An Automated Classification System for the Differentiation of Obstructive Lung Diseases based on the Textural Analysis of HRCT images

PURPOSE: To develop an automated classification system for the differentiation of obstructive lung diseases based on the textural analysis of HRCT images, and to evaluate the accuracy and usefulness of the system. MATERIALS AND METHODS: For textural analysis, histogram features, gradient features, run length encoding, and a co-occurrence matrix were employed. A Bayesian classifier was used for automated classification. The images (image number n=256) were selected from the HRCT images obtained from 17 healthy subjects (n=67), 26 patients with bronchiolitis obliterans (n=70), 28 patients with mild centrilobular emphysema (n=65), and 21 patients with panlobular emphysema or severe centrilobular emphysema (n=63). An five-fold cross-validation method was used to assess the performance of the system. Class-specific sensitivities were analyzed and the overall accuracy of the system was assessed with kappa statistics. RESULTS: The sensitivity of the system for each class was as follows: normal lung 84.9%, bronchiolitis obliterans 83.8%, mild centrilobular emphysema 77.0%, and panlobular emphysema or severe centrilobular emphysema 95.8%. The overall performance for differentiating each disease and the normal lung was satisfactory with a kappa value of 0.779. CONCLUSION: An automated classification system for the differentiation between obstructive lung diseases based on the textural analysis of HRCT images was developed. The proposed system discriminates well between the various obstructive lung diseases and the normal lung.

Reproducibility of Computer-Aided Detection Marks in Digital Mammography

OBJECTIVE: To evaluate the performance and reproducibility of a computer-aided detection (CAD) system in mediolateral oblique (MLO) digital mammograms taken serially, without release of breast compression. MATERIALS AND METHODS: A CAD system was applied preoperatively to the full-field digital mammograms of two MLO views taken without release of breast compression in 82 patients (age range: 33-83 years; mean age: 49 years) with previously diagnosed breast cancers. The total number of visible lesion components in 82 patients was 101: 66 masses and 35 microcalcifications. We analyzed the sensitivity and reproducibility of the CAD marks. RESULTS: he sensitivity of the CAD system for first MLO views was 71% (47/66) for masses and 80% (28/35) for microcalcifications. The sensitivity of the CAD system for second MLO views was 68% (45/66) for masses and 17% (6/35) for microcalcifications. In 84 ipsilateral serial MLO image sets (two patients had bilateral cancers), identical images, regardless of the existence of CAD marks, were obtained for 35% (29/84) and identical images with CAD marks were obtained for 29% (23/78). Identical images, regardless of the existence of CAD marks, for contralateral MLO images were 65% (52/80) and identical images with CAD marks were obtained for 28% (11/39). The reproducibility of CAD marks for the true positive masses in serial MLO views was 84% (42/50) and that for the true positive microcalcifications was 0% (0/34). CONCLUSION: The CAD system in digital mammograms showed a high sensitivity for detecting masses and microcalcifications. However, reproducibility of microcalcification marks was very low in MLO views taken serially without release of breast compression. Minute positional change and patient movement can alter the images and result in a significant effect on the algorithm utilized by the CAD for detecting microcalcifications.

Volumetric Measurements of Lung Nodules with Multi-Detector Row CT: Effect of Changes in Lung Volume

OBJECTIVE: To evaluate how changes in lung volume affect volumetric measurements of lung nodules using a multi-detector row CT. MATERIALS AND METHODS: Ten subjects with asthma or chronic bronchitis who had one or more lung nodules were included. For each subject, two sets of CT images were obtained at inspiration and at expiration. A total of 33 nodules (23 nodules > or = 3 mm) were identified and their volume measured using a semiautomatic volume measurement program. Differences between nodule volume on inspiration and expiration were compared using the paired t-test. Percent differences, between on inspiration and expiration, in nodule attenuation, total lung volume, whole lung attenuation, and regional lung attenuation, were computed and compared with percent difference in nodule volume determined by linear correlation analysis. RESULTS: The difference in nodule volume observed between inspiration and expiration was significant (p or = 3 mm. The volume of nodules was measured to be larger on expiration CT than on inspiration CT (28 out of 33 nodules; 19 out of 23 nodules > or = 3 mm). A statistically significant correlation was found between the percent difference of lung nodule volume and lung volume or regional lung attenuation (p or = 3 mm. CONCLUSION: Volumetric measurements of pulmonary nodules were significantly affected by changes in lung volume. The variability in this respiration-related measurement should be considered to determine whether growth has occurred in a lung nodule.

Reproducibility of Computer-Aided Detection System in Digital Mammograms

PURPOSE: To evaluate the reproducibility of the computer-aided detection (CAD) system for digital mammograms. MATERIALS AND METHODS: We applied the CAD system (ImageChecker M1000-DM, version 3.1; R2 Technology) to full field digital mammograms. These mammograms were taken twice at an interval of 10-45 days (mean: 25 days) for 34 preoperative patients (breast cancer n=27, benign disease n=7, age range: 20-66 years, mean age: 47.9 years). On the mammograms, lesions were visible in 19 patients and these were depicted as 15 masses and 12 calcification clusters. We analyzed the sensitivity, the false positive rate (FPR) and the reproducibility of the CAD marks. RESULTS: The broader sensitivities of the CAD system were 80% (12 of 15), 67% (10 of 15) for masses and those for calcification clusters were 100% (12 of 12). The strict sensitivities were 50% (15 of 30) and 50% (15 of 30) for masses and 92% (22 of 24) and 79% (19 of 24) for the clusters. The FPR for the masses was 0.21-0.22/image, the FPR for the clusters was 0.03-0.04/image and the total FPR was 0.24-0.26/image. Among 132 mammography images, the identical images regardless of the existence of CAD marks were 59% (78 of 132), and the identical images with CAD marks were 22% (15 of 69). The reproducibility of the CAD marks for the true positive mass was 67% (12 of 18) and 71% (17 of 24) for the true positive cluster. The reproducibility of CAD marks for the false positive mass was 8% (4 of 53), and the reproducibility of CAD marks for the false positive clusters was 14% (1 of 7). The reproducibility of the total mass marks was 23% (16 of 71), and the reproducibility of the total cluster marks was 58% (18 of 31). CONCLUSION: CAD system showed higher sensitivity and reproducibility of CAD marks for the calcification clusters which are related to breast cancer. Yet the overall reproducibility of CAD marks was low; therefore, the CAD system must be applied considering this limitation.

A Simple Text Searching System for Radiology Reports Based on World Wide Web Technology

PURPOSE: The authors have developed a system that searches for radiologic reports, and we have evaluated the utility of this system for clinical application, research and medical education. MATERIALS AND METHODS: The system retrieves radiologic reports from the OCS (ordering communication system) server and stores them in a separate computer. A CGI (common gateway interface) program was written to search the database with variable keywords given by the user. The system satisfied the requirements of functionality, accessibility and security. RESULTS: Users could search for radiologic reports with variable keywords using the web browser, and the user could refine the results using combinations of these keywords. The system has been running successfully for 2 years, and radiologists have used this system for searching or collecting cases, keeping reports and gathering statistics. CONCLUSION: The simple text searching system has been developed in a short time period with only small costs and a bit of hard effort, and this system is practical for clinical use, research and medical education.

Performance of Computer-aided Detection in False-Negative Screening Mammograms of Breast Cancers

PURPOSE: To analyze retrospectively the abnormalities visible on the false-negative screening mammograms of patients with breast cancer and to determine the performance of computer-aided detection (CAD) in the detection of cancers. MATERIALS AND METHODS: Of 108 consecutive cases of breast cancer diagnosed over a period of 6 years, of which previous screening mammograms were available, 32 retrospectively visible abnormalities (at which locations cancer later developed) were found in the previous mammograms, and which were originally reported as negative. These 32 patients ranged in age from 38 to 72 years (mean 52 years). We analyzed their previous mammographic findings, and assessed the ability of CAD to mark cancers in previous mammograms, according to the clinical presentation, the type of abnormalities and the mammographic parenchymal density. RESULTS: In these 32 previous mammograms of breast cancers (20 asymptomatic, 12 symptomatic), the retrospectively visible abnormalities were identified as densities in 22, calcifications in 8, and densities with calcifications in 2. CAD marked abnormalities in 20 (63%) of the 32 cancers with false-negative screening mammograms; 14 (70%) of the 20 subsequent screening-detected cancers, 5 (50%) of the 10 interval cancers, and 1 (50%) of the 2 cancers palpable after the screening interval. CAD marked 12 (50%) of the 24 densities and 9 (90%) of the 10 calcifications. CAD marked abnormalities in 7 (50%) of the 14 predominantly fatty breasts, and 13 (72%) of the 18 dense breasts. CONCLUSION: CAD-assisted diagnosis could potentially decrease the number of false-negative mammograms caused by the failure to recognize the cancer in the screening program, although its usefulness in the prevention of interval cancers appears to be limited.

Impact of Image Post-Processing on PACS Workstation: Dynamic Range Suppression(DRS) of Chest Radiographs

PURPOSE: To investigate the impact of post-processing on a PACS workstation before and after use of thedynamic range suppression method for the normal chest radiographs. MATERIALS AND METHODS: Forty normal chestradiographs of healthy adult volunteers aged 20 to 33 (average 27; M:F = 29:11) were acquired by FCR using adigital interface and then transferred to an in-house-developed PACS workstation. The image size of computed chestradiographs was 7.5MB with 1760 x 2140 matrix. An image enhancement processing named DRS, developed by theauthors, was applied to the acquired images and generated a total of 40 chest radiographs. These were presented tothree groups of observers, each consisting of one radiologist and one technician on the PACS workstation, whichhad two monitors of 1712 x 2100 resolution. So that external light would not affect the visibility of imagesduring observation, these were displayed in a light-controlled room. The J.J.Vucich method, suitably modified, wasused to evaluate the anatomical structures and physical parameters of processed and unprocessed radiographs. Usinga percentage scale, the observers evaluated both anatomical sections (seven anatomical items : cortical margins ofribs, left diaphragms, thoracic vertebrae, trachea, pulmonary vasculature, trabeculae of ribs and clavicle,diaphragm outline) and physical sections (four items : contrast, graininess, density, detail). The results for thethree groups, both before and after DRS processing, were then compared. RESULTS: There was a statisticallysignificant difference between the three groups: in the anatomical section, 78.64 before DRS and 82.55 after ; andin the physical section, 75.48 and 79.78 (p<0.05). The average values of all items were 77.06 before DRS and 81.17after (p<0.05). CONCLUSION: Post-processing of computed chest radiographs on the PACS workstation improves boththe visibility of anatomical features and general image quality. Thus, in a PACS environment, it can be a usefultool for enhancing the diagnostic efficacy of radiography.

Impact of Image Post-Processing on PACS Workstation: Dynamic Range Suppression(DRS) of Chest Radiographs

PURPOSE: To investigate the impact of post-processing on a PACS workstation before and after use of thedynamic range suppression method for the normal chest radiographs. MATERIALS AND METHODS: Forty normal chestradiographs of healthy adult volunteers aged 20 to 33 (average 27; M:F = 29:11) were acquired by FCR using adigital interface and then transferred to an in-house-developed PACS workstation. The image size of computed chestradiographs was 7.5MB with 1760 x 2140 matrix. An image enhancement processing named DRS, developed by theauthors, was applied to the acquired images and generated a total of 40 chest radiographs. These were presented tothree groups of observers, each consisting of one radiologist and one technician on the PACS workstation, whichhad two monitors of 1712 x 2100 resolution. So that external light would not affect the visibility of imagesduring observation, these were displayed in a light-controlled room. The J.J.Vucich method, suitably modified, wasused to evaluate the anatomical structures and physical parameters of processed and unprocessed radiographs. Usinga percentage scale, the observers evaluated both anatomical sections (seven anatomical items : cortical margins ofribs, left diaphragms, thoracic vertebrae, trachea, pulmonary vasculature, trabeculae of ribs and clavicle,diaphragm outline) and physical sections (four items : contrast, graininess, density, detail). The results for thethree groups, both before and after DRS processing, were then compared. RESULTS: There was a statisticallysignificant difference between the three groups: in the anatomical section, 78.64 before DRS and 82.55 after ; andin the physical section, 75.48 and 79.78 (p<0.05). The average values of all items were 77.06 before DRS and 81.17after (p<0.05). CONCLUSION: Post-processing of computed chest radiographs on the PACS workstation improves boththe visibility of anatomical features and general image quality. Thus, in a PACS environment, it can be a usefultool for enhancing the diagnostic efficacy of radiography.

Image Viewing Station for MR and SPECT: Using Personal Computer

PURPOSE: Macro language was programmed to analyze and process on Macintosh personal computers GE MR imagesdigitally transferred from the MR main computer, with special interest in the interpretation of information such as patients data and imaging parameters under each image header. By this method, raw data(files) of certain patients may be digitally stored on a hard disk or CD ROM, and the quantitative analysis, interpretation anddisplay is possible. MATERIALS AND METHODS: Patients and images were randomly selected. 4.X MR images were transferred through FTP using the ethernet network. 5.X and SPECT images were transferred using floppy diskets. Toprocess transferred images, an freely distributed software for Macintosh namely NIH Image, with its macrolanguage, was used to import images and translate header information. To identify necessary information, aseparate window named "Info-txt", was made for each image series. MacLC, Centris650, and PowerMac 6100/CD,7100/CD, 8100/CD models with 256 color and RAM over 8 Mbyte were used. RESULTS: Different versions of MR images and SPECT images were displayed simultaneously and a separate window named "Info-txt" was used to show all necessary information(name of the patient, unit number, date, TR, TE, FOV etc.). Additional information(diagnosis,pathologic report etc.) was stored in another text box in "Info-txt". The size of the file for each image planewas about 149 Kbytes and the images were stored in a step-like file folders. CONCLUSION: 4.X and 5.X GE Signa 1.5T images were successfully processed with Macintosh computer and NIH Image. This result may be applied to manyfields and there is hope of a broader area of application with the linkage of NIH Image and a database program.