Search

In electrical patient record system， reference images are used for physicians， not for primary diagnosis. Lossy image compression method such as JPEG is often used for reference images. The purpose of this study is to develop a new image compression method to apply different compression ratio to the inside of ROIs according to normal JPEG scheme. ROIs can be often defined as more significant regions than others by radiologists. By using our method，two quantization matrices were used in the phase of encoding. We studied the quantization matrix for outside of ROI to minimize the degradation of image quality. The image size was 7.3% smaller than the normal JPEG images. It is easy to introduce our method into PACS， since any additional software was not required to decode image data compressed by the proposed method.

医用画像診断装置で撮影される画像において画質評価は重要な課題の1つである。医用画像の視覚特性の影響も考慮した心理評価手法の一つにC-Dダイアグラムがあるが，全ての撮影装置に対して評価用のBurgerファントムが準備されているわけではない。本論文ではすへての撮影装置を対象として，画像評価の物理特性であるMTFとウィナースペクトルのデータを元に仮想Burgerファントムを作成し，実際にX線診断装置にて撮影した画像データと比較してその妥当性を確認した。この仮想BurgerファントムがX線診断装置以外の画像診断装置に適用できるかどうかの評価は今後の重要な研究課題である。, Providing good image quality is considered to be one of the most important requirements for diagnostic imaging systems.Methods for quantitatively evaluating image quality can be classified into physical methods and psychological methods. Although physical methods are generally well established, psychological methods （in which the influence of the visual characteristics of diagnostic images is also taken into consideration） often suffer from difficulties in quantification.One psychological evaluation method involves the use of a C-D diagram with a Burger phantom. However, this method is not applicable to modalities other than X-ray systems and parameter setting is difficult.In the present study, a virtual Burger phantom was constructed on a PC based on the MTF and Wiener spectrum data of the imaging system which is the physical characteristic of image evaluation.The validity of this method was confirmed by comparing phantom image data against actually acquired image data using X-ray systems.Further research will be needed to confirm that this virtual Burger phantom should be applicable to other modalities than X-ray system.

金沢大学医薬保健研究域保健学系

Medical images, including images obtained X-ray computed Tomography (CT), magnetic resonance imaging (MR), position emission tomography (PET), and so on, give clinicians and surgeons significant information for diagnosis and surgical operation planning. Registration between images and between images and a patient, is an essential technique in order to make good use of those images. So far various kinds of methods are proposerd to solve the problem. In this paper, image registration methods used in clinical application are reviewed in several aspects, and the state of the art of medical image registration us described.

Supervisor:松澤照男 教授, 情報科学研究科, 博士

A magnetic resonance (MR) image detabase system has been developed. The purpose of this paper is to present this system and its application. It consists of an image processing system and database system to handle data of clinical cases and a medical knowledge base. Using the image processing system and the database system with stored data of clinical cases, the MR image data of cases can be retrieved by diagnosis as well as ID number, patient's name, etc., and MR images can be processed. By using the database system with the medical knowledge base, differential diagnoses can be offered.