All issues
Volume 11 (1994)
- Issue 3 Pages 87-
- Issue 2 Pages 35-
- Issue 1 Pages 1-
Volume 11, Issue 3
Displaying 1-6 of 6 articles from this issue
- |<
- <
- 1
- >
- >|
-
Mitsuo UNO, Masatoshi TSUZAKA, Hiroshi FUJITA, Michito SAWADA1994 Volume 11 Issue 3 Pages 87-90
Published: 1994
Released on J-STAGE: August 27, 2012
JOURNAL FREE ACCESSWe present a procedure to automatically classify the slice positions in chest X-ray CT images for the purpose of increasing the true-positive candidates of lymphnodes and also decreasing the false-positive ones detected by our algorithm developed previously. Image features determined from the position and the shape of vessel and trachea on the CT slices were extracted in our approach. The classification of the slices into three categories (A to C) was successful by use of the trachea information. Detailed classification into nine categories in terms of the vessel information was fairly good for A and B categories (93%), but it was difficult to classify in the C category, which demonstrates the necessity of a new approach to the C as shown in the following paper of this issue.View full abstractDownload PDF (1726K) -
Takeshi HARA, Hiroshi FUJITA, Masatoshi TSUZAKA, Michito SAWADA1994 Volume 11 Issue 3 Pages 91-95
Published: 1994
Released on J-STAGE: August 27, 2012
JOURNAL FREE ACCESSWe have developed an automated classification method for slice positions in chest X-ray CT images, which is based on the template matching by applying a genetic algorithm (GA), in order to increase the true positives and to reduce the false positives of mediastinal lymphnodes detected by our detection algorithm. Combining the GA method with a feature extraction one which we have already reported yielded 96% classification rate.View full abstractDownload PDF (1703K) -
Kenichi HIRAKO, Hiroshi FUJITA, Tokiko ENDO, Katsuhei HORITA, Choichir ...1994 Volume 11 Issue 3 Pages 96-100
Published: 1994
Released on J-STAGE: August 27, 2012
JOURNAL FREE ACCESSWe have developed a new automated-detection algorithm for clustered microcalcifications on digital mammograms. In our technique, the vectors of density gradient were firstly calculated within the area of breast which was segmented automatically. Second, three "circular-shape" filters were developed to extract the specific features for microcalcifications pattern from the vectors. The sen-sitivity and specificity of our algorithm were 81% and 63% with 0.53 false detection per image for our database of 100 mammograms.View full abstractDownload PDF (702K) -
Du-Yih TSAI, Eiji WATANABE, Katsuyuki KOJIMA, Kou FUJISAWA, Kensaku KA ...1994 Volume 11 Issue 3 Pages 101-107
Published: 1994
Released on J-STAGE: August 27, 2012
JOURNAL FREE ACCESSAn accurate and efficient method for automated measurement of right and left ventricular volumes in magnetic resonance imaging is proposed. The method can be divided into four major stages. They are preprocessing, adaptive thresholding,2-dimensional boundary extraction, and 3-dimensional display and volume calculation. A simulation study using a generated sphere with known dimension and an anatomic specimen study using an excised pig heart were performed to validate the accuracy of our algorithm for computerized measurement, followed by a clinical study of a healthy-volunteer heart. The measurement results were compared to that obtained with manual tracing by two human experts. Our preliminary results show that the proposed method provides acceptable accuracy and is clinically applicable.View full abstractDownload PDF (5199K) -
Masami UEDA, Hiroshi INATSU, Komyou KARIYA, Suguru UCHIDA1994 Volume 11 Issue 3 Pages 108-115
Published: 1994
Released on J-STAGE: August 27, 2012
JOURNAL FREE ACCESSThe relationship between the number of signals averaged, slice thickness, and image quality in a magnetic resonance imaging system were assessed using the relative efficiency of transmission, η, and the conditioal entropy, Hx(Y), using the enropy method. The value of η is the measurement related to both contrast and noise of the images. Hx(Y) represents the noise component. The value of η increased in form of a S curve as the number of signals averaged was increased.
By contrast, the relationship between the slice thickness and the value of η was approximatcly linear. Also, the value of Hx(Y) decreased as the number of signals averaged and the slice thickness were increased.
The evaluation of image quality by η agreed well with visual observation. Thus, η and Hx(Y) were effective for the evaluation of magnetic resonance images.View full abstractDownload PDF (2113K) -
Du-Yih TSAI, Daisuke FUKUOKA, Katsuyuki KOJIMA, Kou FUJISAWA, Kensaku ...1994 Volume 11 Issue 3 Pages 116-123
Published: 1994
Released on J-STAGE: August 27, 2012
JOURNAL FREE ACCESSIn this paper, the classification of echocardiographic images is studied by making use of some texture features, including the angular second moment, the contrast, the correlation, and the entropy which are obtained from a gray-level coocurrence matrix. Features of these types are used to classify two sets of echocardiographic images?normal and abnormal (cardiomyopathy) hearts (18 and 13 samples, respectively). The minimum distance classifier and the evaluation index are employed to evaluate the performance of these features. Implementation of our algorithm is performed on a PC-386 personal computer and produces about 90% correct classification for the two sets of echocardiographic images. Our preliminary results suggest that this method of feature-based image analysis has potential use for computer-aided diagnosis of heart diseases.View full abstractDownload PDF (2971K)
- |<
- <
- 1
- >
- >|