Generative Adversarial Network-Based Image Conversion Among Different Computed Tomography Protocols and Vendors: Effects on Accuracy and Variability in Quantifying Regional Disease Patterns of Interstitial Lung Disease

OBJECTIVE

To assess whether computed tomography (CT) conversion across different scan parameters and manufacturers using a routable generative adversarial network (RouteGAN) can improve the accuracy and variability in quantifying interstitial lung disease (ILD) using a deep learning-based automated software.

MATERIALS AND METHODS

This study included patients with ILD who underwent thin-section CT. Unmatched CT images obtained using scanners from four manufacturers (vendors A-D), standard- or low-radiation doses, and sharp or medium kernels were classified into groups 1-7 according to acquisition conditions. CT images in groups 2-7 were converted into the target CT style (Group 1: vendor A, standard dose, and sharp kernel) using a RouteGAN. ILD was quantified on original and converted CT images using a deep learning-based software (Aview, Coreline Soft). The accuracy of quantification was analyzed using the dice similarity coefficient (DSC) and pixel-wise overlap accuracy metrics against manual quantification by a radiologist. Five radiologists evaluated quantification accuracy using a 10-point visual scoring system.

RESULTS

Three hundred and fifty CT slices from 150 patients (mean age: 67.6 ± 10.7 years; 56 females) were included. The overlap accuracies for quantifying total abnormalities in groups 2-7 improved after CT conversion (original vs. converted: 0.63 vs. 0.68 for DSC, 0.66 vs. 0.70 for pixel-wise recall, and 0.68 vs. 0.73 for pixel-wise precision; P < 0.002 for all). The DSCs of fibrosis score, honeycombing, and reticulation significantly increased after CT conversion (0.32 vs. 0.64, 0.19 vs. 0.47, and 0.23 vs. 0.54, P < 0.002 for all), whereas those of ground-glass opacity, consolidation, and emphysema did not change significantly or decreased slightly. The radiologists' scores were significantly higher (P < 0.001) and less variable on converted CT.

CONCLUSION

CT conversion using a RouteGAN can improve the accuracy and variability of CT images obtained using different scan parameters and manufacturers in deep learning-based quantification of ILD.

Relationships Between High-Resolution Computed Tomographic Features and Lung Function Trajectory in Patients With Asthma

PURPOSE

A subset of asthmatics suffers from persistent airflow limitation, known as remodeled asthma, despite optimal treatment. Typical quantitative scoring methods to evaluate structural changes of airway remodeling on high-resolution computed tomography (HRCT) are time-consuming and laborious. Thus, easier and simpler methods are required in clinical practice. We evaluated the clinical usefulness of a simple, semi-quantitative method based on 8 HRCT parameters by comparing asthmatics with a persistent decline of post-bronchodilator (BD)-FEV1 to those with a BD-FEV1 that normalized over time and evaluated the relationships of the parameters with BD-FEV1.

METHODS

Asthmatics (n = 59) were grouped into 5 trajectories (Trs) according to the changes of BD-FEV1 over 1 year. After 9-12 months of guideline-based treatment, HRCT parameters including emphysema, bronchiectasis, anthracofibrosis, bronchial wall thickening (BWT), fibrotic bands, mosaic attenuation on inspiration, air-trapping on expiration, and centrilobular nodules were classified as present (1) or absent (0) in 6 zones.

RESULTS

The Tr5 group (n = 11) was older and exhibited a persistent decline in BD-FEV1. The Tr5 and Tr4 groups (n = 12), who had a lower baseline BD-FEV1 that normalized over time, had longer durations of asthma, frequent exacerbations, and higher doses of steroid use compared to the Tr1-3 groups (n = 36), who had a normal baseline BD-FEV1. The Tr5 group had higher emphysema and BWT scores than the Tr4 (P = 8.25E-04 and P = 0.044, respectively). Scores for the other 6 parameters were not significantly different among the Tr groups. BD-FEV1 was inversely correlated with the emphysema and BWT scores in multivariate analysis (P = 1.70E-04, P = 0.006, respectively).

CONCLUSIONS

Emphysema and BWT are associated with airway remodeling in asthmatics. Our simple, semi-quantitative scoring system based on HRCT may be an easy-to-use method for estimating airflow limitation.

Epidural abscess formation after chemoradiation therapy for esophageal cancer: A case report and literature review

RATIONALE

Esophageal cancer is one of the leading causes of death worldwide; the treatments vary according to the stage at diagnosis. Advanced esophageal cancer is usually treated by concurrent chemoradiation which is associated with complications including esophagitis, esophageal stricture or perforation, radiation pneumonitis, and/or cardiac toxicity. Herein, we describe epidural abscess, which is a very rare but severe complication that can occur after concurrent chemoradiation therapy for advanced esophageal cancer.

PATIENT CONCERNS

A 75-year-old man developed a fever during concurrent chemoradiation therapy for advanced esophageal cancer, which progressed to neurological deficit and paraplegia. Enhanced chest computed tomography and C-spine magnetic resonance imaging were performed.

DIAGNOSIS

Chest computed tomography revealed a poorly enhanced necrotic change in the cervical esophageal cancer, with mottled dirty material and fluid collection. C-spine magnetic resonance imaging revealed a prevertebral abscess with pyogenic spondylitis at the C6-T2 level. In addition, an anterior epidural abscess at the C6-7 level compressed the spinal cord.

INTERVENTIONS

The patient underwent emergency anterior cervical discectomy and decompression corpectomy.

OUTCOMES

: After surgery, the neurological symptoms gradually improved.

LESSONS

Pyogenic spondylitis with an epidural abscess is a rare but life-threatening complication that can develop after concurrent chemoradiation therapy for advanced esophageal cancer. Rapid, accurate diagnosis and prompt surgical treatment are important to ensure a favorable long-term prognosis and a good quality of life.

Effectiveness of radiologist training in improving reader agreement for Lung-RADS 4X categorization

OBJECTIVES

To identify the agreement on Lung CT Screening Reporting and Data System 4X categorization between radiologists and an expert-adjudicated reference standard and to investigate whether training led to improvement of the agreement measures and diagnostic potential for lung cancer.

METHODS

Category 4 nodules in the Korean Lung Cancer Screening Project were identified retrospectively, and each 4X nodule was matched with one 4A or 4B nodule. An expert panel re-evaluated the categories and determined the reference standard. Nineteen radiologists were asked to determine the presence of CT features of malignancy and 4X categorization for each nodule. A review was performed in two sessions, and training material was given after session 1. Agreement on 4X categorization between radiologists and the expert-adjudicated reference standard and agreement between radiologist-assessed 4X categorization and lung cancer diagnosis were evaluated.

RESULTS

The 48 expert-adjudicated 4X nodules and 64 non-4X nodules were evenly distributed in each session. The proportion of category 4X decreased after training (56.4% ± 16.9% vs. 33.4% ± 8.0%; p < 0.001). Cohen's κ indicated poor agreement (0.39 ± 0.16) in session 1, but agreement improved in session 2 (0.47 ± 0.09; p = 0.03). The increase in agreement in session 2 was observed among inexperienced radiologists (p < 0.05), and experienced and inexperienced reviewers exhibited comparable agreement performance in session 2 (p > 0.05). All agreement measures between radiologist-assessed 4X categorization and lung cancer diagnosis increased in session 2 (p < 0.05).

CONCLUSION

Radiologist training can improve reader agreement on 4X categorization, leading to enhanced diagnostic performance for lung cancer.

KEY POINTS

• Agreement on 4X categorization between radiologists and an expert-adjudicated reference standard was initially poor, but improved significantly after training. • The mean proportion of 4X categorization by 19 radiologists decreased from 56.4% ± 16.9% in session 1 to 33.4% ± 8.0% in session 2. • All agreement measures between the 4X categorization and lung cancer diagnosis increased significantly in session 2, implying that appropriate training and guidance increased the diagnostic potential of category 4X.

Extravalidation and reproducibility results of a commercial deep learning-based automatic detection algorithm for pulmonary nodules on chest radiographs at tertiary hospital

INTRODUCTION

To extra validate and evaluate the reproducibility of a commercial deep convolutional neural network (DCNN) algorithm for pulmonary nodules on chest radiographs (CRs) and to compare its performance with radiologists.

METHODS

This retrospective study enrolled 434 CRs (normal to abnormal ratio, 246:188) from 378 patients that visited a tertiary hospital. DCNN performance was compared with two radiology residents and two thoracic radiologists. Abnormality assessment (using the area under the receiver operating ch3cteristics (AUROC)) and nodule detection (using jackknife alternative free-response ROC (JAFROC)) were compared among three groups (DCNN only, radiologist without DCNN and radiologist with DCNN). A subset of 56 paired cases, having two CRs taken within a 7-day period, were assessed for intraobserver reproducibility using the intraclass correlation coefficient. Independent characteristics of pulmonary nodules detected by DCNN were assessed by multiple logistic regression analysis.

RESULTS

The AUROC for abnormality detection for the three groups were 0.87, 0.93 and 0.96, respectively (P < 0.05), whereas the JAFROC analysis of nodule detection was 0.926, 0.929 and 0.964. Reproducibility for the three groups was 0.80, 0.67 and 0.80, which shows an increase in radiologists using DCNN (P < 0.05). Nodules detected by DCNN were more solid, round-shaped and well marginated, not masked and laterally located (P < 0.05).

CONCLUSIONS

Extra validation results of DCNN showed high ROC results and there was a significant improvement in the performance when radiologists used DCNN. Reproducibility by DCNN alone showed good agreement, and there was an improvement from moderate to good agreement for radiologists using DCNN.

Intralobar pulmonary sequestration with cystic degeneration mimicking a bronchogenic cyst in an elderly patient: A case report and literature review

INTRODUCTION

Pulmonary sequestration (PS) is a rare congenital malformation defined as nonfunctioning lung tissue supplied by systemic circulation. It is uncommonly diagnosed in adults. Herein, we describe a clinical case of PS with cystic degeneration mimicking a bronchogenic cyst in an elderly patient.

PATIENT CONCERNS

A huge cystic mass was incidentally found in a 65-year-old man on chest computed tomography (CT) scans during preoperative workup for a hand laceration. A 15-cm-sized round cystic mass was detected in the right lower lobe.

DIAGNOSIS

After reviewing the chest CT scan, we decided to perform contrast-enhanced chest magnetic resonance imaging (MRI) and CT-guided lung aspiration biopsy. On MRI, the lesion had the appearance of a cystic mass with hemorrhagic clots, such as an intrapulmonary bronchogenic cyst. The aspirated specimen was nondiagnostic; thus, we decided to surgically remove the mass.

INTERVENTIONS

Upon right lower lobectomy, the mass was diagnosed as a PS. A thin systemic artery supplying the cystic mass was visualized during surgery.

OUTCOMES

The patient is undergoing regular follow-up at the outpatient clinic.

CONCLUSIONS

PS should be considered as a differential diagnosis in patients with a cystic lung mass. Identification of a systemic artery on radiologic imaging is important in the diagnosis of PS before preoperative workup to prevent unpredicted massive bleeding during surgery.

Diagnostic value of using epicardial fat measurement on screening low-dose chest CT for the prediction of metabolic syndrome: A cross-validation study

There has been a marked increase in the use of low-dose computed tomography (LDCT) for lung cancer screening. However, the potential of LDCT to predict metabolic syndrome (MetS) has not been well-documented in this risk-sharing population. We assessed the reliability of epicardial fat volume (EFV) and epicardial fat area (EFA) measurements on chest LDCT for prediction of MetS.A total of 130 (mean age, 50.2 ± 10.77 years) asymptomatic male who underwent nonelectrocardiography (ECG)-gated LDCT were divided into 2 groups for the main analysis (n = 75) and validation (n = 55). Each group was further divided into subgroups with or without MetS. EFV and EFA were calculated semiautomatically using commercially available software with manual assistance. The area under the curve (AUC) on receiver operating characteristic (ROC) analysis and cutoff values to predict MetS on LDCT were then calculated and validated. Female data were not available for analysis due to small sample size in this self-referred lung cancer screening program.In the analysis group, the mean EFV was 123.12 ± 42.29 and 67.30 ± 20.68 cm for the MetS and non-MetS subgroups, respectively (P < .001), and the mean EFA was 7.95 ± 3.10 and 4.04 ± 1.73 cm, respectively (P < .001). Using 93.65 and 4.94 as the cutoffs for EFV and EFA, respectively, the sensitivity, specificity, positive and negative predictive values, and accuracy for predicting MetS were 84.2% and 84.2%, and 92.9% and 64.3% (P < .001); 80% and 44.4% (P = .01); 94.5% and 92.3%; and 90.7% and 69.3% (P < .001), respectively. The AUC for EFV and EFA for predicting MetS was 0.909 and 0.808 (95% confidence interval, 0.819-1.000 and 0.702-0.914, respectively) (P = .02). Using the same cutoff values in the analysis group, there was no significant difference in diagnostic performance using EFV and EFA between the analysis and validation sets.Although quantification of both EFA and EFV is feasible on non-ECG-gated LDCT, EFV may be used to reliably predict MetS with fairly high and better diagnostic performance in selected population.

Myocardial CT perfusion imaging for pre- and postoperative evaluation of myocardial ischemia in a patient with myocardial bridging: A case report

RATIONALE

Although myocardial bridging (MB) is usually considered as benign, initial medical therapy and following surgical treatment in drug-refractory cases has been widely accepted for managing symptomatic MB. Before the patient proceeds to percutaneous or surgical intervention, however, the presence of objective ischemia in the corresponding myocardial territory should be documented.

PATIENT CONCERN AND INTERVENTION

We herein report a 43-year-old male complaining of chest pain in whom cardiac CT with myocardial perfusion (cCTP) showed an MB of left anterior descending artery (LAD) with preoperative perfusion defect in corresponding myocardium and normalization of perfusion after supra-arterial myotomy.

DIAGNOSIS

Myocardial bridging-induced ischemia.

LESSONS

This case illustrates the potential utility of cCTP for the simultaneous assessment of MB and its hemodynamic significance for treatment planning and post-therapeutic evaluation although further research is needed to establish the clinical usefulness of this technique.

The Impact of Iterative Reconstruction in Low-Dose Computed Tomography on the Evaluation of Diffuse Interstitial Lung Disease

OBJECTIVE

To evaluate the impact of iterative reconstruction (IR) on the assessment of diffuse interstitial lung disease (DILD) using CT.

MATERIALS AND METHODS

An American College of Radiology (ACR) phantom (module 4 to assess spatial resolution) was scanned with 10-100 effective mAs at 120 kVp. The images were reconstructed using filtered back projection (FBP), adaptive statistical iterative reconstruction (ASIR), with blending ratios of 0%, 30%, 70% and 100%, and model-based iterative reconstruction (MBIR), and their spatial resolution was objectively assessed by the line pair structure method. The patient study was based on retrospective interpretation of prospectively acquired data, and it was approved by the institutional review board. Chest CT scans of 23 patients (mean age 64 years) were performed at 120 kVp using 1) standard dose protocol applying 142-275 mA with dose modulation (high-resolution computed tomography [HRCT]) and 2) low-dose protocol applying 20 mA (low dose CT, LDCT). HRCT images were reconstructed with FBP, and LDCT images were reconstructed using FBP, ASIR, and MBIR. Matching images were randomized and independently reviewed by chest radiologists. Subjective assessment of disease presence and radiological diagnosis was made on a 10-point scale. In addition, semi-quantitative results were compared for the extent of abnormalities estimated to the nearest 5% of parenchymal involvement.

RESULTS

In the phantom study, ASIR was comparable to FBP in terms of spatial resolution. However, for MBIR, the spatial resolution was greatly decreased under 10 mA. In the patient study, the detection of the presence of disease was not significantly different. The values for area under the curve for detection of DILD by HRCT, FBP, ASIR, and MBIR were as follows: 0.978, 0.979, 0.972, and 0.963. LDCT images reconstructed with FBP, ASIR, and MBIR tended to underestimate reticular or honeycombing opacities (-2.8%, -4.1%, and -5.3%, respectively) and overestimate ground glass opacities (+4.6%, +8.9%, and +8.5%, respectively) compared to the HRCT images. However, the reconstruction methods did not differ with respect to radiologic diagnosis.

CONCLUSION

The diagnostic performance of LDCT with MBIR was similar to that of HRCT in typical DILD cases. However, caution should be exercised when comparing disease extent, especially in follow-up studies with IR.

Single nodular opacity of granulomatous pneumocystis jirovecii pneumonia in an asymptomatic lymphoma patient

The radiologic findings of a single nodule from Pneumocystis jirovecii pneumonia (PJP) have been rarely reported. We described a case of granulomatous PJP manifesting as a solitary pulmonary nodule with a halo sign in a 69-year-old woman with diffuse large B cell lymphoma during chemotherapy. The radiologic appearance of the patient suggested an infectious lesion such as angioinvasive pulmonary aspergillosis or lymphoma involvement of the lung; however, clinical manifestations were not compatible with the diseases. The nodule was confirmed as granulomatous PJP by video-assisted thoracoscopic surgery biopsy.

Ultra-Low-Dose Chest CT in Patients with Neutropenic Fever and Hematologic Malignancy: Image Quality and Its Diagnostic Performance

Purpose

The aim of this study was to evaluate the image quality of ultra-low-dose computed tomography (ULDCT) and its diagnostic performance in making a specific diagnosis of pneumonia in febrile neutropenic patients with hematological malignancy.

Materials and Methods

ULDCT was performed prospectively in 207 febrile neutropenic patients with hematological malignancy. Three observers independently recorded the presence of lung parenchymal abnormality, and also indicated the cause of the lung parenchymal abnormality between infectious and noninfectious causes. If infectious pneumonia was considered the cause of lung abnormalities, they noted the two most appropriate diagnoses among four infectious conditions, including fungal, bacterial, viral, and Pneumocystis pneumonia. Sensitivity for correct diagnoses and receiver operating characteristic (ROC) curve analysis for evaluation of diagnostic accuracy were calculated. Interobserver agreements were determined using intraclass correlation coefficient.

Results

Of 207 patients, 139 (67%) had pneumonia, 12 had noninfectious lung disease, and 56 had no remarkable chest computed tomography (CT) (20 with extrathoracic fever focus and 36 with no specific disease). Mean radiation expose dose of ULDCT was 0.60±0.15 mSv. Each observer regarded low-dose CT scans as unacceptable in only four (1.9%), one (0.5%), and three (1.5%) cases of ULDCTs. Sensitivity and area under the ROC curve in making a specific pneumonia diagnosis were 63.0%, 0.65 for reader 1; 63.0%, 0.61 for reader 2; and 65.0%, 0.62 for reader 3; respectively

conclusion

ULDCT, with a sub-mSv radiation dose and acceptable image quality, provides ready and reasonably acceptable diagnostic information for pulmonary infection in febrile neutropenic patients with hematologic malignancy

DCE and DSC MR perfusion imaging in the differentiation of recurrent tumour from treatment-related changes in patients with glioma

AIM

To retrospectively compare the utility of perfusion magnetic resonance imaging (MRI) in distinguishing treatment-related changes from recurrent disease in glioma patients.

MATERIALS AND METHODS

Thirty-one patients with histologically diagnosed gliomas and increased enhancement after or during concurrent (chemo-) radiation therapy were enrolled. They underwent dynamic contrast-enhanced (DCE) permeability MRI followed by dynamic susceptibility contrast (DSC) perfusion MRI. The vascular transfer constant (rK(trans)) and initial areas under the concentration curve (riAUC) were obtained from DCE MRI, and cerebral blood volume (rCBV) was obtained from DSC MRI. Patients were classified as having treatment-related changes or recurrent tumours based on clinicoradiological results or pathological results from surgery.

RESULTS

Nineteen patients were diagnosed as having recurrences and 12 patients as having treatment-related changes. The rK(trans), riAUC, and rCBV values in the recurrent group were significantly higher than the values in the group with treatment-related changes (p < 0.05). For all 31 patients, there was no significant difference between DSC MRI and DCE MRI for the differentiating power between recurrence and treatment-related changes (p = 0.7227). However, when including only the 24 patients with concordant values of rK(trans) and riAUC, DCE MRI showed a significant AUC value of 0.786 in the receiver operating characteristic (ROC) curve analysis (p = 0.003), whereas DSC MRI did not (AUC = 0.643, p = 0.229).

CONCLUSION

MRI perfusion images appear to show promise in distinguishing treatment-related changes from recurrent tumours. When both rK(trans) and riAUC show concordant values, DCE MRI seems to be more powerful than DSC MRI in the differentiation of recurrence from treatment-related changes.

Interpretation of digital chest radiographs: comparison of light emitting diode versus cold cathode fluorescent lamp backlit monitors

Objective

To compare the diagnostic performance of light emitting diode (LED) backlight monitors and cold cathode fluorescent lamp (CCFL) monitors for the interpretation of digital chest radiographs.

Materials and Methods

We selected 130 chest radiographs from health screening patients. The soft copy image data were randomly sorted and displayed on a 3.5 M LED (2560 × 1440 pixels) monitor and a 3 M CCFL (2048 × 1536 pixels) monitor. Eight radiologists rated their confidence in detecting nodules and abnormal interstitial lung markings (ILD). Low dose chest CT images were used as a reference standard. The performance of the monitor systems was assessed by analyzing 2080 observations and comparing them by multi-reader, multi-case receiver operating characteristic analysis. The observers reported visual fatigue and a sense of heat. Radiant heat and brightness of the monitors were measured.

Results

Measured brightness was 291 cd/m² for the LED and 354 cd/m² for the CCFL monitor. Area under curves for nodule detection were 0.721 ± 0.072 and 0.764 ± 0.098 for LED and CCFL (p = 0.173), whereas those for ILD were 0.871 ± 0.073 and 0.844 ± 0.068 (p = 0.145), respectively. There were no significant differences in interpretation time (p = 0.446) or fatigue score (p = 0.102) between the two monitors. Sense of heat was lower for the LED monitor (p = 0.024). The temperature elevation was 6.7℃ for LED and 12.4℃ for the CCFL monitor.

Conclusion

Although the LED monitor had lower maximum brightness compared with the CCFL monitor, soft copy reading of the digital chest radiographs on LED and CCFL showed no difference in terms of diagnostic performance. In addition, LED emitted less heat.

MR staging accuracy for endometrial cancer based on the new FIGO stage

BACKGROUND

Magnetic resonance imaging (MRI) has been frequently used to determine a preoperative treatment plan for gynecologic cancers. However, the MR accuracy for staging an endometrial cancer is not satisfactory based on the old FIGO staging system.

PURPOSE

To evaluate MR accuracy for staging endometrial cancer using the new FIGO staging system.

MATERIAL AND METHODS

Between January 2005 and May 2009, 199 women underwent surgery due to endometrial cancer. In each patient, an endometrial cancer was staged using MR findings based on the old FIGO staging system and then repeated according to the new FIGO staging system for comparison. Histopathologic findings were used as a standard of reference.

RESULTS

The accuracy of MRI in the staging of endometrial carcinoma stage I, II, III, and IV using the old FIGO staging system were 80% (159/199), 89% (178/199), 90% (179/199), and 99% (198/199), respectively, compared to 87% (174/199), 97% (193/199), 90% (179/199), and 99% (198/199), respectively, when using the new FIGO staging criteria. The overall MR accuracy of the old and new staging systems were 51% (101/199) and 81% (161/199), respectively.

CONCLUSION

MRI has become a more useful tool in the preoperative staging of endometrial cancers using the new FIGO staging system compared to the old one with increased accuracy.

Age and academic-professional honors

This report analyzes variations in the age distribution of individuals at selection to three prominent positions in the academic-professional community. Nobel prizewinner, president of a scientific or professional association, and president of a major research-oriented university. The mean ages of the individuals in each of the categories are compared with those of the presumed pool of scientists. The cross-sectional variations in age at selection seem to be correlated with variations in the levels of codification among the disciplines. There has been a tendency toward older selectees over time. The aging of the pool of scientists cannot fully account for this aging trend. A backlog of candidates for recognition seems to be implicated.