XClose

Institute of Archaeology

Home
Menu

Karina Andersson

Machine Learning as a Method of Sex Estimation from Cranial CT Scans

 

Email: karina.andersson.17@ucl.ac.uk
Section: Archaeological Sciences

Supervisors:

Profile

Machine Learning as a Method of Sex Estimation from Cranial CT Scans

Much of the work of a forensic anthropologist is focused on establishing a biological profile from skeletal remains--including estimations of age, sex, stature, and ancestry. Additionally, many techniques for establishing other aspects of the biological profile are reliant on an accurate sex estimation. The skull and pelvis are viewed as the most accurate skeletal elements for sex, as the pelvis exhibits reproductive differences and the cranium displays differences in size and morphology. However, ascertaining sex from the cranium is often based on a visual evaluation of macroscopic traits such as glabellar and nuchal crest prominence, size of the mastoid process, and gonial angle on the mandible. Unfortunately, recent studies have demonstrated that these visual methods are prone to inter and intra-observer error and are highly subject to practitioner bias.

Since the 1990s, there has been a push for forensic sciences, including anthropology, to move towards more objective methods with known error rates and this has led to higher standards for forensic evidence presented by expert witnesses in court. While some metric methods have been developed in an attempt to overcome these issues, they have also been proven to be lacking in consistency and many cranial measurements are poorly defined in the literature.

As the reliability and accuracy of current sex estimation methods are called into question, there is an increasing need for forensic anthropology to develop more objective and standardised practices in order to remain relevant. As such, the proposed project would attempt to use artificial intelligence (AI) in order to evaluate sex from cranial CT scans of known sex individuals. The goals of my research include developing a novel AI sex estimation method and comparing the accuracy and validity to traditional visual assessments.

Education

  • BSc, Anthropology, University of Calgary, 2016
  • BA, Archaeology, University of Calgary, 2016
  • MSc, Bioarchaeology and Forensic Anthropology, University College London, 2018
    Conference papers

    Andersson, K. 2021. Machine Learning as a Method of Sex Estimation from Cranial CT Scans. Poster session presented at 1st Annual UCL Collaborative Social Science Domain PhD Conference. University College London. (London, United Kingdom).  

    Andersson, K. 2018. Epidemiology of Talipes at Great Ormond Street Hospital Between 1854-1918: Utilising Digital Hospital Records.  Poster session presented at the Digital Dilemma Conference. University College London. (London, United Kingdom).

    Andersson, K., and A.M. Logan. 2014. Dental Calculus: A Sample Analysis from the Bögöz Site. Second International Student Colloquium on Osteology and Bioarchaeology. Session I: Adult Osteology. Haáz Rezső Múzeum, Odorheiu Secuiesc (Harghita, România).

    Andersson, K. 2014. Dental Calculus: A Sample Analysis from the Bögöz Site. Poster session presented at the Student Union Undergraduate Research Symposium. University of Calgary. (Calgary, Canada).