Estimating human age from tooth samples is critical for forensic analysis. However, forensic experts are not readily available, especially in rural areas and small towns.
To help investigators get results faster, can we automate the process using machine learning algorithms?
Hemalatha Balan from the KGISL Institute of Technology, Coimbatore, collaborated with researchers from Saudi Arabia, Egypt and the US to tackle the problem.
From the Kovai Scan Center, Coimbatore, they collected orthopantomograms, X-rays of the upper and lower jaw. They also took images of a hundred healthy Indians, aged 4 to 18, and matched them with the gender and age of the individuals.
To reduce noise in the images, they used an isotropic diffusion filter. While preserving important image features, such as edges and boundaries, the filter selectively evens out image regions, based on their local properties, such as image colours and curvature.
Then, they segmented the dental images using a convolutional neural network, a deep learning algorithm used for image recognition and classification.
The researchers used the neural multi-kernel algorithm, a machine learning technique to improve accuracy in pixels. Then they trained the neural multi-kernel support vector machine to filter features from the images.
The image and feature vectors were processed by the spike neuron-based convolutional neural network to classify the images according to sex and age.
Image: Nazim Nazeer
The researchers compared the performance of their automatic age and sex classification system with results from existing techniques. Their technique’s accuracy in categorising was higher. The accuracy of gender classification was about100 percent, while age classification accuracy was above 90 percent.
With this method, forensic investigators can quickly gather information from teeth found at crime scenes.
Applied Artificial Intelligence 36 (1): 2022;
DOI: 10.1080/08839514.2022.2073724
Reported by Nazim Nazeer
ICMR- National Institute for Research in Environmental Health
*This report was written in the third online workshop on science writing organised by Current Science.
STEAMindiaReports 
Leave a comment