Pattern Recognition 39 (2006) 2157 – 2165 www.elsevier.com/locate/patcog Image languages in intelligent radiological palm diagnostics Marek R. Ogiela a , ∗ , Ryszard Tadeusiewicz a , Lidia Ogiela b a Institute of Automatics, AGH University of Science and Technology, Al. Mickiewicza 30, PL-30-059, Krakow, Poland b Faculty of Management, AGH University of Science and Technology, Al. Mickiewicza 30, PL-30-059, Krakow, Poland Abstract This paper presents a new technique of computer-aided analysis and recognition of pathological wrist bone lesions. This method uses artificial intelligence (AI) techniques and mathematical linguistics allowing to evaluate automatically and analyse the structure of the said bones, based on palm radiological images. Possibilities of computer interpretation of selected images, based on the methodology of automatic medical image understanding, as introduced by the authors, were created owing to the introduction of an original relational description of individual palm (wrist) bones. This description has been built with the use of graph linguistic formalisms already applied in artificial intelligence. These were, however, developed and adjusted to the needs of automatic medical image understanding in earlier works of the authors, as specified in the bibliography section of this paper. The research described in this paper has demonstrated that the for needs of palm (wrist) bone diagnostics, specialist linguistic tools such as expansive graph grammars and EDT-label graphs are particularly well-suited. Defining a graph image language adjusted to the specific features of the scientific problem here-described allowed for a semantic description of correct palm bone structures (with consideration to idiosyncratic features). It also enabled interpretation of images showing some in-born lesions, such as additional bones; or acquired lesions such as their incorrect junctions resulting from injuries and synostoses.  2006 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved. Keywords: Pattern recognition; Image understanding; Medical image analysis; Computer-aided diagnosis; Wrist disease diagnostics 1. Introduction A significant breakthrough in the field of medical IT has brought about the elaboration of new, AI-based methods that provide opportunities to conduct automatic analysis of two- dimensional and three-dimensional medical images [1,2]. These methods were based on a newly established class of linguistic formalisms, based on the terms of ETPL (k)-class graph grammars, EDG and IE graphs, used as the new tool for description and discovering of important diagnostic fea- tures of broad range of various medical images. The new methodology of automatic understanding of medical images, developed by authors during the last ten years and described in details in book [3] can be considered as the next step on the way from image processing algorithms (e.g. contrast ∗ Corresponding author. Tel.: +48 12 617 38 54; fax: +48 12 634 15 68. E-mail addresses: mogiela@agh.edu.pl (M.R. Ogiela), rtad@agh.edu.pl (R. Tadeusiewicz), logiela@agh.edu.pl (L. Ogiela). 0031-3203/$30.00  2006 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.patcog.2006.03.014 enhancement), throw image analysis (e.g. densitometry) and automatic recognition or classification. The linguistic ap- proach for medical images description, analysis, classifica- tion and understanding can be practically performed thanks the development of methods and algorithms of an effec- tive syntactic analysers operation created for such grammar types. The implementation of such techniques brought about such important progress, that the application of syntactic recognition methods for medical images became not only possible, but also extremely effective. We use these tech- niques for the analysis of a wide range of medical images, describing obtained results in many papers [3–7]. We discovered syntactic methods of pattern recognition as the most effective method for diagnostic analysis of complex medical images. This is in particular true about those imaging a number of structures at the same time and it has a significant diagnostic meaning, mainly due to semantic information contained therein. Classic image analysis techniques were strongly limited in this respect