• Korean Journal of Radiology
• /
• v.24 no.6
• /
• pp.541-552
• /
• 2023

Objective: Detection of pneumoperitoneum using abdominal radiography, particularly in the supine position, is often challenging. This study aimed to develop and externally validate a deep learning model for the detection of pneumoperitoneum using supine and erect abdominal radiography. Materials and Methods: A model that can utilize "pneumoperitoneum" and "non-pneumoperitoneum" classes was developed through knowledge distillation. To train the proposed model with limited training data and weak labels, it was trained using a recently proposed semi-supervised learning method called distillation for self-supervised and self-train learning (DISTL), which leverages the Vision Transformer. The proposed model was first pre-trained with chest radiographs to utilize common knowledge between modalities, fine-tuned, and self-trained on labeled and unlabeled abdominal radiographs. The proposed model was trained using data from supine and erect abdominal radiographs. In total, 191212 chest radiographs (CheXpert data) were used for pre-training, and 5518 labeled and 16671 unlabeled abdominal radiographs were used for fine-tuning and self-supervised learning, respectively. The proposed model was internally validated on 389 abdominal radiographs and externally validated on 475 and 798 abdominal radiographs from the two institutions. We evaluated the performance in diagnosing pneumoperitoneum using the area under the receiver operating characteristic curve (AUC) and compared it with that of radiologists. Results: In the internal validation, the proposed model had an AUC, sensitivity, and specificity of 0.881, 85.4%, and 73.3% and 0.968, 91.1, and 95.0 for supine and erect positions, respectively. In the external validation at the two institutions, the AUCs were 0.835 and 0.852 for the supine position and 0.909 and 0.944 for the erect position. In the reader study, the readers' performances improved with the assistance of the proposed model. Conclusion: The proposed model trained with the DISTL method can accurately detect pneumoperitoneum on abdominal radiography in both the supine and erect positions.

• Korean Journal of Radiology
• /
• v.24 no.5
• /
• pp.395-405
• /
• 2023

Objective: This study aimed to develop and validate models using radiomics features on a native T1 map from cardiac magnetic resonance (CMR) to predict left ventricular reverse remodeling (LVRR) in patients with nonischemic dilated cardiomyopathy (NIDCM). Materials and Methods: Data from 274 patients with NIDCM who underwent CMR imaging with T1 mapping at Severance Hospital between April 2012 and December 2018 were retrospectively reviewed. Radiomic features were extracted from the native T1 maps. LVRR was determined using echocardiography performed ≥ 180 days after the CMR. The radiomics score was generated using the least absolute shrinkage and selection operator logistic regression models. Clinical, clinical + late gadolinium enhancement (LGE), clinical + radiomics, and clinical + LGE + radiomics models were built using a logistic regression method to predict LVRR. For internal validation of the result, bootstrap validation with 1000 resampling iterations was performed, and the optimism-corrected area under the receiver operating characteristic curve (AUC) with 95% confidence interval (CI) was computed. Model performance was compared using AUC with the DeLong test and bootstrap. Results: Among 274 patients, 123 (44.9%) were classified as LVRR-positive and 151 (55.1%) as LVRR-negative. The optimism-corrected AUC of the radiomics model in internal validation with bootstrapping was 0.753 (95% CI, 0.698-0.813). The clinical + radiomics model revealed a higher optimism-corrected AUC than that of the clinical + LGE model (0.794 vs. 0.716; difference, 0.078 [99% CI, 0.003-0.151]). The clinical + LGE + radiomics model significantly improved the prediction of LVRR compared with the clinical + LGE model (optimism-corrected AUC of 0.811 vs. 0.716; difference, 0.095 [99% CI, 0.022-0.139]). Conclusion: The radiomic characteristics extracted from a non-enhanced T1 map may improve the prediction of LVRR and offer added value over traditional LGE in patients with NIDCM. Additional external validation research is required.

• Korean Journal of Radiology
• /
• v.24 no.2
• /
• pp.133-144
• /
• 2023

Objective: Cyclin-dependent kinase inhibitor (CDKN)2A/B homozygous deletion is a key molecular marker of isocitrate dehydrogenase (IDH)-mutant astrocytomas in the 2021 World Health Organization. We aimed to investigate whether qualitative and quantitative MRI parameters can predict CDKN2A/B homozygous deletion status in IDH-mutant astrocytomas. Materials and Methods: Preoperative MRI data of 88 patients (mean age ± standard deviation, 42.0 ± 11.9 years; 40 females and 48 males) with IDH-mutant astrocytomas (76 without and 12 with CDKN2A/B homozygous deletion) from two institutions were included. A qualitative imaging assessment was performed. Mean apparent diffusion coefficient (ADC), 5th percentile of ADC, mean normalized cerebral blood volume (nCBV), and 95th percentile of nCBV were assessed via automatic tumor segmentation. Logistic regression was performed to determine the factors associated with CDKN2A/B homozygous deletion in all 88 patients and a subgroup of 47 patients with histological grades 3 and 4. The discrimination performance of the logistic regression models was evaluated using the area under the receiver operating characteristic curve (AUC). Results: In multivariable analysis of all patients, infiltrative pattern (odds ratio [OR] = 4.25, p = 0.034), maximal diameter (OR = 1.07, p = 0.013), and 95th percentile of nCBV (OR = 1.34, p = 0.049) were independent predictors of CDKN2A/B homozygous deletion. The AUC, accuracy, sensitivity, and specificity of the corresponding model were 0.83 (95% confidence interval [CI], 0.72-0.91), 90.4%, 83.3%, and 75.0%, respectively. On multivariable analysis of the subgroup with histological grades 3 and 4, infiltrative pattern (OR = 10.39, p = 0.012) and 95th percentile of nCBV (OR = 1.24, p = 0.047) were independent predictors of CDKN2A/B homozygous deletion, with an AUC accuracy, sensitivity, and specificity of the corresponding model of 0.76 (95% CI, 0.60-0.88), 87.8%, 80.0%, and 58.1%, respectively. Conclusion: The presence of an infiltrative pattern, larger maximal diameter, and higher 95th percentile of the nCBV may be useful MRI biomarkers for CDKN2A/B homozygous deletion in IDH-mutant astrocytomas.

• Korean Journal of Radiology
• /
• v.23 no.10
• /
• pp.939-948
• /
• 2022

Objective: Evidence supports the efficacy of coronary computed tomography angiography (CCTA)-based risk scores in cardiovascular risk stratification of patients with suspected coronary artery disease (CAD). We aimed to compare two CCTA-based risk score algorithms, Leiden and Confirm scores, in patients with diabetes mellitus (DM) and suspected CAD. Materials and Methods: This single-center prospective cohort study consecutively included 1241 DM patients (54.1% male, 60.2 ± 10.4 years) referred for CCTA for suspected CAD in 2015-2017. Leiden and Confirm scores were calculated and stratified as < 5 (reference), 5-20, and > 20 for Leiden and < 14.3 (reference), 14.3-19.5, and > 19.5 for Confirm. Major adverse cardiovascular events (MACE) were defined as the composite outcomes of cardiovascular death, nonfatal myocardial infarction (MI), stroke, and unstable angina requiring hospitalization. The Cox model and Kaplan-Meier method were used to evaluate the effect size of the risk scores on MACE. The area under the curve (AUC) at the median follow-up time was also compared between score algorithms. Results: During a median follow-up of 31 months (interquartile range, 27.6-37.3 months), 131 of MACE were recorded, including 17 cardiovascular deaths, 28 nonfatal MIs, 64 unstable anginas requiring hospitalization, and 22 strokes. An incremental incidence of MACE was observed in both Leiden and Confirm scores, with an increase in the scores (log-rank p < 0.001). In the multivariable analysis, compared with Leiden score < 5, the hazard ratios for Leiden scores of 5-20 and > 20 were 2.37 (95% confidence interval [CI]: 1.53-3.69; p < 0.001) and 4.39 (95% CI: 2.40-8.01; p < 0.001), respectively, while the Confirm score did not demonstrate a statistically significant association with the risk of MACE. The Leiden score showed a greater AUC of 0.840 compared to 0.777 for the Confirm score (p < 0.001). Conclusion: CCTA-based risk score algorithms could be used as reliable cardiovascular risk predictors in patients with DM and suspected CAD, among which the Leiden score outperformed the Confirm score in predicting MACE.

• Korean Journal of Radiology
• /
• v.23 no.10
• /
• pp.949-958
• /
• 2022

Objective: To investigate the feasibility of using a deep learning-based analysis of auscultation data to predict significant stenosis of arteriovenous fistulas (AVF) in patients undergoing hemodialysis requiring percutaneous transluminal angioplasty (PTA). Materials and Methods: Forty patients (24 male and 16 female; median age, 62.5 years) with dysfunctional native AVF were prospectively recruited. Digital sounds from the AVF shunt were recorded using a wireless electronic stethoscope before (pre-PTA) and after PTA (post-PTA), and the audio files were subsequently converted to mel spectrograms, which were used to construct various deep convolutional neural network (DCNN) models (DenseNet201, EfficientNetB5, and ResNet50). The performance of these models for diagnosing ≥ 50% AVF stenosis was assessed and compared. The ground truth for the presence of ≥ 50% AVF stenosis was obtained using digital subtraction angiography. Gradient-weighted class activation mapping (Grad-CAM) was used to produce visual explanations for DCNN model decisions. Results: Eighty audio files were obtained from the 40 recruited patients and pooled for the study. Mel spectrograms of "pre-PTA" shunt sounds showed patterns corresponding to abnormal high-pitched bruits with systolic accentuation observed in patients with stenotic AVF. The ResNet50 and EfficientNetB5 models yielded an area under the receiver operating characteristic curve of 0.99 and 0.98, respectively, at optimized epochs for predicting ≥ 50% AVF stenosis. However, Grad-CAM heatmaps revealed that only ResNet50 highlighted areas relevant to AVF stenosis in the mel spectrogram. Conclusion: Mel spectrogram-based DCNN models, particularly ResNet50, successfully predicted the presence of significant AVF stenosis requiring PTA in this feasibility study and may potentially be used in AVF surveillance.

• Korean Journal of Radiology
• /
• v.23 no.10
• /
• pp.986-997
• /
• 2022

Objective: Whether metabolic redistribution occurs in patients with white matter hyperintensities (WMHs) on magnetic resonance imaging (MRI) is unknown. This study aimed 1) to propose a measure of the brain metabolic network for an individual patient and preliminarily apply it to identify impaired metabolic networks in patients with WMHs, and 2) to explore the clinical and imaging features of metabolic redistribution in patients with WMHs. Materials and Methods: This study included 50 patients with WMHs and 70 healthy controls (HCs) who underwent ¹⁸F-fluorodeoxyglucose-positron emission tomography/MRI. Various global property parameters according to graph theory and an individual parameter of brain metabolic network called "individual contribution index" were obtained. Parameter values were compared between the WMH and HC groups. The performance of the parameters in discriminating between the two groups was assessed using the area under the receiver operating characteristic curve (AUC). The correlation between the individual contribution index and Fazekas score was assessed, and the interaction between age and individual contribution index was determined. A generalized linear model was fitted with the individual contribution index as the dependent variable and the mean standardized uptake value (SUV_mean) of nodes in the whole-brain network or seven classic functional networks as independent variables to determine their association. Results: The means ± standard deviations of the individual contribution index were (0.697 ± 10.9) × 10^-3 and (0.0967 ± 0.0545) × 10^-3 in the WMH and HC groups, respectively (p < 0.001). The AUC of the individual contribution index was 0.864 (95% confidence interval, 0.785-0.943). A positive correlation was identified between the individual contribution index and the Fazekas scores in patients with WMHs (r = 0.57, p < 0.001). Age and individual contribution index demonstrated a significant interaction effect on the Fazekas score. A significant direct association was observed between the individual contribution index and the SUV_mean of the limbic network (p < 0.001). Conclusion: The individual contribution index may demonstrate the redistribution of the brain metabolic network in patients with WMHs.

• Korean Journal of Radiology
• /
• v.23 no.5
• /
• pp.505-516
• /
• 2022

Objective: To evaluate whether artificial intelligence (AI) for detecting breast cancer on mammography can improve the performance and time efficiency of radiologists reading mammograms. Materials and Methods: A commercial deep learning-based software for mammography was validated using external data collected from 200 patients, 100 each with and without breast cancer (40 with benign lesions and 60 without lesions) from one hospital. Ten readers, including five breast specialist radiologists (BSRs) and five general radiologists (GRs), assessed all mammography images using a seven-point scale to rate the likelihood of malignancy in two sessions, with and without the aid of the AI-based software, and the reading time was automatically recorded using a web-based reporting system. Two reading sessions were conducted with a two-month washout period in between. Differences in the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and reading time between reading with and without AI were analyzed, accounting for data clustering by readers when indicated. Results: The AUROC of the AI alone, BSR (average across five readers), and GR (average across five readers) groups was 0.915 (95% confidence interval, 0.876-0.954), 0.813 (0.756-0.870), and 0.684 (0.616-0.752), respectively. With AI assistance, the AUROC significantly increased to 0.884 (0.840-0.928) and 0.833 (0.779-0.887) in the BSR and GR groups, respectively (p = 0.007 and p < 0.001, respectively). Sensitivity was improved by AI assistance in both groups (74.6% vs. 88.6% in BSR, p < 0.001; 52.1% vs. 79.4% in GR, p < 0.001), but the specificity did not differ significantly (66.6% vs. 66.4% in BSR, p = 0.238; 70.8% vs. 70.0% in GR, p = 0.689). The average reading time pooled across readers was significantly decreased by AI assistance for BSRs (82.73 vs. 73.04 seconds, p < 0.001) but increased in GRs (35.44 vs. 42.52 seconds, p < 0.001). Conclusion: AI-based software improved the performance of radiologists regardless of their experience and affected the reading time.

• Korean Journal of Radiology
• /
• v.24 no.10
• /
• pp.983-995
• /
• 2023

Objective: To compare the diagnostic performance and interobserver agreement between contrast-enhanced computed tomography (CECT) and contrast-enhanced magnetic resonance imaging (CE-MRI) with magnetic resonance cholangiopancreatography (MRCP) for evaluating the resectability in patients with extrahepatic cholangiocarcinoma (eCCA). Materials and Methods: This retrospective study included treatment-naïve patients with pathologically confirmed eCCA, who underwent both CECT and CE-MRI with MRCP using extracellular contrast media between January 2015 and December 2020. Among the 214 patients (146 males; mean age ± standard deviation, 68 ± 9 years) included, 121 (56.5%) had perihilar cholangiocarcinoma. R0 resection was achieved in 108 of the 153 (70.6%) patients who underwent curative-intent surgery. Four fellowship-trained radiologists independently reviewed the findings of both CECT and CE-MRI with MRCP to assess the local tumor extent and distant metastasis for determining resectability. The pooled area under the receiver operating characteristic curve (AUC), sensitivity, and specificity of CECT and CE-MRI with MRCP were compared using clinical, surgical, and pathological findings as reference standards. The interobserver agreement of resectability was evaluated using Fleiss kappa (κ). Results: No significant differences were observed between CECT and CE-MRI with MRCP in the pooled AUC (0.753 vs. 0.767), sensitivity (84.7% [366/432] vs. 90.3% [390/432]), and specificity (52.6% [223/424] vs. 51.4% [218/424]) (P > 0.05 for all). The AUC for determining resectability was higher when CECT and CE-MRI with MRCP were reviewed together than when CECT was reviewed alone in patients with discrepancies between the imaging modalities or with indeterminate resectability (0.798 [0.754-0.841] vs. 0.753 [0.697-0.808], P = 0.014). The interobserver agreement for overall resectability was fair for both CECT (κ = 0.323) and CE-MRI with MRCP (κ = 0.320), without a significant difference (P = 0.884). Conclusion: CECT and CE-MRI with MRCP showed no significant differences in the diagnostic performance and interobserver agreement in determining the resectability in patients with eCCA.

• Korean Journal of Radiology
• /
• v.23 no.4
• /
• pp.466-478
• /
• 2022

Objective: ¹⁸F-fluorodeoxyglucose (FDG) PET/CT is often used for detecting malignancy in patients with newly diagnosed hemophagocytic lymphohistiocytosis (HLH), with acceptable sensitivity but relatively low specificity. The aim of this study was to improve the diagnostic ability of ¹⁸F-FDG PET/CT in identifying malignancy in patients with HLH by combining ¹⁸F-FDG PET/CT and clinical parameters. Materials and Methods: Ninety-seven patients (age ≥ 14 years) with secondary HLH were retrospectively reviewed and divided into the derivation (n = 71) and validation (n = 26) cohorts according to admission time. In the derivation cohort, 22 patients had malignancy-associated HLH (M-HLH) and 49 patients had non-malignancy-associated HLH (NM-HLH). Data on pretreatment ¹⁸F-FDG PET/CT and laboratory results were collected. The variables were analyzed using the Mann-Whitney U test or Pearson's chi-square test, and a nomogram for predicting M-HLH was constructed using multivariable binary logistic regression. The predictors were also ranked using decision-tree analysis. The nomogram and decision tree were validated in the validation cohort (10 patients with M-HLH and 16 patients with NM-HLH). Results: The ratio of the maximal standardized uptake value (SUVmax) of the lymph nodes to that of the mediastinum, the ratio of the SUVmax of bone lesions or bone marrow to that of the mediastinum, and age were selected for constructing the model. The nomogram showed good performance in predicting M-HLH in the validation cohort, with an area under the receiver operating characteristic curve of 0.875 (95% confidence interval, 0.686-0.971). At an appropriate cutoff value, the sensitivity and specificity for identifying M-HLH were 90% (9/10) and 68.8% (11/16), respectively. The decision tree integrating the same variables showed 70% (7/10) sensitivity and 93.8% (15/16) specificity for identifying M-HLH. In comparison, visual analysis of ¹⁸F-FDG PET/CT images demonstrated 100% (10/10) sensitivity and 12.5% (2/16) specificity. Conclusion: ¹⁸F-FDG PET/CT may be a practical technique for identifying M-HLH. The model constructed using ¹⁸F-FDG PET/CT features and age was able to detect malignancy with better accuracy than visual analysis of ¹⁸F-FDG PET/CT images.

• Korean Journal of Radiology
• /
• v.23 no.11
• /
• pp.1078-1088
• /
• 2022

Objective: To develop and validate a model using radiomics features from apparent diffusion coefficient (ADC) map to diagnose local tumor recurrence in head and neck squamous cell carcinoma (HNSCC). Materials and Methods: This retrospective study included 285 patients (mean age ± standard deviation, 62 ± 12 years; 220 male, 77.2%), including 215 for training (n = 161) and internal validation (n = 54) and 70 others for external validation, with newly developed contrast-enhancing lesions at the primary cancer site on the surveillance MRI following definitive treatment of HNSCC between January 2014 and October 2019. Of the 215 and 70 patients, 127 and 34, respectively, had local tumor recurrence. Radiomics models using radiomics scores were created separately for T2-weighted imaging (T2WI), contrast-enhanced T1-weighted imaging (CE-T1WI), and ADC maps using non-zero coefficients from the least absolute shrinkage and selection operator in the training set. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance of each radiomics score and known clinical parameter (age, sex, and clinical stage) in the internal and external validation sets. Results: Five radiomics features from T2WI, six from CE-T1WI, and nine from ADC maps were selected and used to develop the respective radiomics models. The area under ROC curve (AUROC) of ADC radiomics score was 0.76 (95% confidence interval [CI], 0.62-0.89) and 0.77 (95% CI, 0.65-0.88) in the internal and external validation sets, respectively. These were significantly higher than the AUROC values of T2WI (0.53 [95% CI, 0.40-0.67], p = 0.006), CE-T1WI (0.53 [95% CI, 0.40-0.67], p = 0.012), and clinical parameters (0.53 [95% CI, 0.39-0.67], p = 0.021) in the external validation set. Conclusion: The radiomics model using ADC maps exhibited higher diagnostic performance than those of the radiomics models using T2WI or CE-T1WI and clinical parameters in the diagnosis of local tumor recurrence in HNSCC following definitive treatment.