Study on the detection of CT image based on multi-task network method of global attention for fine-fracture / 中国医学装备

Objective:To improve the perception of computed tomography(CT)images in detecting fine fracture through multi-task network of global attention,and to realize the detection of the target of fine fracture at case level through multi-task,and to quickly and accurately identify and locate fracture from a large number of CT images,so as to assist doctors to timely conduct treatment.Methods:A grouped Non-local network method was introduced to calculate the remote dependency relationship between each position of CT image continuous sections and channel.A single-stage detector of multi-objective detection model three dimension(3D)RetinaNet was integrated with the medical image semantic segmentation architecture(3D U-Net).A end-to-end multi-task 3D convolutional network was realized,which realized the detection of case level for fine fracture through multi-task collaboration.Select 600 CT scan images from the Rib Frac Dataset of rib fractures provided by the MICCAI 2020 Challenge,and they were divided into training set(500 cases)and test set(100 cases)as the ratio of 5:1 to test the precise performance of multi-task 3D convolutional network.Results:The precise performance of multi-task 3D convolutional network method was better than that of single-task FracNet,3D RetinaNet and 3D Retina U-Net in detection,which average precision was respectively higher 7.8%and 11.4%than 3D RetinaNet and 3D Retina U-Net.It was better than two kinds of single-task network detection method included 3D Faster R-CNN and 3D Mask R-CNN,and the average precision of that was respectively higher 6.7%and 3.1%than them.Conclusion:The integrated different modules of global attention multi-task network can improve the detection performance of fine fracture.The introduction of grouped Non-local network method can further improve the precise performance for the targets of fine fractures in detection.

The Relationship between Subjective and Objective Parameters in CT Phantom Image Evaluation

OBJECTIVE: To evaluate whether there is a relationship between subjective parameters determined by a reviewer (spatial resolution, low contrast resolution, and artifacts) and objective parameters (the CT number of water, noise, and image uniformity) in CT phantom image evaluations. MATERIALS AND METHODS: We reviewed the CT results of phantom image evaluations conducted by Korean Institute for Accreditation of Medical Image (KIAMI) from May 2007 to June 2007. We compared the objective parameters against the pass or fail groups for the subjective parameters. We also evaluated whether there is a relationship between the artifact types and the other subjective parameters. RESULTS: The mean noise value was significantly higher in the fail groups for the subjective parameters compared to the pass groups (p = 0.006). Specifically, noise and low contrast resolution were found to have a statistically significant positive correlation (r = 0.183, p < 0.001). In the fail group for low contrast resolution, the failure due to artifacts was significantly higher than the pass group (p < 0.001). In contrast, no statistically significant differences were found for the mean CT number of water, noise, or image uniformity based on the types of artifacts. CONCLUSION: Subjective CT image parameters evaluated by a reviewer correlate with objectively measured parameters, especially noise. Therefore, a stricter noise standard might be able to improve the subjective parameters results, such as low contrast resolution.

Feasibility of Automated Quantification of Regional Disease Patterns Depicted on High-Resolution Computed Tomography in Patients with Various Diffuse Lung Diseases

OBJECTIVE: This study was designed to develop an automated system for quantification of various regional disease patterns of diffuse lung diseases as depicted on high-resolution computed tomography (HRCT) and to compare the performance of the automated system with human readers. MATERIALS AND METHODS: A total of 600 circular regions-of-interest (ROIs), 10 pixels in diameter, were utilized. The 600 ROIs comprised 100 ROIs that represented six typical regional patterns (normal, ground-glass opacity, reticular opacity, honeycombing, emphysema, and consolidation). The ROIs were used to train the automated classification system based on the use of a Support Vector Machine classifier and 37 features of texture and shape. The performance of the classification system was tested with a 5-fold cross-validation method. An automated quantification system was developed with a moving ROI in the lung area, which helped classify each pixel into six categories. A total of 92 HRCT images obtained from patients with different diseases were used to validate the quantification system. Two radiologists independently classified lung areas of the same CT images into six patterns using the manual drawing function of dedicated software. Agreement between the automated system and the readers and between the two individual readers was assessed. RESULTS: The overall accuracy of the system to classify each disease pattern based on the typical ROIs was 89%. When the quantification results were examined, the average agreement between the system and each radiologist was 52% and 49%, respectively. The agreement between the two radiologists was 67%. CONCLUSION: An automated quantification system for various regional patterns of diffuse interstitial lung diseases can be used for objective and reproducible assessment of disease severity.

Nontraumatic Subperiosteal Orbital Hematoma Associated with Ethmoid Sinusitis: A Case Report

Subperiosteal hematoma of the orbit is a rare disease and most of the cases occur in young adult males as a result of direct facial or orbital trauma. In the absence of direct facial or orbital trauma, nontranmatic subperiosteal orbital hematoma has rarely been reported in association with a sudden elevation of cranial venous pressure or venous congestion, systemic diseases associated with a bleeding diathesis and paranasal sinusitis. We report here on a rare case of subperiosteal orbital hematoma associated with ethmoid sinusitis, as was seen on CT imaging.

1024 Matrix Image Reconstruction: Usefulness in High Resolution chest CT

PURPOSE: We tried to evaluate whether high resolution chest CT with a 1,024 matrix has a significant advantage in image quality compared to a 512 matrix. MATERIALS AND METHODS: Each set of 512 and 1024 matrix high resolution chest CT scans with both 0.625 mm and 1.25 mm slice thickness were obtained from 26 patients. Seventy locations that contained twenty-four low density lesions without sharp boundary such as emphysema, and forty-six sharp linear densities such as linear fibrosis were selected; these were randomly displayed on a five mega pixel LCD monitor. All the images were masked for information concerning the matrix size and slice thickness. Two chest radiologists scored the image quality of each arrowed lesion as follows: (1) undistinguishable, (2) poorly distinguishable, (3) fairly distinguishable, (4) well visible and (5) excellently visible. The scores were compared from the the aspects of matrix size, slice thickness and the different observers by using ANOVA tests. RESULTS: The average and standard deviation of image quality were 3.09 (+/-.92) for the 0.625 mmx512 matrix, 3.16 (+/-.84) for the 0.625 mmx1024 matrix, 2.49 (+/-1.02) for the 1.25 mmx512 matrix, and 2.35 (+/-1.02) for the 1.25 mmx1024 matrix, respectively. The image quality on both matrices of the high resolution chest CT scans with a 0.625 mm slice thickness was significantly better than that on the 1.25 mm slice thickness (p < 0.001). However, the image quality on the 1024 matrix high resolution chest CT scans was not significantly different from that on the 512 matrix high resolution chest CT scans (p = 0.678). The interobserver variation between the two observers was not significant (p = 0.691). CONCLUSION: We think that 1024 matrix image reconstruction for high resolution chest CT may not be clinically useful.

Added Value of Coronal Reformations for Duty Radiologists and for Referring Physicians or Surgeons in the CT Diagnosis of Acute Appendicitis

OBJECTIVE: To assess the added value of coronal reformation for radiologists and for referring physicians or surgeons in the CT diagnosis of acute appendicitis. MATERIALS AND METHODS: Contrast-enhanced CT was performed using 16-detector-row scanners in 110 patients, 46 of whom had appendicitis. Transverse (5-mm thickness, 4-mm increment), coronal (5-mm thickness, 4-mm increment), and combined transverse and coronal sections were interpreted by four radiologists, two surgeons and two emergency physicians. The area under the receiver operating characteristic curve (Az value), sensitivity, specificity (McNemar test), diagnostic confidence and appendiceal visualization (Wilcoxon signed rank test) were compared. RESULTS: For radiologists, the additional coronal sections tended to increase the Az value (0.972 vs. 0.986, p = 0.076) and pooled sensitivity (92% [95% CI: 88, 96] vs. 96% [93, 99]), and enhanced appendiceal visualization in true-positive cases (p = 0.031). For non-radiologists, no such enhancement was observed, and the confidence for excluding acute appendicitis declined (p = 0.013). Coronal sections alone were inferior to transverse sections for diagnostic confidence as well as appendiceal visualization for each reader group studied (p < 0.05). CONCLUSION: The added value of coronal reformation is more apparent for radiologists compared to referring physicians or surgeons in the CT diagnosis of acute appendicitis.

The Value of Coronal Image Reconstructions of HRCT using MDCT for the Assessment of Bronchiectasis: Experiment with 64 MDCTs

PURPOSE: The aim of our study was to evaluate the value of coronal image reconstructions of HRCT with using 64 MDCT scans for the assessment of bronchiectasis. MATERIALS AND METHODS: Chest CT scans (0.6-mm collimation, table speed of 14 mm/sec and a rotation time of 0.5 sec) that employed 64 MDCT images (Somatom Sensation 64, Siemens) without contrast media were performed in 56 patients (21 males and 35 females, mean age: 55 years) who displayed hemoptysis. The images were reconstructed with a 1 mm slice thickness in the axial (10 mm apart) and coronal (10 mm apart) planes with using a high frequency algorithm, and they were sent to PACS monitors. The axial images were assessed with and without the coronal images by two radiologists at two separate occasions. The presence of bronchiectasis was decided upon by consensus diagnosis of the two radiologists. The detection rates of bronchiectasis were compared between the readings with using the axial images alone and the readings with using both the axial and coronal images. RESULTS: The detection rate of bronchiectasis was significantly higher with using both the axial and coronal images than with using with axial images alone (82.1%, 46/56 patients Vs 64.3%, 36/56 patients, respectively, p=0.001). The detection rates for all the lobes, except for the superior division of the left upper lobe, were significantly improved with using both the axial and coronal images (RUL; p=0.013, RML; p=0.002, RLL; 0.024, Lt lingular segment; p=0.004, LLL; p=0.018). CONCLUSION: The coronal images of HRCT with using 64 MDCT improved the detection rate of bronchiectasis in the patients with hemoptysis when they were used in conjunction with the standard axial images. We suggest that HRCT with the coronal images should be obtained for the patients with hemoptysis, despite that the simple chest radiographs are often normal or they have non-specific findings.

The Value of Coronal Image Reconstructions of HRCT using MDCT for the Assessment of Bronchiectasis: Experiment with 64 MDCTs

PURPOSE: The aim of our study was to evaluate the value of coronal image reconstructions of HRCT with using 64 MDCT scans for the assessment of bronchiectasis. MATERIALS AND METHODS: Chest CT scans (0.6-mm collimation, table speed of 14 mm/sec and a rotation time of 0.5 sec) that employed 64 MDCT images (Somatom Sensation 64, Siemens) without contrast media were performed in 56 patients (21 males and 35 females, mean age: 55 years) who displayed hemoptysis. The images were reconstructed with a 1 mm slice thickness in the axial (10 mm apart) and coronal (10 mm apart) planes with using a high frequency algorithm, and they were sent to PACS monitors. The axial images were assessed with and without the coronal images by two radiologists at two separate occasions. The presence of bronchiectasis was decided upon by consensus diagnosis of the two radiologists. The detection rates of bronchiectasis were compared between the readings with using the axial images alone and the readings with using both the axial and coronal images. RESULTS: The detection rate of bronchiectasis was significantly higher with using both the axial and coronal images than with using with axial images alone (82.1%, 46/56 patients Vs 64.3%, 36/56 patients, respectively, p=0.001). The detection rates for all the lobes, except for the superior division of the left upper lobe, were significantly improved with using both the axial and coronal images (RUL; p=0.013, RML; p=0.002, RLL; 0.024, Lt lingular segment; p=0.004, LLL; p=0.018). CONCLUSION: The coronal images of HRCT with using 64 MDCT improved the detection rate of bronchiectasis in the patients with hemoptysis when they were used in conjunction with the standard axial images. We suggest that HRCT with the coronal images should be obtained for the patients with hemoptysis, despite that the simple chest radiographs are often normal or they have non-specific findings.

CT Densitometry of Lung Mass: The Effect of Reconstruction Algorithm

PURPOSE: To evaluate the effect of reconstruction algorithms on the CT measurement of mean lung mass density and normal thoracic structures. MATERIALS AND METHODS: Forty-six patients with a 2-9cm-sized lung mass underwent thoracic CT examinations with intravenous contrast enhancement and using a CT HiSpeed Advantage scanner (GE Medical Systems). In each examination, the axial image of the lung mass was reconstructed using soft, standard, detail, and bone algorithms. The mean value and standard deviation of mass density in Hounsfield Units (HU) were measured using ROIs of three different sizes (50 mm2, 200 mm2, and 350 mm2 or more), and the same method was used to measure the density of normal lung, muscle, bone, and vessels. In 21 patients, mass density was also measured on unenhanced and delayed enhanced images and the degree of enhancement was calculated. RESULTS: The average maximum difference in mean mass density in the images of the four different algorithms was less than 1 (range, 0.1 -1.9) HU (ROI size, 350 mm2 or more), 0 -4.2 HU (200 mm 2), and 0.1 -3.6 HU (50mm2). The average maximum difference in the degree of lung mass enhancement was 0.5 -1.2 (range, 0 -1.6 )HU (ROI size, 350 mm2 or more). The mean density of the four normal thoracic structures was highest in images reconstructed with the bone algorithm, though there was no significant difference between the four different algorithms (p = 1.000). CONCLUSION: The measured mean CT density of a lung mass larger than 2 cm does not significantly change according to the reconstruction algorithm used. When using a small ROI, however, the density difference may increase.

The Usefulness of CT and MR Virtual Colonoscopy in the Evaluation of Experimental Polyp Models

PURPOSE: The purpose of this study was to evaluate the diagnostic accuracy of CT and MR virtual colonoscopy using an experimental model of colonic polyposis. MATERIALS AND METHODS: Eleven pig colons ranging from 15 to 20cm in length were cleansed and a total of 99 variable sized polyps ranging from 3 to 15mm in diameter were created. The polyps were divided into three groups according to size: <5mm(n=23), 5-9mm(n=44), and ?10mm(n=32) in diameter. Each specimen was scanned using spiral CT and MRI. Each CT and MR virtual colonoscopy was independently evaluated by two radiologists blinded to the location, size and numbers of polyps, but aware that polyps were present. Interobserver agreement regarding the detection rate of the polyps during the two types of virtual colonoscopy was measured and the diagnostic accuracy of CT and MR virtual colonoscopy was compared. Results: In both CT and MR virtual colonoscopy, the detection rates determined by the two radiologists coincided in all three groups(p<0.05). The detection rates of polyps less than 5mm, between 5 and 9mm, and 10mm or larger in diameter were 74%(17/23), 89%(39/44), and 100%(32/32), respectively, in CT colonoscopy, and 61%(14/23), 89%(39/44), and 100%(32/32), respectively in MR colonoscopy. In polyps 5mm or larger, the results of the two types of virtual colonscopy coincided but in those less than 5mm in diameter, the results of MR virtual colonoscopy were slightly inferior to those of CT colonoscopy. Conclusion: Both CT and MR virtual colonoscopy provide high detection rates of colonic polyps 5mm or larger in diameter and these techniques can therefore be used to diagnose colonic mass lesions.

The Usefulness of CT and MR Virtual Colonoscopy in the Evaluation of Experimental Polyp Models

PURPOSE: The purpose of this study was to evaluate the diagnostic accuracy of CT and MR virtual colonoscopy using an experimental model of colonic polyposis. MATERIALS AND METHODS: Eleven pig colons ranging from 15 to 20cm in length were cleansed and a total of 99 variable sized polyps ranging from 3 to 15mm in diameter were created. The polyps were divided into three groups according to size: <5mm(n=23), 5-9mm(n=44), and ?10mm(n=32) in diameter. Each specimen was scanned using spiral CT and MRI. Each CT and MR virtual colonoscopy was independently evaluated by two radiologists blinded to the location, size and numbers of polyps, but aware that polyps were present. Interobserver agreement regarding the detection rate of the polyps during the two types of virtual colonoscopy was measured and the diagnostic accuracy of CT and MR virtual colonoscopy was compared. Results: In both CT and MR virtual colonoscopy, the detection rates determined by the two radiologists coincided in all three groups(p<0.05). The detection rates of polyps less than 5mm, between 5 and 9mm, and 10mm or larger in diameter were 74%(17/23), 89%(39/44), and 100%(32/32), respectively, in CT colonoscopy, and 61%(14/23), 89%(39/44), and 100%(32/32), respectively in MR colonoscopy. In polyps 5mm or larger, the results of the two types of virtual colonscopy coincided but in those less than 5mm in diameter, the results of MR virtual colonoscopy were slightly inferior to those of CT colonoscopy. Conclusion: Both CT and MR virtual colonoscopy provide high detection rates of colonic polyps 5mm or larger in diameter and these techniques can therefore be used to diagnose colonic mass lesions.

Leiomyoma of the Trachea: A Case Report

Tracheal leiomyoma is a rare benign tumor arising from the smooth muscle in the tracheal wall. As in other cases of benign tracheal tumors, the obstructive symptoms due to leiomyoma are similar to those of asthma and bronchitis, and diagnostic delay is therefore common. We describe the case of a patient with leiomyoma arising from the trachea and review the literature, with particular emphasis on the two dimensional CT images.