Benchmark Data Set and Method for Depth Estimation from Light Field Images.

Convolutional Neural Networks (CNN) have performed extremely well for many image analysis tasks. However, supervised training of deep CNN architectures requires huge amounts of labelled data which is unavailable for light field images. In this paper, we leverage on synthetic light field images and propose a two stream CNN network that learns to estimate the disparities of multiple correlated neighbourhood pixels from their Epipolar Plane Images (EPI). Since the EPIs are unrelated except at their intersection, a two stream network is proposed to learn convolution weights individually for the EPIs and then combine the outputs of the two streams for disparity estimation. The CNN estimated disparity map is then refined using the central RGB light field image as a prior in a variational technique. We also propose a new real world dataset comprising light field images of 19 objects captured with the Lytro Illum camera in outdoor scenes and their corresponding 3D pointclouds, as ground truth, captured with the 3dMD scanner. This dataset will be made public to allow more precise 3D pointcloud level comparison of algorithms in the future which is currently not possible. Experiments on the synthetic and real world datasets show that our algorithm outperforms existing state-of-the-art for depth estimation from light field images.

Fibrinolysis in critically ill patients.

Impaired fibrinolysis may contribute to development of adult respiratory distress syndrome (ARDS). Pathologic increases in endogenous plasminogen activator inhibitor (PAI-1) may blunt normal fibrinolysis and unmask alternate fibrinolytic mechanisms, such as elastase-induced fibrin degradation. We measured PAI-1 and elastase-induced fibrin(ogen) degradation products in 69 critically ill patients in our medical intensive care unit (MICU) and in nine healthy volunteers. Factor VIII-related antigen protein (VIII:Ag), a reported marker of acute lung injury, and alpha-1-protease inhibitor (alpha-1-PI), an acute phase reactant, were also measured. MICU patients included 24 control patients with no known risk of ARDS, 35 patients with risk factors for ARDS including sepsis, pneumonia, aspiration, and shock, and 12 patients with ARDS including two patients from at-risk groups who developed ARDS. Plasma PAI-1 was determined by chromogenic assay, elastase-induced peptides by a new radioimmunoassay, VIII:Ag by immunoelectrophoresis, and alpha-1-PI by immunodiffusion. When compared to normal volunteers, MICU control patients had elevated PAI-1, VIII:Ag, elastase-induced peptides, and alpha-1-PI. Patients with ARDS had significantly higher PAI-1 and VIII:Ag than did MICU control patients; elastase-induced peptides and alpha-1-PI were not higher. However, at-risk patients who did not develop ARDS also had high PAI-1 or VIII:Ag. Although these data cannot refute the possible role of these compounds in the pathogenesis of ARDS, they demonstrate that PAI-1 and VIII:Ag may be elevated in many critically ill patients but may not be useful markers for the subsequent development of ARDS.

Central venous pressure and pulmonary capillary wedge pressure as estimates of left atrial pressure: effects of positive end-expiratory pressure and catheter tip malposition.

We compared CVP and pulmonary capillary wedge pressure (WP) measurements with left atrial pressure (LAP) in postcoronary bypass surgical patients with preserved cardiopulmonary function. Measurements were obtained under normal conditions and conditions likely to induce WP-LAP discrepancies (PEEP and catheter tip malposition). Patients were in both supine and lateral positions; the catheter tip was placed vertically below (tip down; n = 12) or above (tip up; n = 5) the left atrium. Our data showed that both CVP and WP correlated well with LAP at all PEEP levels in the supine and tip down lateral positions. However, in the tip up lateral position, WP overestimated LAP (13.3 +/- 3.4 vs. 8.0 +/- 2 mm Hg; p less than .01) at 20 cm H2O of PEEP, whereas CVP (8.8 +/- 2.1 mm Hg) closely reflected LAP. Thus, by placing the catheter tip vertically below the left atrium in supinely and laterally positioned patients, CVP and WP both produced reliable estimates of LAP despite an acute increase in alveolar pressure. When the pulmonary artery catheter tip was vertically above the left atrium, WP overestimated LAP. Under these conditions, CVP remained a reliable estimate of LAP. We conclude that CVP measurement as an estimate of LAP in this patient population could be used and not ignored. This is true in patients with previously documented good LAP-CVP correlation who are subjected subsequently to conditions which may produce an LAP-WP discrepancy (high PEEP and catheter tip malposition).

Topographic location of the left atrium by computed tomography: reducing pulmonary artery catheter calibration error.

Two potential errors of pulmonary artery wedge pressure measurement that have received little attention are improper anatomic referencing of the transducer and nondependent placement of the pulmonary artery catheter tip. Transducers are often referenced to the midaxillary line of the supine patient. We utilized CT of the chest to determine accurately the topographic location of the left atrium to evaluate the accuracy of referencing the transducer in this position and to investigate atrial location in the lateral decubitus positions vis-à-vis external landmarks, so that more accurate referencing could be performed with a pulmonary artery catheter tip in the dependent lung. A prospective group consisting of ten patients referred for chest CT as well as retrospective review of 40 chest CT scans served as the study population. This study demonstrated a discrepancy between the perceived midaxillary line and the true location of the midleft atrium. Lateral decubitus CT scanning revealed a more readily localizable surface anatomic landmark which consistently and accurately predicts midleft atrial location. In addition, the appropriate decubitus position guarantees dependent catheter tip placement (Zone 3).

Effects of antiplatelet drugs on pulmonary responses to thrombin in anesthetized rabbits.

Platelet aggregation and fibrin deposition in the pulmonary circulation may contribute to the pathogenesis of lung injury in the adult respiratory distress syndrome (ARDS). We evaluated the effect of two antiplatelet drugs (forskolin & dipyridamole) on pulmonary responses to intravenous infusion of 100 NIH units of thrombin per kg bw in anesthetized, and ventilated rabbits treated with fibrinolysis inhibitor. Thrombin infusion resulted in pulmonary hypertension and increased arterial CO2 tension (PaCO2) and dead space ventilation (VD/VT). Arterial oxygen tension (PaO2) and numbers of circulating leukocytes and platelets dropped after thrombin infusion. These early hemodynamic changes correlated with histological evidence of entrapped leukocytes in the pulmonary microcirculation and transient alveolar edema. Microthrombi were rarely observed in animals that received thrombin. There was little evidence for endothelial damage or progressive lung water accumulation. Treatment with forskolin or dipyridamole reversed thrombin-induced changes in pulmonary artery pressure, PaCO2, VD/VT and systemic oxygenation. Moreover, forskolin and dipyridamole blunted the drop in circulating leukocytes and prevented the development of alveolar edema following thrombin. The beneficial actions of these agents may be due to interference with the release of mediators from leukocytes or platelets.

Nontraumatic fat embolization. A rare cause of new pulmonary infiltrates in an immunocompromised patient.

Diffuse pulmonary infiltrates and hypoxemia are common in immunocompromised patients. We describe a patient with lymphoma who developed hypoxemia and diffuse pulmonary infiltrates during treatment with corticosteroids. Open lung biopsy and postmortem examination indicated that the cause of the infiltrates was nontraumatic fat embolization (NTFE). Most previous cases of NTFE have implicated a fatty liver as the source of emboli; however, this patient had no fatty changes of the liver. The diagnosis of NTFE in an immunocompromised patient is difficult to make because its distinguishing features, such as hypoxemia, petechiae, and altered mental status, are nonspecific in this setting.

Influence of lung injury on pulmonary wedge-left atrial pressure correlation during positive end-expiratory pressure ventilation.

The correlation between pulmonary artery wedge pressure (Pw) and left atrial pressure (Pla) requires a continuous fluid column between the catheter tip and the left atrium. We hypothesized that lung injury may protect the fluid column from the collapsing effects of increased airway pressure. Correlation between Pw and Pla would then depend on catheter tip location in injured versus normal lung regions. In 7 anesthetized dogs with unilateral acid pneumonitis, we compared Pla and simultaneous Pw measurements from pulmonary artery catheters located in injured and normal lungs at different levels of positive end-expiratory pressure (PEEP). Studies were repeated in 10 dogs with normal lungs and 5 dogs with bilateral acid pneumonitis. In supine dogs with unilateral lung injury, Pw from the injured lung more accurately reflected Pla than did Pw obtained from the normal lung at PEEP levels of 7 mmHg or higher, in contrast to data from dogs with normal lungs or equally injured lungs. Discrepancies between Pw and Pla at PEEP levels of 7 and 11 mmHg from the normal lung were corrected when that lung was placed in the dependent position to increase venous pressure at the catheter tip. A good Pw-Pla correlation was not guaranteed by catheter tip location below the level of the left atrium during PEEP ventilation. We conclude that the continuity of the fluid column was protected by lung injury. Although Pw-Pla differences from the normal lung were modest at the levels of PEEP that are usually optimal for gas exchange in uneven lung injury, it is recommended that the injured lung should not be avoided during insertion of the balloon-tipped catheter.

Lateral position improves wedge-left atrial pressure correlation during positive-pressure ventilation.

Because lateral position can be used to locate a pulmonary artery catheter tip in lung regions where venous pressure exceeds alveolar pressure, we studied the effect of lateral position on the correlation between pulmonary artery occlusion pressure (Pw) and left atrial pressure (Pla) at various increments of positive-end expiratory pressure (PEEP). In ten normal anesthetized pigs, catheters were placed in the left atrium and right and left pulmonary arteries; simultaneous measurements of Pla and Pw from both catheters were obtained in the supine, right lateral, and left lateral positions. Pw obtained in the lateral position when the catheter tip was vertically located below the left atrium ("lower") more accurately reflected Pla changes than Pw obtained from catheters above the left atrium, at PEEP levels less than 20 cm H2O. Although most catheter tips were located below the left atrium in the supine position, our data from the lower catheter more accurately assessed Pla than did the Pw obtained in the supine position; this was particularly evident when the catheter tip was located vertically above the left atrium. These findings suggest that the lateral position can be useful in measuring Pw during continuous positive-pressure ventilation. However, further studies in humans are needed before this maneuver can be recommended.

Effect of catheter position on thermodilution cardiac output during continuous positive-pressure ventilation.

PEEP may decrease regional perfusion to nondependent lung regions, thereby creating different zones of thermal dissipation in dependent and nondependent zones of the lung. Under these conditions, the measurement of cardiac output by thermodilution may, thus, be influenced by the vertical position of the pulmonary artery catheter in the lung. We investigated this hypothesis in 7 healthy, anesthetized pigs by comparing cardiac output measurements from thermistors located in dependent and nondependent lung regions at varying levels of PEEP. Our data from thermistors in these 2 positions were similar, suggesting that the measurement of cardiac output by thermodilution is not influenced by the vertical position of the thermal sensor with respect to the left atrium.

Lung function and structure after Escherichia coli endotoxin in rabbits: effect of dose and rate of administration.

We evaluated the effect of increasing doses of Escherichia coli endotoxin and its rate of administration on systemic blood pressure, alveolar-arterial oxygen gradient (A- aDO2 1.0), dynamic compliance (Cdyn), circulating platelets and leukocytes, and postmortem bloodless wet to dry ratios in anesthetized rabbits. Infusion of endotoxin resulted in systemic hypotension, diminished Cdyn, thrombocytopenia, and leukopenia, but did not influence venous admixture. These parameters were not affected by the rapidity of administration, but changes in Cdyn and circulating platelets were dose-dependent. High (15 mg/kg), but not low (0.5 mg/kg), doses of endotoxin resulted in an early but transient increase in lung water, but bloodless wet to dry weight ratios were not increased at 4-6 h following endotoxin even when high doses were injected. Ultrastructural studies done in six rabbits showed an early but transient platelet sequestration in pulmonary capillaries, progressive increase in intracapillary leukocytes, interstitial edema, and focal, although minimal, endothelial injury at 4 h after injection. Thus, infusion of E coli endotoxin in rabbits does not result in increased lung water and intrapulmonary shunting acutely; this tolerance to endotoxin is not related to the dose or rates of administration studied.

Effect of metoprolol on cardiac and pulmonary function in chronic obstructive pulmonary disease.

We evaluated the effects of oral metoprolol, 200 mg daily, on cardiac and pulmonary function in 6 patients with chronic reversible airways obstruction and no cardiac dysfunction. The patients were clinically stable. In all patients, baseline forced expiratory volume in 1 second (FEV1) was less than 60% predicted, and increased at least 15% after isoproterenol inhalation. Resting control first pass right and left ventricular radionuclide ejection fractions were normal or only slightly depressed. Compared to placebo, metoprolol did not significantly affect FEV1 or forced vital capacity. Metoprolol did not significantly alter left or right ventricular ejection fraction, measured by first pass radionuclide technique. On patient experienced increased dyspnea on metoprolol, which was not accompanied by changes in clinical exam, spirometry, or ejection fraction. We conclude that metoprolol may be safely administered to a subset of stable patients with chronic reactive airways disease.

Resolution of pulmonary dysfunction following acute chlorine exposure.

Following acute exposure to chlorine gas, 18 asymptomatic subjects were evaluated for early pulmonary dysfunction. Airway obstruction was evident in all exposed individuals immediately after chlorine exposure. The severity, course of the obstructive defect, and clinical picture correlated with the chief complaint on admission. The obstructive abnormalities resolved within 1 wk after exposure to chlorine in 12 subjects whose chief complaint was cough. A slower resolution of the physiologic changes, clinical signs, and symptoms was noted in 6 subjects whose initial chief complaint was dyspnea. In this group, maximum mid-expiratory flow rate (FEF25-75%), and forced expiratory flow after exhaling 50% and 75% of the vital capacity (FEF50% and FEF25%, respectively) were still diminished 2 wk after chlorine exposure. The slow rate of resolution in the dyspnea group is best explained by increased individual susceptibility since a past medical history of smoking or asthma and "wheezing" was more prevalent in this group.

Effect of positive end-expiratory pressure and body position in unilateral lung injury.

Positive end-expiratory pressure (PEEP), by increasing lung volume in acute lung injury, may recruit terminal air spaces in the involved regions, but may also distend noninvolved regions increasing extravascular lung water and worsening gas exchange. We investigated the effect of increasing levels of PEEP on arterial oxygenation in 26 anesthesized dogs with unilateral acid pneumonitis and studied the influences of gravity and distribution of the injury on this effect. Arterial PO2 was consistently higher when the noninjured lung was dependent than in the supine or injured lung-dependent positions. Low levels of PEEP (5, 10 cmH2O) improved arterial oxygenation and reduced intrapulmonary physiological shunt. However, 15 cmH2O PEEP resulted in worsening of gas exchange, increased dead space ventilation, and diminished static compliance. The adverse effects of high levels of PEEP on arterial oxygenation were similar whether the injured lung was dependent or not and were evident a lower levels of PEEP in one group of dogs in which the unilateral injury was more diffuse and in which the upper and middle lobes were also involved. Thus, the compressive effects of high levels of PEEP on alveolar capillaries in the noninjured lung are influenced by the extent and distribution of injury in the injured lung, but not by local forces governing regional blood flow distribution.

Effect of intravenous metoprolol on reversible obstructive airways disease.

We gave increasing doses of metoprolol intravenously to seven subjects with stable chronic obstructive pulmonary disease (COPD) who were also receiving their usual bronchodilators. Six of the seven tolerated up to 0.2mg/kg metoprolol without adverse effects, although there were declines in forced expiratory volume in 1 sec (FEV1). At 0.15 mg/kg mean FEV1 fell 12% (p less than 0.025), and at 0.2 mg/kg mean decline in FEV1 was 15% (p less than 0.01). These findings suggest that 0.2 mg/kg metoprolol may be given intravenously to most patients with COPD in addition to previously administered bronchodilators without precipitating clinically significant adverse effects. Any side effects that develop can be reversed by beta agonists.

Epidemiologic observations on eosinophilia and its relation to respiratory disorders.

The percentage of eosinophils (%EOS), determined from a differential blood smear, was measured in 2,311 subjects enrolled in a general population study in Tucson, Arizona. A subgroup of 290 subjects was tested in more detail during a later evaluation in which absolute eosinophil counts, leukocyte counts, and nasal smears for eosinophils were obtained. In men, but not in women, there was a significant tendency for the %EOS to decrease with age. The highest %EOS was noted during the months of February through May, the time when most plants in this region are in bloom. Blood eosinophils were significantly related to allergy skin test reactivity, circulating IgE concentrations, several respiratory symptoms and disease diagnoses, as well as to reduced ventilatory function. Among subjects younger than 55 yr of age, however, ventilatory function was significantly low, and symptom rates increased only when there was allergy skin test reactivity in addition to eosinophilia. Neither allergy skin test reactivity nor eosinophilia alone was related to ventilatory function in this age group. Among older subjects, blood eosinophilia was associated with definite impairment of ventilatory function, regardless of skin test reactivity and independent of smoking habits. The presence of eosinophilia identified a predominantly female group of elderly nonsmokers with markedly impaired ventilatory function. These subjects appeared to fall into the clinical category of "asthmatic bronchitis".

The usefulness of fiberoptic bronchoscopy in evaluating new pulmonary lesions in the compromised host.

Thirty-four fiberoptic bronchoscopies employing various bronchoscopic technics were carried out in 33 immune-compromised patients for the evaluation of new pulmonary lesions. Transbronchial biopsy was performed only with fluoroscopic guidance and was omitted in patients with a bleeding tendency. Bronchial brushing and bronchial washing were successfully carried out despite the presence of contraindications to biopsy. Brushing and washing were diagnostically useful in 66 and 74 per cent of the cases, respectively, compared to 71 per cent for forceps biopsy. The combined over-all yield was 88 per cent, with no serious complications encountered. The most common etiology of new infiltrates was opportunistic infection. Among bacterial infections, gram-negative organisms were the most common, and among fungal etiologies, Coccidioides immitis was the predominant pathogen in this series from Tucson, Arizona. Although the roentgenographic pattern was not helpful in predicting the etiology of the new infiltrates, diffuse lesions were more frequently evaluated correctly by fiberoptic bronchoscopy than localized lesions. The low incidence of complications and the high over-all yield indicate that fiberoptic bronchoscopy, employing bronchial brushing and washing as supplements to transbronchial biopsy (and as a replacement to biopsy in patients with a bleeding diathesis), can be very useful in evaluating new pulmonary lesions in the immune-compromised patient. When used together, these technics significantly increase the diagnostic yield and eliminate the risks associated with performing more invasive diagnostic procedures in the compromised host.

Asbestos exposure and related neoplasia. The 28 year experience of a major urban hospital.

In a retrospective study of 49 cases of asbestosis, a steady increase in the frequency of diagnosis of asbestosis and asbestos-related neoplasia is documented from a major urban hospital since 1960. Although in the majority of cases the subjects were exposed to asbestos in a neighboring shipyard, in 20 per cent of the cases, asbestos exposure was in industries not related to shipbuilding, reflecting its widespread use. This selective population of patients with asbestosis more often than not had an associated neoplasm. The most likely accompanying tumor was pleural mesothelioma, and among cell types of lung cancer, adenocarcinoma was notably frequent.

The association of adenocarcinoma of the lung and blastomycosis from an unusual geographical location.

A case of North American blastomycosis acquired outside the American continent and presenting with a left hilar mass is reported. The progressive hilar enlargement, despite adequate antifungal treatment, raised the possibility of drug resistance, concomitant tuberculosis and carcinoma. Non-invasive diagnostic studies including cultures, cytology and fibreoptic bronchoscopy were negative. On thoracotomy, both blastomy cosis and a poorly undifferentiated adenocarcinoma were evident.

Reversibility of severe bleomycin-induced pneumonitis.

The use of bleomycin sulfate as an antineoplastic agent has been limited by its substantial pulmonary toxic effects. The exact incidence and prognosis of bleomycin-pneumonitis is unresolved. Although bleomycin pulmonary toxicity is thought to be dose-related, recent reports have emphasized severe reactions at low doses. Furthermore, severe pulmonary toxicity has been suggested to be progressive, irreversible, and ultimately, fatal. We report clinical, roentgenographic, and pathophysiologic recovery after severe, bleomycin-induced pneumonitis.