The biomass, composition and size-structure of phytoplankton communities directly control many key ocean processes, such as ocean carbon cycling, food web length, trophic transfer efficiencies and biogeochemical processes. Phytoplankton communities are structured through complex interactions of bottom-up (physical processes; nutrient, light, trace metal availability) and top-down (biological processes; grazing, viral lyses, mortality) forcing that are constantly working in concert. In this dissertation I address the biomass structure and environmental relationships for phytoplankton communities in the southern California Current and adjacent ocean ecosystems. I begin by examining the microbial communities in the eastern Equatorial Pacific (EEP), a high-nitrate, low-chlorophyll open-ocean upwelling region, where I developed and first applied an advanced digital epifluorescence microscopy system. By combing results obtained from epifluorescence microscopy with flow cytometry (FCM) I am able to obtain a fairly complete picture of the composition, biomass and size-structure of the entire microbial community throughout the euphotic zone. My finding from the EEP show that upwelling modestly alters the phytoplankton community size-structure, giving areas with enhanced diatom and dinoflagellate biomass and also increases in nano- and pico-phytoplankton. Then I applied these same methods to investigate changes in the microbial community across a strong frontal zone in the California Current Ecosystem (CCE) where the most striking features were a sharp transition between the phototrophic bacteria Prochlorococcus (PRO) and Synechococcus (SYN) and a very large (̃7 fold) increase in micro-sized (>20 [mu]m) biomass directly at the front. Lastly, I compiled data on microbial communities from four disparate regions of the central to eastern North Pacific Ocean in order to investigate interregional commonalities and differences, and to test hypothesized relationships between phytoplankton size structure and total phytoplankton biomass along trophic gradients. Here I show that there are coherent patterns and variability in the phytoplankton community size-structure and composition that are clear across a wide range of trophic states and ecosystem types

Background

Pneumothorax can precipitate a life-threatening emergency due to lung collapse and respiratory or circulatory distress. Pneumothorax is typically detected on chest X-ray; however, treatment is reliant on timely review of radiographs. Since current imaging volumes may result in long worklists of radiographs awaiting review, an automated method of prioritizing X-rays with pneumothorax may reduce time to treatment. Our objective was to create a large human-annotated dataset of chest X-rays containing pneumothorax and to train deep convolutional networks to screen for potentially emergent moderate or large pneumothorax at the time of image acquisition.

Methods and findings

In all, 13,292 frontal chest X-rays (3,107 with pneumothorax) were visually annotated by radiologists. This dataset was used to train and evaluate multiple network architectures. Images showing large- or moderate-sized pneumothorax were considered positive, and those with trace or no pneumothorax were considered negative. Images showing small pneumothorax were excluded from training. Using an internal validation set (n = 1,993), we selected the 2 top-performing models; these models were then evaluated on a held-out internal test set based on area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and positive predictive value (PPV). The final internal test was performed initially on a subset with small pneumothorax excluded (as in training; n = 1,701), then on the full test set (n = 1,990), with small pneumothorax included as positive. External evaluation was performed using the National Institutes of Health (NIH) ChestX-ray14 set, a public dataset labeled for chest pathology based on text reports. All images labeled with pneumothorax were considered positive, because the NIH set does not classify pneumothorax by size. In internal testing, our "high sensitivity model" produced a sensitivity of 0.84 (95% CI 0.78-0.90), specificity of 0.90 (95% CI 0.89-0.92), and AUC of 0.94 for the test subset with small pneumothorax excluded. Our "high specificity model" showed sensitivity of 0.80 (95% CI 0.72-0.86), specificity of 0.97 (95% CI 0.96-0.98), and AUC of 0.96 for this set. PPVs were 0.45 (95% CI 0.39-0.51) and 0.71 (95% CI 0.63-0.77), respectively. Internal testing on the full set showed expected decreased performance (sensitivity 0.55, specificity 0.90, and AUC 0.82 for high sensitivity model and sensitivity 0.45, specificity 0.97, and AUC 0.86 for high specificity model). External testing using the NIH dataset showed some further performance decline (sensitivity 0.28-0.49, specificity 0.85-0.97, and AUC 0.75 for both). Due to labeling differences between internal and external datasets, these findings represent a preliminary step towards external validation.

Conclusions

We trained automated classifiers to detect moderate and large pneumothorax in frontal chest X-rays at high levels of performance on held-out test data. These models may provide a high specificity screening solution to detect moderate or large pneumothorax on images collected when human review might be delayed, such as overnight. They are not intended for unsupervised diagnosis of all pneumothoraces, as many small pneumothoraces (and some larger ones) are not detected by the algorithm. Implementation studies are warranted to develop appropriate, effective clinician alerts for the potentially critical finding of pneumothorax, and to assess their impact on reducing time to treatment.

The study aimed to determine if computer vision techniques rooted in deep learning can use a small set of radiographs to perform clinically relevant image classification with high fidelity. One thousand eight hundred eighty-five chest radiographs on 909 patients obtained between January 2013 and July 2015 at our institution were retrieved and anonymized. The source images were manually annotated as frontal or lateral and randomly divided into training, validation, and test sets. Training and validation sets were augmented to over 150,000 images using standard image manipulations. We then pre-trained a series of deep convolutional networks based on the open-source GoogLeNet with various transformations of the open-source ImageNet (non-radiology) images. These trained networks were then fine-tuned using the original and augmented radiology images. The model with highest validation accuracy was applied to our institutional test set and a publicly available set. Accuracy was assessed by using the Youden Index to set a binary cutoff for frontal or lateral classification. This retrospective study was IRB approved prior to initiation. A network pre-trained on 1.2 million greyscale ImageNet images and fine-tuned on augmented radiographs was chosen. The binary classification method correctly classified 100 % (95 % CI 99.73-100 %) of both our test set and the publicly available images. Classification was rapid, at 38 images per second. A deep convolutional neural network created using non-radiological images, and an augmented set of radiographs is effective in highly accurate classification of chest radiograph view type and is a feasible, rapid method for high-throughput annotation.

Background

Emphysematous pyelonephritis is a severe infection of the kidney associated with formation of gas in the renal parenchyma and/or collecting system. The purpose of this study was to evaluate outcomes of patients with emphysematous pyelonephritis in a contemporary cohort and to evaluate the impact of urolithiasis on disease severity.

Methods

A search of all imaging reports at University of California San Francisco (UCSF) for the term "emphysematous pyelonephritis" was undertaken from 2003-2014. Patients were included if there was clinical evidence of infection, no recent urologic instrumentation, and computerized tomography (CT) demonstrating gas in the renal parenchyma or collecting system. Clinical and laboratory variables were obtained from medical records.

Results

A total of 14 cases were identified. The majority of patients (57%) had gas confined to the collecting system. Three patients (21%) had gas in the renal parenchyma and 3 patients (21%) had gas extending into perirenal tissues. A total of 8 patients (57%) had concomitant urolithiasis. Seven patients (50%) were managed with antibiotic therapy alone while 6 patients (43%) required percutaneous drainage. No patients required immediate nephrectomy. There were no deaths. Patients with urolithiasis had less severe emphysematous pyelonephritis than patients without urolithiasis (P<0.05).

Conclusions

The majority of patients in this study had gas contained within the collecting system and were treated successfully with antibiotics alone. Percutaneous drainage was successfully utilized in patients with more advanced disease. No patients required emergent nephrectomy. Emphysematous pyelonephritis in patients with urolithiasis was less severe than in patients without urolithiasis.

Venous thromboembolism (VTE) is an extremely common form of vascular disease and impacts a great number of patients worldwide. Acute deep vein thrombosis (DVT) is a subset of VTE and is traditionally been treated with anticoagulation. There is good quality data which suggests the use of catheter directed interventions for the treatment of acute DVT with the aim of reducing post-thrombotic syndrome (PTS). The present review will discuss the various therapies available for acute DVT, focusing on catheter directed interventions, ranging from traditional anticoagulation to the most novel forms of aspiration thrombectomy.

Purpose

To create an algorithm able to accurately detect IVC filters on radiographs without human assistance, capable of being used to screen radiographs to identify patients needing IVC filter retrieval.

Methods

A primary dataset of 5225 images, 30% of which included IVC filters, was assembled and annotated. 85% of the data was used to train a Cascade R-CNN (Region Based Convolutional Neural Network) object detection network incorporating a pre-trained ResNet-50 backbone. The remaining 15% of the data, independently annotated by three radiologists, was used as a test set to assess performance. The algorithm was also assessed on an independently constructed 1424-image dataset, drawn from a different institution than the primary dataset.

Results

On the primary test set, the algorithm achieved a sensitivity of 96.2% (95% CI 92.7-98.1%) and a specificity of 98.9% (95% CI 97.4-99.5%). Results were similar on the external test set: sensitivity 97.9% (95% CI 96.2-98.9%), specificity 99.6 (95% CI 98.9-99.9%).

Conclusion

Fully automated detection of IVC filters on radiographs with high sensitivity and excellent specificity required for an automated screening system can be achieved using object detection neural networks. Further work will develop a system for identifying patients for IVC filter retrieval based on this algorithm.

We previously reported that activated γδ T cells greatly enhance autoimmune responses, particularly the Th17 response. To determine the mechanisms involved, we made a series of comparisons between activated and non-activated γδ T cells. Our results showed that activated γδ T cells expressed greatly increased levels of A2A adenosine receptor (A2AR) and decreased amounts of CD73, as well as increased amounts of T cell activation markers such as CD69, CD44 and CD25. We show that A2AR is a major functional molecule in the enhancing activity of γδ T cells. A2AR-/- γδ T cells (isolated from A2AR-/- mouse), lost their Th17-enhancing activity as did A2AR+/+ γδ T cells (isolated from wt-B6 mouse) after treatment with an A2AR antagonist. Since γδ T cells possess either an enhancing or an inhibiting effect, we also tested whether A2AR expression on γδ T cells is essential to their inhibiting effect. Our results showed that the inhibiting effect of A2AR-/- γδ T cells was as potent as that of A2AR+/+ γδ T cells. In a previous report we showed that the expression of different levels of CD73 molecule allowed γδ T cells to adjust their suppressive activity; in the current study, we show that expression of increased amounts of A2AR allows γδ T cells to more effectively exert their enhancing function.

PURPOSE: To compare treatment response after transarterial chemoembolization (TACE) for hepatocellular carcinoma (HCC) in patients with and without a transjugular intrahepatic portosystemic shunt (TIPS). MATERIALS AND METHODS: A retrospective review of patients who underwent conventional TACE for HCC between January 2005 and December 2009 identified 10 patients with patent TIPS. From the same time period, 23 patients without TIPS were selected to control for comparable Model for End-Stage Liver Disease and Child-Pugh-Turcotte scores. The two groups showed similar distribution of Barcelona Clinic Liver Cancer and United Network of Organ Sharing stages. Target HCC lesions were evaluated according to the modified response evaluation criteria in solid tumors (mRECIST) guidelines. Transplantation rate, time to tumor progression, and overall survival (OS) were documented. RESULTS: After TACE, the rate of complete response was significantly greater in non-TIPS patients compared with TIPS patients (74 vs. 30 %, p = 0.03). Objective response rate (complete and partial response) trended greater in the non-TIPS group (83 vs. 50 %, p = 0.09). The liver transplantation rate was 80 and 74 % in the TIPS and non-TIPS groups, respectively (p = 1.0). Time to tumor progression was similar (p = 0.47) between the two groups. OS favored the non-TIPS group (p = 0.01) when censored for liver transplantation. CONCLUSION: TACE is less effective in achieving complete or partial response using mRECIST criteria in TIPS patients compared with those without a TIPS. Nevertheless, similar clinical outcomes may be achieved, particularly in TIPS patients who are liver-transplantation candidates.