Computer-aided diagnosis (CAD) is defined as a diagnosis made by a radiologist who takes into consideration the results of a computer analysis of radiographic images as a second opinion. The general approach of the CAD scheme is to alert the radiologist by indicating potential lesions and/or by providing objective measures of opacity patterns. This article is to provide an overview of CAD schemes for detection of pulmonary nodules, interstitial diseases and interval changes in digital chest radiographs. ROC analyses indicated that the observer performances for detection of these lesions were significantly improved by using these CAD systems.

View full abstract

X-ray spectra measured with a CdZnTe detector were corrected using response functions calculated with the Monte Carlo method. In the calculation, the Hecht equation was employed to deal with the effects of carrier trapping in a CdZnTe crystal. Corrected spectra showed excellent agreement with x-ray spectra obtained with a HPGe detector. This meant that the response functions calculated in this study were valid. We conclude that a CdZnTe detector can be employed in diagnostic x-ray spectroscopy with proper corrections.

View full abstract

The architectural distortion is a very important finding in interpreting breast cancers as well as microcalcification and mass on mammograms. In addition, it sometimes appears at the previous stage of obvious mass shadow. The treatment of breast cancer at an early stage will be possible if the distorted areas are detected correctly. In despite of the importance for detecting architectural distortions, no algorithms for detecting them have been reported. The purpose of this study was to develop a new detection method for focal retraction that is one of architectural distortion, around skinline. In order to extract the thick mammary gland regions, binarization technique was performed. The black top-hat processing based on morphological operators was applied to determine the suspect depressed regions around skinline. The false positives are eliminated by the features of their sizes and positions. After applying this technique to 51 digitized mammograms, the detection sensitivity was 92% with 3.8 falsepositives detections per image. It was concluded that this technique is effective to detect the architectural distortion.

View full abstract

We have been developing an automated detection scheme for mammographic microcalcifications as a part of computer-aided diagnosis (CAD) systems. The purpose of this study was to confirm the validation of the contrast-correctionm ethod in our CAD scheme by using three databases from 1996 to 2000 because there were obvious differences in contrast of microcalcificationsa mong the databases. In order to investigate the changes of contrast of microcalcifications, the average and standard deviation values of contrast extracted from each microcalcification were calculated, before the detection performance of those three databases was estimated. Although there were some changes in contrast of microcalcificationas ccording to the situations of the background areas, the contrast-correctionm ethod was functioned effectively. These results were confirmed by the performance evaluation by using three databases; corresponding three true-positive rates were over 80% with 0.5 or less false positive per image.

View full abstract

The specification of individuals by the dentistry-view is used for the identification of victims' body in the case of accidents. Then, an objective judgment is required to carry out the correct discernment not only by subjective observation by forensic doctors but quantitive numerical values by a computer. In this study, we propose a method of estimating the degree of identity using the Nth-order autocorrelation features extracted from the teeth radiograms. The autocorrelation features are widely used in the field of image recognition. As an initial result of applying this technique, the eight correct recognition results are obtained among nine cases.

View full abstract