Effects of Implementing Artificial Intelligence-Based Computer-Aided Detection for Chest Radiographs in Daily Practice on the Rate of Referral to Chest Computed Tomography in Pulmonology Outpatient Clinic

Objective@#The clinical impact of artificial intelligence-based computer-aided detection (AI-CAD) beyond diagnostic accuracy remains uncertain. We aimed to investigate the influence of the clinical implementation of AI-CAD for chest radiograph (CR) interpretation in daily practice on the rate of referral for chest computed tomography (CT). @*Materials and Methods@#AI-CAD was implemented in clinical practice at the Seoul National University Hospital. CRs obtained from patients who visited the pulmonology outpatient clinics before (January–December 2019) and after (January–December 2020) implementation were included in this study. After implementation, the referring pulmonologist requested CRs with or without AI-CAD analysis. We conducted multivariable logistic regression analyses to evaluate the associations between using AI-CAD and the following study outcomes: the rate of chest CT referral, defined as request and actual acquisition of chest CT within 30 days after CR acquisition, and the CT referral rates separately for subsequent positive and negative CT results.Multivariable analyses included various covariates such as patient age and sex, time of CR acquisition (before versus after AICAD implementation), referring pulmonologist, nature of the CR examination (baseline versus follow-up examination), and radiology reports presence at the time of the pulmonology visit. @*Results@#A total of 28546 CRs from 14565 patients (mean age: 67 years; 7130 males) and 25888 CRs from 12929 patients (mean age: 67 years; 6435 males) before and after AI-CAD implementation were included. The use of AI-CAD was independently associated with increased chest CT referrals (odds ratio [OR], 1.33; P = 0.008) and referrals with subsequent negative chest CT results (OR, 1.46; P = 0.005). Meanwhile, referrals with positive chest CT results were not significantly associated with AI-CAD use (OR, 1.08; P = 0.647). @*Conclusion@#The use of AI-CAD for CR interpretation in pulmonology outpatients was independently associated with an increased frequency of overall referrals for chest CT scans and referrals with subsequent negative results.

Conventional Versus Artificial Intelligence-Assisted Interpretation of Chest Radiographs in Patients With Acute Respiratory Symptoms in Emergency Department: A Pragmatic Randomized Clinical Trial

Objective@#It is unknown whether artificial intelligence-based computer-aided detection (AI-CAD) can enhance the accuracy of chest radiograph (CR) interpretation in real-world clinical practice. We aimed to compare the accuracy of CR interpretation assisted by AI-CAD to that of conventional interpretation in patients who presented to the emergency department (ED) with acute respiratory symptoms using a pragmatic randomized controlled trial. @*Materials and Methods@#Patients who underwent CRs for acute respiratory symptoms at the ED of a tertiary referral institution were randomly assigned to intervention group (with assistance from an AI-CAD for CR interpretation) or control group (without AI assistance). Using a commercial AI-CAD system (Lunit INSIGHT CXR, version 2.0.2.0; Lunit Inc.). Other clinical practices were consistent with standard procedures. Sensitivity and false-positive rates of CR interpretation by duty trainee radiologists for identifying acute thoracic diseases were the primary and secondary outcomes, respectively. The reference standards for acute thoracic disease were established based on a review of the patient’s medical record at least 30 days after the ED visit. @*Results@#We randomly assigned 3576 participants to either the intervention group (1761 participants; mean age ± standard deviation, 65 ± 17 years; 978 males; acute thoracic disease in 472 participants) or the control group (1815 participants; 64 ± 17 years; 988 males; acute thoracic disease in 491 participants). The sensitivity (67.2% [317/472] in the intervention group vs. 66.0% [324/491] in the control group; odds ratio, 1.02 [95% confidence interval, 0.70–1.49]; P = 0.917) and false-positive rate (19.3% [249/1289] vs. 18.5% [245/1324]; odds ratio, 1.00 [95% confidence interval, 0.79–1.26]; P = 0.985) of CR interpretation by duty radiologists were not associated with the use of AI-CAD. @*Conclusion@#AI-CAD did not improve the sensitivity and false-positive rate of CR interpretation for diagnosing acute thoracic disease in patients with acute respiratory symptoms who presented to the ED.

Effects of Expert-Determined Reference Standards in Evaluating the Diagnostic Performance of a Deep Learning Model: A Malignant Lung Nodule Detection Task on Chest Radiographs

Objective@#Little is known about the effects of using different expert-determined reference standards when evaluating the performance of deep learning-based automatic detection (DLAD) models and their added value to radiologists. We assessed the concordance of expert-determined standards with a clinical gold standard (herein, pathological confirmation) and the effects of different expert-determined reference standards on the estimates of radiologists’ diagnostic performance to detect malignant pulmonary nodules on chest radiographs with and without the assistance of a DLAD model. @*Materials and Methods@#This study included chest radiographs from 50 patients with pathologically proven lung cancer and 50 controls. Five expert-determined standards were constructed using the interpretations of 10 experts: individual judgment by the most experienced expert, majority vote, consensus judgments of two and three experts, and a latent class analysis (LCA) model. In separate reader tests, additional 10 radiologists independently interpreted the radiographs and then assisted with the DLAD model. Their diagnostic performance was estimated using the clinical gold standard and various expertdetermined standards as the reference standard, and the results were compared using the t test with Bonferroni correction. @*Results@#The LCA model (sensitivity, 72.6%; specificity, 100%) was most similar to the clinical gold standard. When expertdetermined standards were used, the sensitivities of radiologists and DLAD model alone were overestimated, and their specificities were underestimated (all p-values < 0.05). DLAD assistance diminished the overestimation of sensitivity but exaggerated the underestimation of specificity (all p-values < 0.001). The DLAD model improved sensitivity and specificity to a greater extent when using the clinical gold standard than when using the expert-determined standards (all p-values < 0.001), except for sensitivity with the LCA model (p = 0.094). @*Conclusion@#The LCA model was most similar to the clinical gold standard for malignant pulmonary nodule detection on chest radiographs. Expert-determined standards caused bias in measuring the diagnostic performance of the artificial intelligence model.

2020 Clinical Practice Guideline for Percutaneous Transthoracic Needle Biopsy of Pulmonary Lesions: A Consensus Statement and Recommendations of the Korean Society of Thoracic Radiology

Percutaneous transthoracic needle biopsy (PTNB) is one of the essential diagnostic procedures for pulmonary lesions. Its role is increasing in the era of CT screening for lung cancer and precision medicine. The Korean Society of Thoracic Radiology developed the first evidence-based clinical guideline for PTNB in Korea by adapting pre-existing guidelines. The guideline provides 39 recommendations for the following four main domains of 12 key questions: the indications for PTNB, pre-procedural evaluation, procedural technique of PTNB and its accuracy, and management of post-biopsy complications. We hope that these recommendations can improve the diagnostic accuracy and safety of PTNB in clinical practice and promote standardization of the procedure nationwide.

Implementation of a Deep Learning-Based ComputerAided Detection System for the Interpretation of Chest Radiographs in Patients Suspected for COVID-19

Objective@#To describe the experience of implementing a deep learning-based computer-aided detection (CAD) system for the interpretation of chest X-ray radiographs (CXR) of suspected coronavirus disease (COVID-19) patients and investigate the diagnostic performance of CXR interpretation with CAD assistance. @*Materials and Methods@#In this single-center retrospective study, initial CXR of patients with suspected or confirmed COVID-19 were investigated. A commercialized deep learning-based CAD system that can identify various abnormalities on CXR was implemented for the interpretation of CXR in daily practice. The diagnostic performance of radiologists with CAD assistance were evaluated based on two different reference standards: 1) real-time reverse transcriptase-polymerase chain reaction (rRT-PCR) results for COVID-19 and 2) pulmonary abnormality suggesting pneumonia on chest CT. The turnaround times (TATs) of radiology reports for CXR and rRT-PCR results were also evaluated. @*Results@#Among 332 patients (male:female, 173:159; mean age, 57 years) with available rRT-PCR results, 16 patients (4.8%) were diagnosed with COVID-19. Using CXR, radiologists with CAD assistance identified rRT-PCR positive COVID-19 patients with sensitivity and specificity of 68.8% and 66.7%, respectively. Among 119 patients (male:female, 75:44; mean age, 69 years) with available chest CTs, radiologists assisted by CAD reported pneumonia on CXR with a sensitivity of 81.5% and a specificity of 72.3%. The TATs of CXR reports were significantly shorter than those of rRT-PCR results (median 51 vs. 507 minutes; p < 0.001). @*Conclusion@#Radiologists with CAD assistance could identify patients with rRT-PCR-positive COVID-19 or pneumonia on CXR with a reasonably acceptable performance. In patients suspected with COVID-19, CXR had much faster TATs than rRT-PCRs.

Clinical Implementation of Deep Learning in ThoracicRadiology: Potential Applications and Challenges

Chest X-ray radiography and computed tomography, the two mainstay modalities in thoracic radiology, are under activeinvestigation with deep learning technology, which has shown promising performance in various tasks, including detection,classification, segmentation, and image synthesis, outperforming conventional methods and suggesting its potential forclinical implementation. However, the implementation of deep learning in daily clinical practice is in its infancy and facingseveral challenges, such as its limited ability to explain the output results, uncertain benefits regarding patient outcomes, andincomplete integration in daily workflow. In this review article, we will introduce the potential clinical applications of deeplearning technology in thoracic radiology and discuss several challenges for its implementation in daily clinical practice.

Quantitative Thoracic Magnetic Resonance Criteria for the Differentiation of Cysts from Solid Masses in the Anterior Mediastinum

OBJECTIVE: To evaluate quantitative magnetic resonance imaging (MRI) parameters for differentiation of cysts from and solid masses in the anterior mediastinum. MATERIALS AND METHODS: The development dataset included 18 patients from two institutions with pathologically-proven cysts (n = 6) and solid masses (n = 12) in the anterior mediastinum. We measured the maximum diameter, normalized T1 and T2 signal intensity (nT1 and nT2), normalized apparent diffusion coefficient (nADC), and relative enhancement ratio (RER) of each lesion. RERs were obtained by non-rigid registration and subtraction of precontrast and postcontrast T1-weighted images. Differentiation criteria between cysts and solid masses were identified based on receiver operating characteristics analysis. For validation, two separate datasets were utilized: 15 patients with 8 cysts and 7 solid masses from another institution (validation dataset 1); and 11 patients with clinically diagnosed cysts stable for more than two years (validation dataset 2). Sensitivity and specificity were calculated from the validation datasets. RESULTS: nT2, nADC, and RER significantly differed between cysts and solid masses (p = 0.032, 0.013, and 0.63; nT2 > 0.39. In validation dataset 1, the sensitivity of the RER, nADC, and nT2 criteria was 87.5%, 100%, and 75.0%, and the specificity was 100%, 40.0%, and 57.4%, respectively. In validation dataset 2, the sensitivity of the RER, nADC, and nT2 criteria was 90.9%, 90.9%, and 72.7%, respectively. CONCLUSION: Quantitative MRI criteria using nT2, nADC, and particularly RER can assist differentiation of cysts from solid masses in the anterior mediastinum.

Significant Abnormalities Other than Lung Cancer in Korean Lung Cancer CT Screening

A low-dose chest CT is performed for early detection of lung cancer, but the CT scan frequently shows several incidental abnormalities. Identification of the incidental findings may enable early detection of diseases other than lung cancer, thereby improving the survival of the individual undergoing screening. However, insignificant incidental abnormalities may cause unnecessary additional examination and costs. It is crucial for radiologists to appropriately comprehend and report significant incidental abnormalities other than lung cancer for successful implementation of the national lung cancer screening program in Korea.

Open Bronchus Sign on CT: A Risk Factor for Hemoptysis after Percutaneous Transthoracic Biopsy

OBJECTIVE: We hypothesized that open bronchi within target pulmonary lesions are associated with percutaneous transthoracic needle biopsy (PTNB)-related hemoptysis. We sought to analyze and compare patient characteristics and target features as well as biopsy-related factors between patients with and without PTNB-related hemoptysis. MATERIALS AND METHODS: We retrospectively analyzed 1484 patients (870 males and 614 females; median age, 66 years) who had undergone 1569 cone-beam CT (CBCT)-guided PTNBs. Patient characteristics (sex, age, and pathologic diagnosis), nodule features (nodule type, size, location, and presence of an open bronchus in target nodules), and biopsy-related factors (biopsy needle size, pleura-to-target distance, blood test results, open bronchus unavoidability [OBU] index, etc.) were investigated. OBU index, which was assessed using the pre-procedural CBCT, was a subjective scoring system for the probability of needle penetration into the open bronchus. Univariate analysis and subsequent multivariate logistic regression analysis were conducted to reveal the independent risk factors for PTNB-related hemoptysis. For a subgroup of nodules with open bronchi, a trend analysis between the occurrence of hemoptysis and the OBU index was performed. RESULTS: The independent risk factors for hemoptysis were sex (female; odds ratio [OR], 1.918; p < 0.001), nodule size (OR, 0.837; p < 0.001), open bronchus (OR, 2.101; p < 0.001), and pleura-to-target distance (OR, 1.135; p = 0.003). For the target nodules with open bronchi, a significant trend between hemoptysis and OBU index (p < 0.001) was observed. CONCLUSION: An open bronchus in a biopsy target is an independent predictor of hemoptysis, and careful imaging review may potentially reduce PTNB-related hemoptysis.

Digital Tomosynthesis for Evaluating Metastatic Lung Nodules: Nodule Visibility, Learning Curves, and Reading Times

OBJECTIVE: To evaluate nodule visibility, learning curves, and reading times for digital tomosynthesis (DT). MATERIALS AND METHODS: We included 80 patients who underwent computed tomography (CT) and DT before pulmonary metastasectomy. One experienced chest radiologist annotated all visible nodules on thin-section CT scans using computer-aided detection software. Two radiologists used CT as the reference standard and retrospectively graded the visibility of nodules on DT. Nodule detection performance was evaluated in four sessions of 20 cases each by six readers. After each session, readers were unblinded to the DT images by revealing the true-positive markings and were instructed to self-analyze their own misreads. Receiver-operating-characteristic curves were determined. RESULTS: Among 414 nodules on CT, 53.3% (221/414) were visible on DT. The main reason for not seeing a nodule on DT was small size (93.3%, 0.8, and the average detection rate for malignant nodules was 85% (210/246). The inter-session analysis of the AUC showed no significant differences among the readers, and the detection rate for malignant nodules did not differ across sessions. A slight improvement in reading times was observed. CONCLUSION: Most malignant nodules > 5 mm were visible on DT. As nodule detection performance was high from the initial session, DT may be readily applicable for radiology residents and board-certified radiologists.

Monopolar Radiofrequency Ablation Using a Dual-Switching System and a Separable Clustered Electrode: Evaluation of the In Vivo Efficiency

OBJECTIVE: To determine the in vivo efficiency of monopolar radiofrequency ablation (RFA) using a dual-switching (DS) system and a separable clustered (SC) electrode to create coagulation in swine liver. MATERIALS AND METHODS: Thirty-three ablation zones were created in nine pigs using a DS system and an SC electrode in the switching monopolar mode. The pigs were divided into two groups for two experiments: 1) preliminary experiments (n = 3) to identify the optimal inter-electrode distances (IEDs) for dual-switching monopolar (DSM)-RFA, and 2) main experiments (n = 6) to compare the in vivo efficiency of DSM-RFA with that of a single-switching monopolar (SSM)-RFA. RF energy was alternatively applied to one of the three electrodes (SSM-RFA) or concurrently applied to a pair of electrodes (DSM-RFA) for 12 minutes in in vivo porcine livers. The delivered RFA energy and the shapes and dimensions of the coagulation areas were compared between the two groups. RESULTS: No pig died during RFA. The ideal IEDs for creating round or oval coagulation area using the DSM-RFA were 2.0 and 2.5 cm. DSM-RFA allowed more efficient RF energy delivery than SSM-RFA at the given time (23.0 +/- 4.0 kcal vs. 16.92 +/- 2.0 kcal, respectively; p = 0.0005). DSM-RFA created a significantly larger coagulation volume than SSM-RFA (40.4 +/- 16.4 cm3 vs. 20.8 +/- 10.7 cm3; p < 0.001). Both groups showed similar circularity of the ablation zones (p = 0.29). CONCLUSION: Dual-switching monopolar-radiofrequency ablation using an SC electrode is feasible and can create larger ablation zones than SSM-RFA as it allows more RF energy delivery at a given time.

Portable high-intensity focused ultrasound system with 3D electronic steering, real-time cavitation monitoring, and 3D image reconstruction algorithms: a preclinical study in pigs

PURPOSE: The aim of this study was to evaluate the safety and accuracy of a new portable ultrasonography-guided high-intensity focused ultrasound (USg-HIFU) system with a 3-dimensional (3D) electronic steering transducer, a simultaneous ablation and imaging module, real-time cavitation monitoring, and 3D image reconstruction algorithms. METHODS: To address the accuracy of the transducer, hydrophones in a water chamber were used to assess the generation of sonic fields. An animal study was also performed in five pigs by ablating in vivo thighs by single-point sonication (n=10) or volume sonication (n=10) and ex vivo kidneys by single-point sonication (n=10). Histological and statistical analyses were performed. RESULTS: In the hydrophone study, peak voltages were detected within 1.0 mm from the targets on the y- and z-axes and within 2.0-mm intervals along the x-axis (z-axis, direction of ultrasound propagation; y- and x-axes, perpendicular to the direction of ultrasound propagation). Twenty-nine of 30 HIFU sessions successfully created ablations at the target. The in vivo porcine thigh study showed only a small discrepancy (width, 0.5-1.1 mm; length, 3.0 mm) between the planning ultrasonograms and the pathological specimens. Inordinate thermal damage was not observed in the adjacent tissues or sonic pathways in the in vivo thigh and ex vivo kidney studies. CONCLUSION: Our study suggests that this new USg-HIFU system may be a safe and accurate technique for ablating soft tissues and encapsulated organs.

The Protective Effects of Allopurinol, PAF Antagonist, LT Inhibitor, Superoxide Dismutase-Polyethylene Glycol, and Dexamethasone on Noise Induced Shift of Hearing Threshold in Guinea Pigs / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: Prevention of noise induced hearing loss (NIHL) can be achieved by environmental control, however, recently, there are some suggestions for the protective effects of various drugs. To find an effective drug on the protection of NIHL, we investigated the protective effects of allopurinol, platelet activating factor (PAF) antagonist, leukotriene (LT) inhibitor, superoxide dismutase-polyethylene glycol (SOD-PEG), and dexamethasone on the noise induced threshold shift in guinea pigs. MAERIALS AND METHODS: Thirty healthy guinea pigs (60 ears) were divided into control and different treatment groups to consist of five guinea pigs each. Guinea pigs were exposed to 100 dBA white noise for eight hours. Auditory Brainstem Response (ABR) were recorded before and after the noise exposure. RESULTS: Changes in ABR thresholds were 48.1+/-9.6 dB SPL in the control group, 33.1+/-3.7 dB SPL in the allopurinol group, 39.37+/-17.8 dB SPL in the PAF antagonist group, 36.25+/-10.26 dB SPL in the LT inhibitor group, 38.0+/-13.98 dB SPL in the SOD-PEG group and 46.9+/-8.8 dB SPL in the dexamethasone group. Only the allopurinol group demonstrated significantly different threshold change compared to the control group. CONCLUSION: The results of this study suggest that allopurinol may be effective in protection of noise induced threshold shift, whereas PAF antagonist, LT inhibitor, SOD, dexamethasone may not be as effective.

Properties of SOAEs and Their Correlation with TEOAEs in Neonates / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: Transient evoked otoacoustic emissions (TEOAEs) have been used as a screening test for early detection of hearing loss in newborns and infants, because most of the normal hearing show TEOAEs. But the lack of understanding about the properties of spontaneous otoacoustic emissions (SOAEs) has limited their usefulness on the clinical basis. This study was designed with the purpose to find out the properties of SOAEs and their correlation with TEOAEs. MATERIALS AND METHODS: TEOAE and SOAE tests were performed in 95 neonates (190 ears, male:female=47:48) of well baby clinic in DKUH. RESULTS: 1) The prevalence of SOAEs were 86.8% (165/190 ears):87.2% (82/94 ears) in male and 86.5% (83/96 ears) in female, 91% (86/95 ears) in the right ears and 83% (75/95 ears) in the left ears. 2) The number of peaks in SOAEs was 4.14+/-1.92 overally, 4.38+/-2.01 in right ears, 3.87+/-1.79 in left ears, and 4.34+/-1.95 in males and 3.94+/-1.88 in females. There was a significant difference between the right and the left ears. 3) The level of SOAEs was -18.4 to 22.6 dB SPL. The highest amplitude of SOAEs in each subject was 3.69+/-8.87 dB SPL. 4) The SOAE frequency was mainly between 1000 Hz and 4000 Hz, and most frequently between 3000 Hz and 4000 Hz. The mean frequency of the highest amplitude of SOAE was 2907+/-1092 Hz. 5) The highest amplitude of SOAE was correlated with the overall response amplitude of TEOAE (r=0.619, p<0.01), but not the number of peaks (r=0.086, p=0.272). CONCLUSION: Neonates have robust SOAEs, which correlated with their TEOAEs.

Effect of IL-1beta on Hearing and Penetration through Round Window Membrane in Guinea Pig / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: Interleukin (IL)-1beta has been detected in human middle ear effusion (MEE) and the MEEs of experimental otitis media. The object of this study was to investigate whether this cytokine 1) penetrates through the round window membrane (RWM) and 2) induces hearing loss. MATERIALS AND METHODS: Guinea pigs were randomly assigned as control and IL-1beta groups and each was subdivided into three different groups by applying 1 ng, 10 ng, 100 ng of IL-1beta on RWM. They underwent hearing test by auditory brainstem response (ABR) prior to experiments. In the ontrol group, 5nL of phosphate buffered saline (PBS) soaked in a small piece of gelfoam was applied on the RWM. In IL-1beta groups, three different concentrations of IL-1beta 1 ng, 10 ng, 100 ng soaked in a small piece of gelfoam were applied on the RWM. ABR measurements were performed at 3 and 6 hours after application of different concentrations of IL-1beta. At the end of 6 hours, cochleas were harvested and perilymphs were collected for assay of IL-1beta level by the ELISA method. RESULTS: At 6 hours after application, the control group animals did not show any significant hearing loss. The IL-1beta were undetectable in perilymph of control group. In IL-1beta group, IL-1beta was detected in perilymph in dose dependent increasing concentrations. There was a significant hearing loss in this IL-1beta group in dose dependent manner in 3 and 6 hours. The hearing loss in 100 ng was significant comparing to control group. CONCLUSION: The result of this study indicated that IL-1beta applied on RWM penetrate to the perilymph and caused hearing loss. The result of this study suggest that IL-1beta in MEE may cause sensorineural hearing loss in guinea pig.

Monitoring by ABR, TEOAE and DPOAE of Noise Induced Hearing Loss in Guinea Pig / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: Noise-induced hearing loss (NIHL) is associated with the damage in outer hair cell (OHC). Otoacoustic emmision (OAE) is sensitive to the detection of damage in OHC. We intended to evaluate the usefulness of transient evoked OAE (TEOAE) and distortion product OAE (DPOAE) on early detection and monitoring of NIHL in guinea pigs. MATERIALS AND METHODS: Ten healthy guinea pigs were exposed to 90+/-3 dBA white noise 8 hours daily for 9 days. Daily measurements of auditory brainstem response (ABR), TEOAE and DPOAE were performed before the noise exposure and daily after the noise exposure. RESULTS: ABR threshold became poorer gradually until the 8th day of post noise exposure, and some recovery on the 9th day. Wave reproducibility on TEOAE demonstrated an abrupt decrement on the 6th post-exposure day, and a tendency of recovery from the 7th day. While the echo amplitudes in f2=2002 Hz and f2=4004 Hz showed similar pattern of change with ABR threshold, the echo amplitude of f2=6348 Hz showed marked decrease on the 5th day compared to two other frequencies (f2=2002 Hz, 4004 Hz). CONCLUSION: The results of this study suggest that wave reproducibility in TEOAE and echo amplitude in high frequency in DPOAE may have applicability for early detection and monitoring of NIHL.