Thammasat University researchers analyzed CT-derived skull images from 404 Thai individuals, balanced at 202 females and 202 males, to develop a forensic sex-estimation model. A model using six cranial measurements reported 83.9% classification accuracy.
Key findings
- The 12-parameter model was reported at 57.8-77.6% accuracy, while the six-measurement model reached 83.9%. Selecting more discriminative variables may outperform simply adding measurements, but roughly one in six classifications would still be incorrect in the reported setting.
Why this matters globally
Population-specific standards matter because cranial morphology varies across populations and time. CT enables non-destructive measurement and may assist identification when remains are incomplete, but results should be integrated with DNA, dental, and contextual evidence.
Thai researcher contribution
Thammasat University radiological-technology and forensic-science researchers produced a population-specific Thai reference and linked medical imaging with human identification practice.
Limitations to consider
Regional, age, and temporal representativeness are not clear from the abstract. Accuracy may fall in mixed populations, cross-border cases, damaged skulls, or scans from different systems. Nationality verification and external validation are essential, and recorded biological sex does not encompass gender identity.