Automatic detection of pulmonary embolism on computed tomography pulmonary angiogram scan using a three-dimensional convolutional neural network.

OBJECTIVE: To propose a convolutional neural network (EmbNet) for automatic pulmonary embolism detection on computed tomography pulmonary angiogram (CTPA) scans and to assess its diagnostic performance. METHODS: 305 consecutive CTPA scans between January 2019 and December 2021 were enrolled in this study (142 for training, 163 for internal validation), and 250 CTPA scans from a public dataset were used for external validation. The framework comprised a preprocessing step to segment the pulmonary vessels and the EmbNet to detect emboli. Emboli were divided into three location-based subgroups for detailed evaluation: central arteries, lobar branches, and peripheral regions. Ground truth was established by three radiologists. RESULTS: The EmbNet's per-scan level sensitivity, specificity, positive predictive value (PPV), and negative predictive value were 90.9%, 75.4%, 48.4%, and 97.0% (internal validation) and 88.0%, 70.5%, 42.7%, and 95.9% (external validation). At the per-embolus level, the overall sensitivity and PPV of the EmbNet were 86.0% and 61.3% (internal validation), and 83.5% and 57.5% (external validation). The sensitivity and PPV of central emboli were 89.7% and 52.0% (internal validation), and 94.4% and 43.0% (external validation); of lobar emboli were 95.2% and 76.9% (internal validation), and 93.5% and 72.5% (external validation); and of peripheral emboli were 82.6% and 61.7% (internal validation), and 80.2% and 59.4% (external validation). The average false positive rate was 0.45 false emboli per scan (internal validation) and 0.69 false emboli per scan (external validation). CONCLUSION: The EmbNet provides high sensitivity across embolus locations, suggesting its potential utility for initial screening in clinical practice.

Lung adenocarcinoma associated with cystic airspaces: Predictive value of CT features in assessing pathologic invasiveness.

OBJECTIVES: Lung adenocarcinoma associated with cystic airspaces (LACA) is a unique entity with limited understanding. Our aim was to evaluate the radiological characteristics of LACA and to study which criteria were predictive of invasiveness. METHODS: A retrospective monocentric analysis of consecutive patients with pathologically confirmed LACA was performed. The diagnosed adenocarcinomas were classified into preinvasive (atypical adenomatous hyperplasia, adenocarcinoma in situ, or minimally invasive adenocarcinoma) and invasive adenocarcinomas. Eight clinical features and twelve CT features were evaluated. Univariable and multivariable analyses were performed to analyse the correlation between invasiveness, and CT and clinical features. The inter-observer agreement was evaluated using κ statistics and intraclass correlation coefficients. The predictive performance of the model was evaluated using the area under the receiver operating characteristic curve (AUC). RESULTS: A total of 252 patients with 265 lesions (128 men and 124 women; mean age, 58.0 ± 11.1 years) were enrolled. Multivariable logistic regression indicated that multiple cystic airspaces (OR, 5.599; 95 % CI, 1.865-16.802), irregular shape of cystic airspace (OR, 3.236; 95 % CI, 1.073-9.761), entire tumour size (OR, 1.281; 95 % CI, 1.075-1.526), and attenuation (OR, 1.007; 95 % CI, 1.005-1.010) were independent risk factors for invasive LACA. The AUC of the logistic regression model was 0.964 (95 % CI, 0.944-0.985). CONCLUSION: Multiple cystic airspaces, irregular shape of cystic airspace, entire tumour size, and attenuation were identified as independent risk factors for invasive LACA. The prediction model gives a good predictive performance, providing additional diagnostic information.

China lung cancer screening (CLUS) version 2.0 with new techniques implemented: Artificial intelligence, circulating molecular biomarkers and autofluorescence bronchoscopy.

OBJECTIVE: The present study, CLUS version 2.0, was conducted to evaluate the performance of new techniques in improving the implementation of lung cancer screening and to validate the efficacy of LDCT in reducing lung cancer-specific mortality in a high-risk Chinese population. METHODS: From July 2018 to February 2019, high-risk participants from six screening centers in Shanghai were enrolled in our study. Artificial intelligence, circulating molecular biomarkers and autofluorescencebronchoscopy were applied during screening. RESULTS: A total of 5087 eligible high-risk participants were enrolled in the study; 4490 individuals were invited, and 4395 participants (97.9%) finally underwent LDCT detection. Positive screening results were observed in 857 (19.5%) participants. Solid nodules represented 53.6% of all positive results, while multiple nodules were the most common location type (26.8%). Up to December 2020, 77 participants received lung resection or biopsy, including 70 lung cancers, 2 mediastinal tumors, 1 tracheobronchial tumor, 1 malignant pleural mesothelioma and 3 benign nodules. Lung cancer patients accounted for 1.6% of all the screened participants, and 91.4% were in the early stage (stage 0-1). CONCLUSIONS: LDCT screening can detect a high proportion of early-stage lung cancer patients in a Chinese high-risk population. The utilization of new techniques would be conducive to improving the implementation of LDCT screening.

Prediction of future imagery of lung nodule as growth modeling with follow-up computed tomography scans using deep learning: a retrospective cohort study.

Background: Risk prediction models of lung nodules have been built to alleviate the heavy interpretative burden on clinicians. However, the malignancy scores output by those models can be difficult to interpret in a clinically meaningful manner. In contrast, the modeling of lung nodule growth may be more readily useful. This study developed a CT-based visual forecasting system that can visualize and quantify a nodule in three dimensions (3D) in any future time point using follow-up CT scans. Methods: We retrospectively included 246 patients with 313 lung nodules with at least 1 follow-up CT scan. For the manually segmented nodules, we calculated geometric properties including CT value, diameter, volume, and mass, as well as growth properties including volume doubling time (VDT), and consolidation-to-tumor ratio (CTR) at follow-ups. These nodules were divided into growth and non-growth groups by thresholding their VDTs. We then developed a convolutional neural network (CNN) to model the imagery change of the nodules from baseline CT image (combined with the nodule mask) to follow-up CT image with a particular time interval. The model was evaluated on the geometric and radiological properties using either logistic regression or receiver operating characteristic (ROC) curve. Results: The lung nodules consisted of 115 ground glass nodules (GGN) and 198 solid nodules and were followed up for an average of 354 days with 2 to 11 scans. The 2 groups differed significantly in most properties. The prediction of our forecasting system was highly correlated with the ground truth with small relative errors regarding the four geometric properties. The prediction-derived VDTs had an area under the curve (AUC) of 0.857 and 0.843 in differentiating growth and non-growth nodules for GGN and solid nodules, respectively. The prediction-derived CTRs had an AUC of 0.892 in classifying high- and low-risk nodules. Conclusions: This proof-of-concept study demonstrated that the deep learning-based model can accurately forecast the imagery of a nodule in a given future for both GGNs and solid nodules and is worthy of further investigation. With a larger dataset and more validation, such a system has the potential to become a prognostication tool for assessing lung nodules.

Co-amplification of genes in chromosome 8q24: a robust prognostic marker in hepatocellular carcinoma.

BACKGROUND: Hepatocellular carcinoma (HCC) is a leading cause of tumor-associated death worldwide, owing to its high 5-year postoperative recurrence rate and inter-individual heterogeneity. Thus, a prognostic model is urgently needed for patients with HCC. Several researches have reported that copy number amplification of the 8q24 chromosomal region is associated with low survival in many cancers. In the present work, we set out to construct a multi-gene model for prognostic prediction in HCC. METHODS: RNA sequencing and copy number variant data of tumor tissue samples of HCC from The Cancer Genome Atlas (n=328) were used to identify differentially expressed messenger RNAs of genes located on the chromosomal 8q24 region by the Wilcox test. Univariate Cox and Lasso-Cox regression analyses were carried out for the screening and construction of a prognostic multi-gene signature in The Cancer Genome Atlas cohort (n=119). The multi-gene signature was validated in a cohort from the International Cancer Genome Consortium (n=240). A nomogram for prognostic prediction was built, and the underpinning molecular mechanisms were studied by Gene Set Enrichment Analysis. RESULTS: We successfully established a 7-gene prognostic signature model to predict the prognosis of patients with HCC. Using the model, we divided individuals into high-risk and low-risk sets, which showed a significant difference in overall survival in the training dataset (HR =0.17, 95% CI: 0.1-0.28; P<0.001) and in the testing dataset (HR = 0.42, 95% CI: 0.23-0.74; P=0.002). Multivariate Cox regression analysis showed the signature to be an independent prognostic factor of HCC survival. A nomogram including the prognostic signature was constructed and showed a better predictive performance in short-term (1 and 3 years) than in long-term (5 years) survival. Furthermore, Gene Set Enrichment Analysis identified several pathways of significance, which may aid in explaining the underlying molecular mechanism. CONCLUSIONS: Our 7-gene signature is a reliable prognostic marker for HCC, which may provide meaningful information for therapeutic customization and treatment-related decision making.

Prediction of pathologic stage in non-small cell lung cancer using machine learning algorithm based on CT image feature analysis.

PURPOSE: To explore imaging biomarkers that can be used for diagnosis and prediction of pathologic stage in non-small cell lung cancer (NSCLC) using multiple machine learning algorithms based on CT image feature analysis. METHODS: Patients with stage IA to IV NSCLC were included, and the whole dataset was divided into training and testing sets and an external validation set. To tackle imbalanced datasets in NSCLC, we generated a new dataset and achieved equilibrium of class distribution by using SMOTE algorithm. The datasets were randomly split up into a training/testing set. We calculated the importance value of CT image features by means of mean decrease gini impurity generated by random forest algorithm and selected optimal features according to feature importance (mean decrease gini impurity > 0.005). The performance of prediction model in training and testing sets were evaluated from the perspectives of classification accuracy, average precision (AP) score and precision-recall curve. The predictive accuracy of the model was externally validated using lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) samples from TCGA database. RESULTS: The prediction model that incorporated nine image features exhibited a high classification accuracy, precision and recall scores in the training and testing sets. In the external validation, the predictive accuracy of the model in LUAD outperformed that in LUSC. CONCLUSIONS: The pathologic stage of patients with NSCLC can be accurately predicted based on CT image features, especially for LUAD. Our findings extend the application of machine learning algorithms in CT image feature prediction for pathologic staging and identify potential imaging biomarkers that can be used for diagnosis of pathologic stage in NSCLC patients.

Development and validation of a predictive model for the diagnosis of solid solitary pulmonary nodules using data mining methods.

BACKGROUND: The purpose of this study is to develop a predictive model to accurately predict the malignancy of solid solitary pulmonary nodule (SPN) by data mining methods. METHODS: A training cohort of 388 consecutive patients with solid SPNs was used to develop a predictive model to evaluate the malignancy of solid SPNs. By using SPSS Modeler, we utilized logistic regression (LR), artificial neural network (ANN), k-nearest neighbor (KNN), random forest (RF), and support vector machines (SVM) classifiers to build predictive models. Another cohort of 200 consecutive patients with solid SPNs was used to verify the accuracy of the predictive model. Predictive performance was evaluated using the area under the receiver operating characteristic curve (AUC). RESULTS: There was no significant difference in patients' characteristics between the training cohort and the validation cohort. The AUCs of LR, ANN, KNN, RF, and SVM models for the validation cohort were 0.874±0.0280 (P=0.605), 0.833±0.0351 (P=0.104), 0.792±0.0418 (P=0.014), 0.775±0.0400 (P=0.013), and 0.890±0.0323 (reference), respectively. The SVM algorithm had the highest AUC, and the best sensitivity (90.3%), specificity (80.4%), positive predictive value (93.9%), negative predictive value (71.2%) and accuracy (88.0%) for the validation cohort among the five models. CONCLUSIONS: Data mining by SVM might be a useful auxiliary algorithm in predicting malignancy of solid SPNs.

Screening for early stage lung cancer and its correlation with lung nodule detection.

Currently, the most effective way of reducing lung cancer mortality is early diagnosis of lung cancer. The National Lung Screening Trial has proved the efficacy of lung cancer screening using low-dose computed tomography to reduce lung cancer mortality. However, many questions remain surrounding lung cancer screening implementation, among which include how to select the optimal risk population, the personalized screening interval based different levels of risk, methods to improve diagnostic discrimination between malignant and benign disease in detected lung nodules, and the roles of biomolecular markers in stratifying risk and in guiding the management of indeterminate nodules. This review concentrates on the latest developments of lung cancer screening and provides an overview of the main unanswered questions on lung nodule detection.

Predicting malignancy of pulmonary ground-glass nodules and their invasiveness by random forest.

BACKGROUND: The purpose of this study was to develop a predictive model that could accurately predict the malignancy of the pulmonary ground-glass nodules (GGNs) and the invasiveness of the malignant GGNs. METHODS: The authors built two binary classification models that could predict the malignancy of the pulmonary GGNs and the invasiveness of the malignant GGNs. RESULTS: Results of our developed model showed random forest could achieve 95.1% accuracy to predict the malignancy of GGNs and 83.0% accuracy to predict the invasiveness of the malignant GGNs. CONCLUSIONS: The malignancy and invasiveness of pulmonary GGNs could be predicted by random forest.

Community-based lung cancer screening with low-dose CT in China: Results of the baseline screening.

OBJECTIVES: To investigate whether low-dose computed tomography (LDCT) screening is capable of enhancing the detection rate of early-stage lung cancer in high-risk population of China with both smoking and non-smoking related factors. METHODS: From 2013-2014, eligible participants with high-risk factors of lung cancer were randomly assigned to a screening group or a control group with questionnaire inquiries. Any non-calcified nodules or masses with longest diameters of ≥4 mm identified on LDCT images were considered as positive. RESULTS: A total of 6717 eligible participants were randomly enrolled to a study group (3550 to LDCT screening and 3167 to standard care). 3512 participants (98.9%) underwent LDCT screening, and 3145 participants (99.3%) received questionnaire inquiries. A positive screening result was observed in 804 participants (22.9%). In the two-year follow-up period, lung cancer was detected in 51 participants (1.5%) in the LDCT group versus 10 (0.3%) in the control group (stage I: 48 vs 2; stage II to IV or limited stage: 3 vs 8), respectively. Early-stage lung cancer was found in 94.1% vs 20%, respectively. CONCLUSIONS: Compared to usual care, LDCT led to a 74.1% increase in detecting early-stage lung cancer. This study provides insights about the non-smoking related risk factors of lung cancer in the Chinese population.

High-resolution Computed Tomography Features Distinguishing Benign and Malignant Lesions Manifesting as Persistent Solitary Subsolid Nodules.

INTRODUCTION: We retrospectively investigated the high-resolution computed tomography features that distinguish benign lesions (BLs) from malignant lesions (MLs) appearing as persistent solitary subsolid nodules (SSNs). MATERIALS AND METHODS: In 2015, the data from patients treated in our department with persistent solitary SSNs 5 to 30 mm in size were analyzed retrospectively. The demographic data and HRCT findings were analyzed and compared between those with BLs and MLs. RESULTS: Of the 1934 SSNs, 94 were BLs and 1840 were MLs. One half of the MLs (920 SSNs) were randomly selected and analyzed. The BLs were classified into 2 subgroups: 28 pure ground-glass nodules (pGGNs) and 66 part-solid nodules (PSNs). After matching in each group, 56 pGGNs and 132 PSNs in the ML group were selected. In the pGGN subgroup, multivariate analysis found that a well-defined border (odds ratio [OR], 4.320; 95% confidence interval [CI], 1.534-12.168; P = .006; area under the curve, 0.705; 95% CI, 0.583-0.828; P = .002) and a higher average CT value (OR, 1.007; 95% CI, 1.001-1.013; P = .026; area under the curve, 0.715; 95% CI, 0.599-0.831; P = .001) favored the diagnosis of malignancy. In the PSN subgroup, multivariate analysis revealed that a larger size (OR, 1.084; 95% CI, 1.015-1.158; P = .016), a well-defined border (OR, 3.447; 95% CI, 1.675-7.094; P = .001), and a spiculated margin (OR, 2.735; 95% CI, 1.359-5.504; P = .005) favored the diagnosis of malignancy. CONCLUSION: In pGGNs, a well-defined lesion border and a larger average CT value can be valuable discriminators to distinguish between MLs and BLs. In PSNs, a larger size, well-defined border, and spiculated margin had greater predictive value for malignancy.

Characterization and Performance of Hollow Silica for Adsorption of Amoxicillin in Aqueous Solutions.

Two kinds of hollow silica materials, namely H-SiS1 and H-SiS2, were synthesized using the yeast template method and the Pickering emulsion polymerization method, respectively. The adsorbents were synthesized to adsorb amoxicillin (AMX) from an aqueous environment. Characterization results indicated that hollow silica adsorbents exhibited excellent thermal stability even at temperatures above 700 °C. Several batches of static adsorption experiments were prepared to analyze the adsorption performance on AMX. Isotherm data on different adsorbents fitted well with the Langmuir model (from 15 °C to 35 °C), indicating a monolayer molecular adsorption mechanism for AMX. The maximum adsorption capacities of H-SiS1 and H-SiS2 were 8.40 and 3.46 mg/g at 35 °C, respectively. The adsorption kinetics was described well by the pseudo-second-order model, which indicated that chemical interactions were primarily responsible for AMX adsorption and could be the rate-limiting step during adsorption. These results suggested that H-SiS1 could be significantly useful as adsorbents for removal of AMX residuals from aqueous solution.

[Establishment of a predicting model to evaluate the probability of malignancy or benign in patients with solid solitary pulmonary nodules].

OBJECTIVE: To develop a predicting model for evaluating the probability of malignancy or benign in patients with solid solitary pulmonary nodules through analyzing the clinical, radiologic, laboratory examination and radionuclide (18)F-Fluorodeoxyglucose examinations data. METHODS: The data of the 203 patients(110 males and 93 females) with solid SPN who underwent surgical resection with definite postoperative pathological diagnosis from January 2012 to December 2014 in Shanghai Chest Hospital (group A)were retrospectively analyzed. The clinical data included age, gender, history of smoking, history of tumor; radiologic data included diameter in lung window, location, shape, clear border, lobulation, spiculation, vascular convergence, tumor cycle blood vessel, density, calcification, pleura indentation; laboratory examination included five serum tumor markers consisting of CA125, CEA, CYFRAL21-1, NSE, SCC. (18)F-Fluorodeoxyglucose examinations included (18)F-FDG PET-CT or SPECT. The independent predictors of malignancy were estimated through univariate and multivariate analysis, then the predicting model was built. Another 110 patients with solid SPN(group B)from January 2015 to December 2015 with definite pathological diagnosis were used to validate the predictive value of the model. RESULTS: There were 159(78.3%) cases of malignancy and 44(21.7%) cases of benign in group A. Logistic regression analysis showed age, clear border, spiculation, calcification and (18)F-FDG examination were independent predictors of malignancy in patients with solid SPN(P<0.05). A predicting nomogram was built according to the result of the multivariate logistic regression analysis. The area under the ROC curve was 0.890±0.038 for group B. The cut off value was 0.708. The sensitivity in group B was 86%, specificity 80%, accuracy 84.5%. CONCLUSION: Age of patients, clear border, spiculation, calcification and (18)F-FDG examination were independent predictors of malignancy in patients with solid SPN. The model showed good diagnosis efficiency in external validation, and could be applied to make decision for patients with solid SPN.

Selective removal of erythromycin by magnetic imprinted polymers synthesized from chitosan-stabilized Pickering emulsion.

Magnetic imprinted polymers (MIPs) were synthesized by Pickering emulsion polymerization and used to adsorb erythromycin (ERY) from aqueous solution. The oil-in-water Pickering emulsion was stabilized by chitosan nanoparticles with hydrophobic Fe3O4 nanoparticles as magnetic carrier. The imprinting system was fabricated by radical polymerization with functional and crosslinked monomer in the oil phase. Batches of static and dynamic adsorption experiments were conducted to analyze the adsorption performance on ERY. Isotherm data of MIPs well fitted the Freundlich model (from 15 °C to 35 °C), which indicated heterogeneous adsorption for ERY. The ERY adsorption capacity of MIPs was about 52.32 µmol/g at 15 °C. The adsorption kinetics was well described by the pseudo-first-order model, which suggested that physical interactions were primarily responsible for ERY adsorption. The Thomas model used in the fixed-bed adsorption design provided a better fit to the experimental data. Meanwhile, ERY exhibited higher affinity during adsorption on the MIPs compared with the adsorption capacity of azithromycin and chloramphenicol. The MIPs also exhibited excellent regeneration capacity with only about 5.04% adsorption efficiency loss in at least three repeated adsorption-desorption cycles.

The IASLC/ATS/ERS classification of lung adenocarcinoma-a surgical point of view.

Adenocarcinoma has become the most common histologic type of lung cancers. Ground glass nodules (GGN), most of them early stage noninvasive or minimally invasive adenocarcinomas (MIA), have been encountered more frequently with the application of computed tomography (CT) screening. The International Association for the Study of Lung Cancer (IASLC)/American Thoracic Society (ATS)/European Respiratory Society (ERS) histologic lung adenocarcinoma classification combines radiologic, histologic, clinic, and molecular features to form a diagnostic approach for different subgroups of diseases. One of the major focuses of this new classification is the introduction of adenocarcinoma in situ (AIS) and MIA, to replace the old term of bronchioloalveolar carcinoma (BAC). Not all GGNs are malignant lesions that should be surgically resected upon first presentation. A management approach different to solid nodules has been suggested based on the understanding that these lesions tend to have a more indolent nature. Hasty intervention should be avoided and potential surgical risks, radiation exposure, patient psychology, and socio-economical burden must be balanced comprehensively before surgery is decided upon. In the mean time, surgical issues concerning extent of resection and lymphadenectomy should also be carefully contemplated once intervention is deemed necessary. Extremely good prognosis with a near 100% disease-free survival could be expected when a pure GGN is completely resected. This has led to re-evaluation of sublobar resections, including both segmentectomy and big wedge resection, for small (≤2 cm) less invasive histology (AIS or MIA) appearing as GGN on CT scan. Evidences are accumulating that these limited resections are oncologically equivalent to standard lobectomy. And extensive lymph node dissection may not have additional staging or prognostic benefit. These would add new meaning to the contemporary definition of minimally invasive surgery for lung cancers. Overall, joint effort from a multiple disciplinary team is imperative, and decision making should be based on both anatomical and biological nature of the disease.

[Values in preoperative assessment of occlusion and intra-operative guidance of 64 row spiral CT in percutaneous coronary intervention for chronic coronary occlusion].

OBJECTIVE: To instigate the values of 64 row spiral CT in pre-operative assessment of the occlusion and intra-operative guidance in percutaneous coronary intervention for chronic total occlusion (CTO) in coronary heart disease. METHODS: Fifteen coronary disease patients planned to receive percutaneous coronary intervention underwent 64-row spiral CT-coronary angiography and coronary angiography (CAG). The diagnostic effects of these 2 techniques were compared. RESULTS: Seventeen CTO lesions were confirmed. MSCT succeeded to show the lengths of the 17 CTO lesions with a calcification identification rate of 76.4%, significantly higher than that of the CAG (41.5%). By cross-section examination, MSCT succeeded to detect the occlusion degree of the calcified lesions, and showed that 3 CTO lesions were occluded at a rate < 50%, and 10 lesions at a rate > or = 50%. Twelve complete occlusion lesions in 11 patients underwent PCT, success was seen in 6 of which and failure in the other 6. Univariate analysis showed that the length of lesion, branching at the proximal site, formation of bridging lateral branch, form of occlusion end, and calcification were all not significantly related to the success or failure of intervention. The percentage of the calcification area > or = 50% in the intervention failure group was 83.3%, significantly higher than that in the intervention success group (16.7%, P = 0.05). 3-D images of coronary artery could be obtained by MSCT to show all the complete occlusive lesions. CONCLUSION: 64-MSCT demonstrates a remarkable ability to identify silicified lesions, can re-establish 3-D images of coronary artery, and effectively guide the intervention therapy.