Volumetric segmentation of ground glass nodule based on 3D attentional cascaded residual U-Net and conditional random field.

BACKGROUND: Ground glass nodule (GGN) segmentation is one of the important and challenging tasks in diagnosing early-stage lung adenocarcinomas. Manually delineating of 3D GGN in a computed tomography (CT) image is a subjective, laborious, and tedious task, which presents poor repeatability. PURPOSE: To reduce the annotation burden and improve the segmentation performance, this study proposes a 3D deep learning-based volumetric segmentation model to segment the GGN in CT images. METHODS: A total of 379 GGNs were retrospectively collected from the public database, Shanghai Pulmonary Hospital (SHPH), and Fudan University Shanghai Cancer Center (FUSCC). First, a series of image preprocessing techniques involving image resampling, intensity normalization, 3D nodule patch cropping, and data augmentation, were adopted to generate the input images for the deep learning model by using CT scans. Then, a 3D attentional cascaded residual network (ACRU-Net) was proposed to develop the deep learning-based segmentation model by using the residual network and the atrous spatial pyramid pooling module. To improve the model performance, a voxel-based conditional random field (CRF) method was used to optimize the segmentation results. Finally, a balanced cross-entropy and Dice combined loss function was applied to train and build the segmentation model. RESULTS: Testing on SHPH and FUSCC datasets, the proposed method generates the Dice coefficients of 0.721 ± 0.167 and 0.733 ± 0.100, respectively, which are higher than those of 3D residual U-Net and ACRU-Net without CRF optimization. CONCLUSIONS: The results demonstrated that combining 3D ACRU-Net and CRF effectively improved the GGN segmentation performance. The proposed segmentation model may provide a potential tool to help the radiologist in the segmentation and diagnosis of 3D GGN.

Health Care Analytics With Time-Invariant and Time-Variant Feature Importance to Predict Hospital-Acquired Acute Kidney Injury: Observational Longitudinal Study.

BACKGROUND: Acute kidney injury (AKI) develops in 4% of hospitalized patients and is a marker of clinical deterioration and nephrotoxicity. AKI onset is highly variable in hospitals, which makes it difficult to time biomarker assessment in all patients for preemptive care. OBJECTIVE: The study sought to apply machine learning techniques to electronic health records and predict hospital-acquired AKI by a 48-hour lead time, with the aim to create an AKI surveillance algorithm that is deployable in real time. METHODS: The data were sourced from 20,732 case admissions in 16,288 patients over 1 year in our institution. We enhanced the bidirectional recurrent neural network model with a novel time-invariant and time-variant aggregated module to capture important clinical features temporal to AKI in every patient. Time-series features included laboratory parameters that preceded a 48-hour prediction window before AKI onset; the latter's corresponding reference was the final in-hospital serum creatinine performed in case admissions without AKI episodes. RESULTS: The cohort was of mean age 53 (SD 25) years, of whom 29%, 12%, 12%, and 53% had diabetes, ischemic heart disease, cancers, and baseline eGFR <90 mL/min/1.73 m2, respectively. There were 911 AKI episodes in 869 patients. We derived and validated an algorithm in the testing dataset with an AUROC of 0.81 (0.78-0.85) for predicting AKI. At a 15% prediction threshold, our model generated 699 AKI alerts with 2 false positives for every true AKI and predicted 26% of AKIs. A lowered 5% prediction threshold improved the recall to 60% but generated 3746 AKI alerts with 6 false positives for every true AKI. Representative interpretation results produced by our model alluded to the top-ranked features that predicted AKI that could be categorized in association with sepsis, acute coronary syndrome, nephrotoxicity, or multiorgan injury, specific to every case at risk. CONCLUSIONS: We generated an accurate algorithm from electronic health records through machine learning that predicted AKI by a lead time of at least 48 hours. The prediction threshold could be adjusted during deployment to optimize recall and minimize alert fatigue, while its precision could potentially be augmented by targeted AKI biomarker assessment in the high-risk cohort identified.

Placental Stem Villus Arterial Remodeling Associated with Reduced Hydrogen Sulfide Synthesis Contributes to Human Fetal Growth Restriction.

Intrauterine fetal growth restriction (IUGR) is often associated with compromised umbilical arterial flow, indicating increased placental vascular resistance. Oxidative stress is causatively implicated. Hydrogen sulfide maintains differentiated smooth muscle in vascular beds, and its synthetic enzyme cystathionine-γ-lyase (CSE) is down-regulated in growth-restricted placentas. We hypothesized that remodeling of resistance arteries in stem villi contributes to IUGR by compromising umbilical blood flow via oxidative stress, reducing hydrogen sulfide signaling. Stem villus arteries in human IUGR placentas displaying absent or reversed end-diastolic flow contained reduced myosin heavy chain, smooth muscle actin, and desmin, and increased markers of dedifferentiation, cellular retinol-binding protein 1, and matrix metalloproteinase 2, compared to term and preterm controls. Wall thickness/lumen ratio was increased, lumen diameter decreased, but wall thickness remained unchanged in IUGR placentas. CSE correlated positively with myosin heavy chain, smooth muscle actin, and desmin. Birth weight correlated positively with CSE, myosin heavy chain, smooth muscle actin, and desmin, and negatively with cellular retinol-binding protein 1 and matrix metalloproteinase 2. These findings could be recapitulated in vitro by subjecting stem villus artery explants to hypoxia-reoxygenation, or inhibiting CSE. Treatment with a hydrogen sulfide donor, diallyl trisulfide, prevented these changes. IUGR is associated with vascular remodeling of the stem villus arteries. Oxidative stress results in reduction of placental CSE activity, decreased hydrogen sulfide production, and smooth muscle cell dedifferentiation in vitro. This vascular remodeling is reversible, and hydrogen sulfide donors are likely to improve pregnancy outcomes.

Cross-sectional and longitudinal analysis of cancer vaccination trials registered on the US Clinical Trials Database demonstrates paucity of immunological trial endpoints and decline in registration since 2008.

INTRODUCTION: Cancer vaccination has been researched as a means of treating and preventing cancer, but successful translational efforts yielding clinical therapeutics have been limited. Numerous reasons have been offered in explanation, pertaining both to the vaccine formulation, and the clinical trial methodology used. This study aims to characterize the tumor vaccine clinical trial landscape quantitatively, and explore the possible validity of the offered explanations including the translational obstacles posed by the current common endpoints. METHODS: We performed a detailed cross-sectional and longitudinal analysis of tumor vaccine trials (n=955) registered in the US Clinical Trials database. RESULTS: The number of tumor vaccine trials initiated per annum has declined 30% since a peak in 2008. In terms of vaccine formulation, 25% of trials use tumor cell/lysate preparations; whereas, 73% of trials vaccinate subjects against defined protein/peptide antigens. Also, 68% of trials do not use vectors for antigen delivery. Both these characteristics of tumor vaccines have remained unchanged since 1996. The top five types of cancer studied are: melanoma (22.6%); cervical cancer (13.0%); breast cancer (11.3%); lung cancer (9.5%); and prostate cancer (9.4%). In addition, 86% of the trials are performed where there is established disease rather than prophylactically, of which 67% are performed exclusively in the adjuvant setting. Also, 42% of Phase II trials do not measure any survival-related endpoint, and only 23% of Phase III trials assess the immune response to vaccination. CONCLUSION: The clinical trial effort in tumor vaccination is declining, necessitating a greater urgency in identifying and removing the obstacles to clinical translation. These obstacles may include: 1) vaccination against a small range of antigens; 2) naked delivery of antigen; 3) investigation of less immunogenic cancer types; and 4) investigation in the setting of established disease. In addition, the prevalence of late phase failure may be due to inadequate assessment of survival-related endpoints in Phase II trials. The clinical trial development of tumor vaccines should include mechanism-based translational endpoints, as well as the discovery of immune biomarkers with which to stratify, monitor, and prognosticate patients.