Machine learning model for prediction on skull’s landmarks.

Abstract

The Skull shape is a key step in many applications such as skull recognition, surgical planning, archaeology, education research. The main problem is locating the landmarks in their accurate place then finding the measurements in skull to recognize gender, predict population type, and predict landmarks placement. Most current research use 2D Computed Tomography (CT) image to locate landmarks in their place and that lead to less accuracy to calculate the cranial measurements. The aim of this research is to build an electronic medical assistance for calculating the cranial measurements based on accurate 3D coordinates that relies on meshgrid model which converts 2D skull’s medical image into 3D version. Based on the calculated measurements, traditional classifiers such as Random Forest classifier, logistic regression, Multi-layer Perceptron (MLP) and Support Vector Machines (SVM) to model for automatically predict gender and population type from the skull’s medical image. Furthermore, the suggested model can be used to detect accurate landmarks for new images to enhance the recognition accuracy. In comparing to a state-of-the-art works for gender recognition, the suggested model achieving 99.62 % accuracy in its prediction in comparison to highest previous work which achieved 96 %.