A study conducted in the Department of Computer Languages and Systems at the University of Granada (Universidad de Granada) , in collaboration with the Department of Computer Science and Artificial Intelligence and the Laboratory of Anthropological Physics of the same university, achieved significant progress in the field of forensic medicine — to apply digital record imaging (RI) methods to automate forensic identification processes, which are currently done by hand.
A doctoral thesis directed by professors Óscar Cordón García and Sergio Damas Arroyo, and carried out by José Santamaría López, professor at the University of Cádiz (Spain), shows that the evolutionary algorithm (a technique used to solve problems inspired by human evolution) called “Scatter Search” is an ideal mechanism to design new RI methods which are more accurate, efficient and reliable than methods currently used in forensic identification.
Recently, the New England Journal of Medicine chose the medical image processing systems as one of the eleven most important innovations of the past thousand years in the field of Medicine, which reflects the importance of certain research lines such as the one opened by the study conducted at the University of Granada.
New RI methods
Digital RI is usually used in the fields of Computer Graphics and Vision, and focuses on matching or superimposing two similar images within the same coordinate system. Thanks to highly complex computer systems and the wide spectrum of application fields in which RI is currently used (Medicine, Entertainment Industry, etc.), and due to ongoing research in the past decades, there are many RI methods nowadays. However, although these methods provide good results, they can only be used in certain conditions.
In order to confirm and evaluate the performance of the RI methods proposed by this doctoral thesis, a wide variety of 3D medical images of human brains obtained from magnetic resonance imaging (MRI) were used. These images included scenarios with and without noise and situations with and without damage, particularly multiple sclerosis. The results were compared to those of the different current RI methods, the Scatter Search technique achieving the best performance.
Santamaría López explains that the next step in his study was to apply the RI methods proposed for the automation of certain tasks of vital importance to forensic experts in identification processes, such as the comparison and the obtaining of 3D models from forensic objects. “In order to carry out the first phase of this study, the Brown Medical School Hospital (Rhode Island, US) provided us with a variety of 3D images of human wrists obtained from computed tomography,” says Dr. Santamaría López.
An automated methodology
The next step was to obtain 3D models. In order to do so, the Laboratory of Anthropological Physics of the University of Granada, in collaboration with researchers Miguel Botella and Inmaculada Alemán, provided human support, technical means (a 3D image range scanner) and mortuary material (forensic objects).
Thanks to the progress achieved by this doctoral thesis, the group Soft Computing and Intelligent Information Systems and the Laboratory of Anthropological Physics, both from the University of Granada, have started new research on the development of an automated methodology for forensic identification by using evolutionary algorithms, diffuse logic and record imaging. This new study, directed by researcher Óscar Cordón García, will be carried out over the following three years within the I+D+I National Plan of the Spanish Ministry of Education and Science.
The results of this study were published in prestigious international journals such as Soft Computing, INFORMS Journal on Computing and Pattern Recognition Letters.
Reference: Prof. José Santamaría López. Department of Computer Languages and Systems of the University of Cádiz (Spain). Telephone: (+34) 956 015 287. E-mail: jose.santamarialopez@uca.es
