuther U, Karolczak M, et al. Forensic microradiology: micro-computed tomography (Micro-CT) and analysis of patterned injuries inside of bone. J Forensic Sci 2003;48(6):1336-1342.

• Thali M, Yen K, Schweitzer W, et al. Virtopsy, a new imaging horizon in forensic pathology: virtual autopsy by postmortem multislice computed tomography (MSCT) and magnetic resonance imaging (MRI)-a feasibility study. J Forensic Sci 2003;48(2):386-403.

• Jackowski C, Sonnenschein M, Thali MJ, et al. Virtopsy: postmortem minimally invasive angiography using cross section techniques-implementation and preliminary results. J Forensic Sci 2005;50(5):1175-1186.

• Ljung P, Winskog C, Persson A, et al. Full-body virtual autopsies using a state-of-the-art volume rendering pipeline, IEEE Trans Vis Comput Graph 2006;12(5):869-876.

• Sylvan E. CT-based measurement of lung volume and attenuation of deceased. Master's Thesis. Linkoping University, 2005:LiU-IMT-EX-05/410.

• Harris LS. Postmortem magnetic resonance images of the injured brain: effective evidence in the courtroom. Forensic Sci Int 1991;50(2):179-185.

• Oliver WR, Chancellor AS, Soitys J, et al. Three-dimensional reconstruction of a bullet path: validation by computed radiography. J Forensic Sci 1995;40(2):321-324.

• Etlik O, Temizos O, Dogan A, et al. Three-dimensional volume rendering imaging in detection of bone fractures. Europ J Gen Med 2004;1(4):48-52.

• Aghayev E, Yen K, Sonnenschein M, et al. Pneumomediastinum and soft tissue emphysema of the neck in postmortem CT and MRI; a new vital sign in hanging? Forensic Sci Int 2005;153(2-3):181-188.

• Wallace SK, Cohen WA, Stern EJ, et al. Judicial hanging: postmortem radiographic CT, and MR imaging features with autopsy confirmation. Radiology 1994;193(1):263-267.

• Yen K, Thali MJ, Aghayev E, et al. Strangulation signs: initial correlation of MRI, MSCT, and forensic neck findings. J Magn Reson Imaging 2005;22(4):501-510.

• Yen K, Vock P, Ender B, et al. Virtopsy: forensic traumatology of the subcutaneous fatty tissue. Multislice computed tomography (MSCT) and magnetic resonance imaging (MRI) as diagnostic tools. J Forensic Sci 2004;49(4):799-806.

• Bruschweiler W, Braun M, Dimhofer R, Thali MJ. Analysis of patterned injuries and injury-causing instruments with forensic 3D/CAD supported photogrammetry (FPHG): an instruction manual for the documentation process. Forensic Sci Int 2003;132(2):130-138.

• Thali M, Braun M, Markwader TM, et al. Bite mark documentation and analysis: The forensic 3D/CAD supported photogrammetry approach. Forensic Sci 2003;12(2):115-121.

• Thali M, Braun M, Wirth J, et al. 3D surface and 3D body documentation in forensic medicine: 3D/CAD photogrammetry merged with 3D radiological scanning. J Forensic Sci 2003;48(6):1356-1365.

• Alfakih K, Plein S, Thiele H, et al. Normal human left and right ventricular dimensions for MRI as assessed by turbo gradient echo and steady-state free precession imaging sequences. J Magn Reson Imaging 2003;17(3):323-329.

• Hsu JC, Johnson GA, Smith WM, et al. Magnetic resonance imaging of chronic myocardial infarcts in formalin-fixed human autopsy hearts. Circulation 1994;1994(5):2133-2140.

• Jackowski C. Macroscopical and histological findings in comparison with CT- and MRI- examinations of isolated autopsy-hearts. Ph.D. thesis. Institute of Forensic Medicine, Otto von Guericke University of Magdeburg, 2003.

• Jackowski C, Schweitzer W, Thali M, et al. Virtopsy: postmortem imaging of the human heart in situ using MSCT and MRI. Forensic Sci Int 2005;149(1):11-23.

• Jauhiainen T, Jarvinen VM, Hekali PE. Evaluation of methods for MR imaging of human right ventricular heart volumes and mass. Acta Radiol 2002;43(6):587-592.

• Kerner T, Frits G, Unterberg A, et al. Pulmonary air embolism in severe head injury. Resuscitation 2003;56(1):111-115.

• Lorenz CH, Walker ES, Morgan VL, et al. Normal human right and left ventricular mass, systolic function and gender differences by cine magnetic resonance imaging. J Cardiovasc Magn Reson 1999;1(1):7-21.

• Ith M, Kreis R, Scheurer E, et al. Using 1H-MR spectroscopy in forensic medicine to estimate the post-mortem interval: a pilot study in an animal model and its application to human brain. Presented at meeting of the International Society for Magnetic Resonance in Medicine, Glasgow, Scotland; April 2001:388.

• Ith M, Bigier P, Scheurer E, et al. Observation and identification of metabolites emerging during postmortem decomposition of brain tissue by means of in situ 1H-magnetic resonance spectroscopy. Magn Reson Med 2002;48(5):915-920.