Fast organ matching templates extracted from large anatomical image database for medical thermographs in realistic 3D and stereo visual environments

Medical thermograph has the ability to indicate metabolic fever in the human body. The areas of local maximum or minimum temperature are usually regarded as abnormal regions. The clinical applications will be greatly improved if specific patterns of abnormal regions can be matched to the corresponding anatomical organs. In this study, an organ matching template using geometric searching technique is developed. The searching result can identify the relationship between abnormal and suspicious regions. The proposed method is cooperatively composed of template image file and template data file. The template image file contains two thousand organs that are compressed into a bitmap image by bitwise encoding technology. Four types of combinations of male/female genders and anterior/posterior parts are generated. And two visualization effects of x-ray gray film and color film are developed in 3D/stereo visual environment. In addition, the template data file has a high compression ratio to record anatomical terminologies and pixel coordinates. More than 500 megabytes organ database is packed to only 15Mb. Twostage fast searching algorithm for identifying suspicious organs is achieved by minimal-box and area overlapping evaluation strategy. The resulting organs are then sorted and displayed by thumbnails. By controlling five visual display parameters, the translucent visual effect with bi-directional displaying order can be manipulated. Finally, this method takes less than five seconds to import template. Furthermore, it successfully matches suspicious organs and sorts them by geometric dependences within 12 seconds. In addition, the algorithm has more realistic sense than the ordinary 2D gray/false color thermographs.