Standardized medical image registration for radiological identification of decedents based on paranasal sinuses

https://doi.org/10.1016/j.jflm.2017.12.003Get rights and content

Highlights

  • Standardized medical image registration looks suitable for radiological identification based on paranasal sinuses.

  • The presented approach is easy-to-use, time-efficient and reliable.

  • The presented approach provides a quick and easy method for human identification.

Abstract

Image registration software is frequently used in clinical radiology, e.g., for follow-up diagnosis. To a certain extent, the radiological identification of decedents (RadID) is comparable to a clinical follow-up diagnosis, in that two datasets from different dates are compared in terms of their anatomical characteristics (e.g., paranasal sinuses) or surgical implants. Due to the increasing use of computed tomography (CT) for head examinations in clinical radiology and the increased use of postmortem CT (PMCT) in forensic imaging, the comparison of three-dimensional (3D) clinical CT (termed as antemortem CT (AMCT) in this article) and PMCT datasets for RadID is becoming increasingly practical. In particular, the comparison of paranasal sinuses in AMCT and PMCT imaging is considered a suitable and reliable modality for RadID. However, previous publications regarding RadID based on comparisons of 3D datasets have not considered the implementation of image registration to provide software-side support for RadID. This article demonstrates and evaluates the use of a standard medical image registration procedure for RadID by comparing paranasal sinuses.

Introduction

The first advances in image registration algorithms occurred several decades ago1, 2, 3, 4, 5; however, even with the implementation of combined positron emission tomography (PET) - computed tomography (CT) scanners in the clinical field for almost two decades, accurate image alignment between the modalities by using medical image registration technologies have evolved constantly.6,7 Currently, image registration has become a common approach for follow-up diagnosis, planning radiation treatment, angiography or dual-energy CT.8, 9, 10, 11, 12, 13 Thus, software tools for image registration have found widespread application in clinical medicine.

Since the implementation of forensic radiology in the field of legal medicine more than a decade ago,14, 15, 16, 17 the use of postmortem CT (PMCT) to supplement autopsy has increased continuously worldwide.18 Furthermore, PMCT has become a suitable method for the radiological identification of decedents (RadID)19,20; PMCT has even been recommended for disaster victim identification (DVI).21,22 Esteemed members of the International Society of Forensic Radiology and Imaging (ISFRI) have published recommendations how PMCT should be used for DVI.23 The increased use of CT head examinations in clinical radiology24 has made the technique of comparing paranasal sinuses in three-dimensional (3D) antemortem CT (AMCT) datasets and 3D PMCT datasets for RadID more practicable.24,25 The development of the frontal sinus is complete until the age of 15–20 years.26 From then on, the anatomy of the frontal sinus remains unchanged until old age, when the likelihood of the occurrence of progressive pneumatisation due to atrophic changes increases.27

Here, we present the use of a standard medical image registration tool for RadID that is based on comparing paranasal sinuses.

Section snippets

Materials and methods

Post-processing, including image registration and fusion, was performed using standard radiological imaging software for 3D reading and advanced visualization (MM Reading, syngo.via, Version VB10B HF03, Siemens Healthcare GmbH, Forchheim, Germany). This post-processing workflow enables image the alignment, registration and fusion of two 3D datasets from various modalities taken at different times. AMCT datasets of the head were acquired using various CT scanners from different vendors. PMCT

Results

No false positive matches or false negative matches (missed positive matches) were made by any rater. Thus, sensitivity, specificity and accuracy revealed 100% for each rater and the intraclass correlation was excellent (ICC = 1.00).

The measured times from starting the workflow until the identification of a positive or negative match in each case per rater were listed in Table 1. Time was indicated in minutes and seconds [mm:ss]. Statistical analyses revealed good intraclass correlations

Discussion

In this study, we aimed to present the use of a standard medical image registration technique for comparing AMCT and PMCT datasets based on paranasal sinuses as an easy-to-use, time-efficient and reliable approach for RadID. The simplicity of the presented approach is demonstrated by raters 2 and 3, who had no previous experience with the applied software or identification procedures at all. Importantly, none of the raters made false positive or false negative identifications. Despite the small

Conclusion

Standardized medical image registration looks suitable for RadID based on paranasal sinuses, and this simple method may present a quick and easy method for RadID.

Conflicts of interest

The authors have no conflict of interest to report.

Ethical approval

This study was performed with human cadavers. Ethical approval was waived by the responsible ethics committee of the Canton of Zurich (waiver number: 2015-0686). This article does not contain any studies with (living) human participants or animals performed by any of the authors.

Acknowledgement

The authors acknowledge Niklaus Zölch for his participation in this study. The authors express their gratitude to Emma Louise Kessler, MD, for her generous donation to the Zurich Institute of Forensic Medicine, University of Zurich, Switzerland.

References (31)

  • C.B.R.B. Soares et al.

    Human identification study by means of frontal sinus imaginological aspects

    Forensic Sci Int

    (2016)
  • A.L. Brough et al.

    The basics of disaster victim identification

    J Forensic Radiol Imaging

    (2015)
  • T.L. Beaini et al.

    Human identification through frontal sinus 3D superimposition: pilot study with cone beam computer tomography

    J Forensic Leg Med

    (2015)
  • D.I. Barnea et al.

    A class of algorithms for fast digital image registration

    IEEE Trans Comput

    (1972)
  • H.S. Baird

    Model-based Image Matching Using Location

    (January 1985)
  • Cited by (22)

    • Artificial intelligence in forensic anthropology: State of the art and Skeleton-ID project

      2023, Methodological and Technological Advances in Death Investigations: Application and Case Studies
    • Maxillary sinuses’ height/width/depth of Brazilian subjects and influence of sex, age, skin color, and nutritional status: A CBCT study

      2022, Forensic Imaging
      Citation Excerpt :

      Considering the particularities of their contours, paranasal sinuses are valuable for human identification, as they provide information and reduce the risk of errors by forensic professionals during investigations [11–14]. Located in the maxillary bones, maxillary sinuses are the first to develop and the largest among paranasal sinuses [15]. The formation of MS starts in the 10th week of fetal development.

    • Developments in postmortem imaging

      2022, Principles of Forensic Pathology: From Investigation to Certification
    • Annular distribution patterns of .357 Magnum fragments in soft tissue simulants after striking hard material that prevented the bullet from exiting

      2022, Journal of Forensic and Legal Medicine
      Citation Excerpt :

      The reconstructed data were visualized and analysed using medical image registration software (syngo.via VB30A_HF07, Siemens Healthineers, Erlangen, Germany). The software enables direct comparison of the CT data from multiple scans.5 For further analysis, CT data were displayed using multiplanar reformation (MPR), maximum intensity projection (MIP), and volume rendering technique (VRT).4

    • Automatic Annotation Algorithm of Medical Radiological Images using Convolutional Neural Network

      2021, Pattern Recognition Letters
      Citation Excerpt :

      Image annotation is an important research point for image content analysis. The quality of the annotation will directly affect the performance of the image application [1–2]. In the background of increasingly mature medical radiological image processing technology, medical radiological images are used for clinical diagnosis and treatment to improve the accuracy of clinical examinations [3].

    View all citing articles on Scopus
    View full text