Publications by authors named "Eiichiro Okumura"
Article Synopsis
- The study developed and validated a new ultrasound index to diagnose primary medial-type knee osteoarthritis (OA) by measuring total osteophyte height (TOH) and distance between bones (DBB).
- A DBB of 5.6mm or less was found necessary to diagnose primary medial-type knee OA, while severe deformity required a TOH of 4.7mm or more along with the same DBB threshold.
- The TOH-DBB index effectively captured disease changes across different stages of primary medial-type knee OA, showing promising diagnostic capability.
[Segmentation of Mass in Mammogram Using Gaze Search Patterns].
Nihon Hoshasen Gijutsu Gakkai Zasshi
May 2024
Article Synopsis
- Radiologists often struggle to detect small or hidden lesions in dense breast tissue on mammograms, leading to the use of computer-aided detection (CAD) systems to enhance image interpretation.
- This study focused on using focus images from an eye-tracking device to accurately segment masses in mammograms, involving experts and technologists analyzing both abnormal and normal cases.
- Results showed that the UNIT method outperformed others in segmentation accuracy, achieving a higher dice coefficient, indicating it is a promising approach for enhancing mammogram analysis, though further improvements and larger sample sizes are needed.
[Effect of Color Temperature on Color Scale Test Patterns in Medical Liquid-crystal Display Monitor].
Nihon Hoshasen Gijutsu Gakkai Zasshi
March 2019
In Japan, medical liquid-crystal display (LCD) and general LCD monitors have color temperatures of 7500 and 6500 K, respectively. The differences in color temperature make it difficult for radiologists to judge whether the same color is being displayed on the monitor. Therefore, the radiologist may overlook lesions.
View Article and Find Full Text PDFArticle Synopsis
- Radiologists find it challenging to classify pneumoconiosis on chest X-rays, prompting the development of a computer-aided diagnosis (CAD) system using a three-stage artificial neural network (ANN) approach.
- The system analyzes regions of interest (ROIs) in the chest X-rays utilizing four texture features to classify pneumoconiosis into categories 0 through 3.
- Performance assessment showed high effectiveness of the ANN with area under the ROC curve values of 0.89 for severe cases and 0.84 for early cases, indicating its potential to assist radiologists in accurate classification.
[Development of an automated patient recognition method for chest CT images using a template-matching technique].
Nihon Hoshasen Gijutsu Gakkai Zasshi
October 2014
If patient information, such as identification number or patient name, has been entered incorrectly in a picture archiving and communication system (PACS) environment, the image may be stored in the wrong place. To prevent such cases of misfiling, we have developed an automated patient recognition system for chest CT images. The image database consisted of 100 cases with present and previous chest CT images.
View Article and Find Full Text PDFThe purpose of this study was to evaluate the detection performance of simulated nodules in chest computed tomography (CT) images and nuclear medicine images with an ordinary liquid crystal display (LCD) and a medical LCD (grayscale standard display function: GSDF) and gamma 2.2. We collected 72 chest CT image slices obtained from an LSCT phantom with simulated signals composed of various sizes and CT values and 78 slices of monochrome and color nuclear medicine images obtained from a digital phantom with a simulated signal composed of various sizes and radiation levels.
View Article and Find Full Text PDFArticle Synopsis
- A computer-aided detection (CAD) system for pneumoconiosis was developed using a combination of rule-based analysis and artificial neural networks (ANN) to analyze power spectra from chest radiographs.
- The study introduced three new enhancement techniques to improve the detection accuracy and reduce false positives/negatives, using a limited image database of normal and abnormal chest x-rays.
- The CAD system showed strong performance with area under the curve (AUC) values of 0.93 for severe cases and 0.72 for early cases, suggesting it could effectively assist radiologists in diagnosing pneumoconiosis.
[Computerized classification of pneumoconiosis radiographs based on grey level co-occurrence matrices].
Nihon Hoshasen Gijutsu Gakkai Zasshi
September 2011
Article Synopsis
- Pneumoconiosis is categorized from 0-4 according to the Pneumoconiosis Law, but physicians often struggle to accurately classify chest images.
- A computerized method was developed to automate this categorization process using extracted rib edge regions and various texture feature analysis techniques.
- The method demonstrated effectiveness, with a 69.7% accuracy in a consistency test and 61.8% in a validation test, highlighting the utility of the proposed features and rib edge removal in classifying disease indicators.
Misregistration errors occur at the periphery of the hepatic region due to respiratory- and interval-related changes in hepatic shape. To reduce these misregistration errors, we developed a temporal and dynamic subtraction technique to enhance small hepatocellular carcinoma (HCC) by using a 3D nonlinear image-warping technique. The study population consisted of 21 patients with HCC.
View Article and Find Full Text PDFArticle Synopsis
- Radiologists struggle to accurately classify pneumoconiosis from small nodules in chest X-rays, prompting the creation of a new computer-aided diagnosis (CAD) system using a combination of rule-based methods and artificial neural networks (ANN).
- The study utilized a database of chest radiographs, including both normal and abnormal cases, to identify regions of interest (ROIs) and apply frequency analysis through Fourier transforms.
- The CAD system showed promising classification performance with an Az value of 0.972±0.012, outperforming both the ANN method alone and the rule-based method, indicating its potential to assist radiologists in diagnosing pneumoconiosis effectively.
Automated image-matching technique for comparative diagnosis of the liver on CT examination.
Nihon Hoshasen Gijutsu Gakkai Zasshi
December 2005
When interpreting enhanced computer tomography (CT) images of the upper abdomen, radiologists visually select a set of images of the same anatomical positions from two or more CT image series (i.e., non-enhanced and contrast-enhanced CT images at arterial and delayed phase) to depict and to characterize any abnormalities.
View Article and Find Full Text PDF