Development of a modified 3D region proposal network for lung nodule detection in computed tomography scans: a secondary analysis of lung nodule datasets.

BACKGROUND: Low-dose computed tomography (LDCT) has been shown useful in early lung cancer detection. This study aimed to develop a novel deep learning model for detecting pulmonary nodules on chest LDCT images. METHODS: In this secondary analysis, three lung nodule datasets, including Lung Nodule Analysis 2016 (LUNA16), Lung Nodule Received Operation (LNOP), and Lung Nodule in Health Examination (LNHE), were used to train and test deep learning models. The 3D region proposal network (RPN) was modified via a series of pruning experiments for better predictive performance. The performance of each modified deep leaning model was evaluated based on sensitivity and competition performance metric (CPM). Furthermore, the performance of the modified 3D RPN trained on three datasets was evaluated by 10-fold cross validation. Temporal validation was conducted to assess the reliability of the modified 3D RPN for detecting lung nodules. RESULTS: The results of pruning experiments indicated that the modified 3D RPN composed of the Cross Stage Partial Network (CSPNet) approach to Residual Network (ResNet) Xt (CSP-ResNeXt) module, feature pyramid network (FPN), nearest anchor method, and post-processing masking, had the optimal predictive performance with a CPM of 92.2%. The modified 3D RPN trained on the LUNA16 dataset had the highest CPM (90.1%), followed by the LNOP dataset (CPM: 74.1%) and the LNHE dataset (CPM: 70.2%). When the modified 3D RPN trained and tested on the same datasets, the sensitivities were 94.6%, 84.8%, and 79.7% for LUNA16, LNOP, and LNHE, respectively. The temporal validation analysis revealed that the modified 3D RPN tested on LNOP test set achieved a CPM of 71.6% and a sensitivity of 85.7%, and the modified 3D RPN tested on LNHE test set had a CPM of 71.7% and a sensitivity of 83.5%. CONCLUSION: A modified 3D RPN for detecting lung nodules on LDCT scans was designed and validated, which may serve as a computer-aided diagnosis system to facilitate lung nodule detection and lung cancer diagnosis.

A modified 3D RPN for detecting lung nodules on CT images that exhibited greater sensitivity and CPM than did several previously reported CAD detection models was established.

Combined model integrating deep learning, radiomics, and clinical data to classify lung nodules at chest CT.

OBJECTIVES: The study aimed to develop a combined model that integrates deep learning (DL), radiomics, and clinical data to classify lung nodules into benign or malignant categories, and to further classify lung nodules into different pathological subtypes and Lung Imaging Reporting and Data System (Lung-RADS) scores. MATERIALS AND METHODS: The proposed model was trained, validated, and tested using three datasets: one public dataset, the Lung Nodule Analysis 2016 (LUNA16) Grand challenge dataset (n = 1004), and two private datasets, the Lung Nodule Received Operation (LNOP) dataset (n = 1027) and the Lung Nodule in Health Examination (LNHE) dataset (n = 1525). The proposed model used a stacked ensemble model by employing a machine learning (ML) approach with an AutoGluon-Tabular classifier. The input variables were modified 3D convolutional neural network (CNN) features, radiomics features, and clinical features. Three classification tasks were performed: Task 1: Classification of lung nodules into benign or malignant in the LUNA16 dataset; Task 2: Classification of lung nodules into different pathological subtypes; and Task 3: Classification of Lung-RADS score. Classification performance was determined based on accuracy, recall, precision, and F1-score. Ten-fold cross-validation was applied to each task. RESULTS: The proposed model achieved high accuracy in classifying lung nodules into benign or malignant categories in LUNA 16 with an accuracy of 92.8%, as well as in classifying lung nodules into different pathological subtypes with an F1-score of 75.5% and Lung-RADS scores with an F1-score of 80.4%. CONCLUSION: Our proposed model provides an accurate classification of lung nodules based on the benign/malignant, different pathological subtypes, and Lung-RADS system.

Non-B, Non-C Hepatocellular Carcinoma in an HBV- and HCV-Endemic Area: A Community-Based Prospective Longitudinal Study.

A large community cohort of adults who participated in a health screening program from 2003 to 2013 were prospectively analyzed for the risk factors of non-B, non-C (NBNC) hepatocellular carcinoma (HCC). The serostatus of hepatitis B and C of 52,642 participants was linked to the mortality and cancer registration data of the Health and Welfare Data Science Center, Ministry of Health and Welfare, Taiwan. During a median follow-up of 6 years, 35 of the 43,545 participants who were negative for both HBsAg and anti-HCV antibody developed HCC. Multivariate Cox regression analysis revealed that old age (hazard ratio, 95% CI: 1.058, 1.019−1.098, p = 0.003); male sex (2.446, 1.200−4.985, p = 0.014); high aspartate aminotransferase levels (6.816, 2.945−15.779, p < 0.001); fibrosis index based on four factor score (1.262, 1.154−1.381, p < 0.001); blood sugar (1.009, 1.002−1.015, p = 0.006); and alpha-fetoprotein ≥15 ng/mL (143.938, 43.094−480.760, p < 0.001) were independent risk factors for HCC. By contrast, triglyceride >150 mg/dL was associated with a decreased risk of HCC (0.216, 0.074−0.625, p = 0.005). This prospective community-based study provided insights into the potential HCC risk factors which may shed some light in HCC prevention and screening.

Socioeconomic Inequality in One-Year Mortality of Elderly People with Hip Fracture in Taiwan.

Hip fracture commonly results in considerable consequences in terms of disability, mortality, long-term institutional care and cost. Taiwan launched its universal health insurance coverage in 1995, which largely removes financial barriers to health care. This study aims to investigate whether socioeconomic inequality in one-year mortality exists among Taiwanese elderly people. This population-based cohort study included 193,158 elderly patients (≥65 years) admitted for hip fracture between 2000 and 2012. With over a one-year follow-up, 10.52% of the participants died from all causes. The mortality rate was low in the northern part of Taiwan and in urban and high-family-income areas. Multiple Poisson regression models further suggested that the level of >Q1-Q3 and >Q3-Max showed significantly reduced odds ratio of one-year mortality at 0.90 (95% confidence interval (CI), 0.87-0.93) and 0.77 (95% CI, 0.74-0.81), respectively, compared with that of the lowest family income level (i.e., Min.-Q1). Despite a monotonic decline in overall one-year mortality during the study period, socioeconomic inequality in one-year mortality rate remained evident. The annual percentage change in one-year mortality was higher (-2.86) in elderly people from families with high income (>Q3-Max.) than that for elderly patients from family with low income (Min.-Q1, -1.94). Accessibility, rather than affordability, to health care for hip fracture is probably responsible for the observed socioeconomic inequality.

The role of video-assisted thoracoscopic surgery in therapeutic lung resection for pulmonary tuberculosis.

BACKGROUND: Video-assisted thoracoscopic surgery (VATS) has been considered an effective diagnostic modality for pulmonary tuberculosis. Its feasibility in therapeutic lung resection, however, has not been validated. METHODS: The medical records of patients who underwent VATS or a thoracotomy for therapeutic resection of pulmonary tuberculosis between January 2007 and March 2011 were reviewed for age, sex, indications for surgery, approach and procedures, preoperative sputum culture status, operative time, blood loss, hospital stay, and complications. RESULTS: One hundred twenty-three patients were enrolled. Sixty-three were successfully treated using VATS and 60 were converted to thoracotomy. The number of VATS wedge resections was significantly higher (p = 0.004). Patients who underwent VATS had significantly less blood loss, shorter hospital stays, and fewer complications (p = 0.031, 0.000, and 0.022, respectively). Lesions treated with a pneumonectomy or that required thoracoplasty failed to be done using VATS (p = 0.054 and 0.002, respectively). Patients who underwent VATS had slightly more isolated lobectomies and significantly (p = 0.005) shorter hospital stays than did thoracotomy patients. Concomitant and isolated segmentectomies were done using VATS, but there were significantly fewer than for thoracotomy patients (p = 0.033). CONCLUSIONS: Video-assisted thoracoscopic surgery is effective for therapeutic wedge resections, isolated lobectomies, and simple segmentectomies and lobectomies combined with wedge resections or segmentectomies for pulmonary tuberculosis. Tuberculosis lesions that require a pneumonectomy or thoracoplasty are still major challenges for VATS.

Thymic neuroendocrine carcinoma and thymoma are both associated with increased risk of extrathymic malignancy: a 20-year review of a single institution.

BACKGROUND: Thymoma has been reported to have increased risk of extrathymic malignancy; thymic carcinoma, however, has not been validated of this association. We retrospectively assessed the incidence of additional malignancy among patients with thymoma and thymic carcinoma, and compared it with that of other solid organ cancers. METHODS: We reviewed the medical records between the years of 1988 and 2008 of 213 patients, including 131 with thymoma and 82 with thymic carcinoma. The overall incidence of additional malignancy in patients with thymic epithelial tumors, lung cancer, hepatocellular carcinoma, colorectal cancer, breast cancer, and cervical cancer between 2003 and 2008 in our institution was also computed. RESULTS: The incidence of extrathymic malignancy in patients with thymic epithelial tumors was 12.2% (16 of 131) for thymoma and 12.2% (10 of 82) for thymic carcinoma. The overall incidence of additional malignancy between 2003 and 2008 was significantly higher among patients with thymoma than among patients with hepatocellular carcinoma, colorectal cancer, breast cancer, and cervical cancer (p=0.017, 0.022, 0.009, and 0.018, respectively). In thymic carcinoma, the likelihood of developing extrathymic malignancy was significantly higher among patients with neuroendocrine carcinoma than among patients with other thymic carcinomas (p=0.000). Extrathymic neoplasm did not pose a significant influence on the overall survival of patients with thymoma (p=0.085) and thymic carcinoma (p=0.814). CONCLUSIONS: Our data demonstrated the increased risk of extrathymic malignancies among patient with thymoma. In thymic carcinoma, this association mainly occurred in patients with neuroendocrine carcinoma. Actions for early detection of extrathymic malignancy should be considered for patients with these thymic epithelial tumors.

Angiogenetic biomarkers in non-small cell lung cancer with malignant pleural effusion: correlations with patient survival and pleural effusion control.

OBJECTIVE: To evaluate how the angiogenetic biomarkers vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and interleukin-8 (IL-8) in the fluid of non-small cell lung cancer (NSCLC) with malignant pleural effusion (MPE) correlate with patient survival and pleural effusion control. DESIGN: Prospective study. PATIENTS AND METHODS: From April 1, 1998 to April 30, 2005, we used thoracoscopic biopsy to collect pleural specimens and pleural effusion from 97 patients with NSCLC and MPE. Paired blood samples were harvested. We used enzyme-linked immunosorbent assays to evaluate levels of angiogenic factors in MPE and blood, and immunohistochemical staining to evaluate them in pleural specimens. Related data, such as patient survival and PE control, were collected for correlation analysis. RESULTS: Smoking and PE VEGF >1350 ng/mL were both significant negative predictors of patient survival. A trapped lung was the only significant factor for poor PE control. The serum level, the amount of PE, and the number of red blood cells in PE correlated well with PE VEGF level. Immunohistochemical staining of pleural samples showed that VEGF was secreted by both mesothelial and tumor cells. The level of PE IL-8 weakly correlated with PE VEGF, and the level of bFGF was not significant. CONCLUSIONS: PE VEGF was a useful angiogenetic factor for the amount of fluid in patients with NSCLC and MPE. In addition to smoking, PE VEGF >1350 ng/mL was a significant negative predictor of patient survival.

Complete laryngotracheal disruption caused by blunt injury.

BACKGROUND: We reviewed the clinical courses and evaluated the surgical results of 7 patients with complete laryngotracheal disruption caused by blunt injury. METHODS: Seven patients with complete laryngotracheal disruption caused by blunt injury were successfully treated in a 13-year period. Six of the seven incidents involved men younger than 30 years on motorcycles. All but one had intact cutaneous tissue of the neck. Six of seven laryngotracheal disruptions were at the cricotracheal junction and the other was at the junction of second and third tracheal ring. In the emergency departments, 4 of these 7 patients underwent endotracheal intubation and three others underwent tracheostomy after failed intubation. Two of 7 patients underwent delayed surgery (posttrauma day 3 and day 5) because of delayed diagnosis. All patients underwent laryngotracheoplasty with (n = 3) or without (n = 4) concomitant tracheostomy. RESULTS: Total hospital stays ranged from 9 to 28 days (average 15 days). Intensive care unit stay ranged from 2 to 10 days (average 5.8 days). All 7 patients had paralysis of bilateral vocal cords that were revealed by postoperative bronchoscopy. In 3 patients who underwent concomitant tracheostomy, the tracheostomy tubes were removed within 3 to 5 months after surgery. In the other 4 patients who underwent laryngotracheoplasty only, the endotracheal tube was used as an airway support for 2 to 6 days (average 3.5 days). All patients had patent airways. Vocal cord function partially recovered in one side (n = 6) or both sides (n = 1). Their voices were audible but still husky 5 months or 1 year later. CONCLUSIONS: Complete laryngotracheal disruption can be treated by laryngotracheoplasty with or without concomitant tracheostomy, and phonation can be partially recovered.