Neural networks for the analysis of small pulmonary nodules.

PURPOSE: Small pulmonary nodules can be readily detected by computed tomography (CT). The goal of this detection is to diagnose early lung cancer as the five year survival at this early stage is over 70% in contradistinction to the overall 5-year survival of around 10%. Critical to the efficacy of CT for early lung cancer detection is the ability to distinguish between benign and malignant nodules. We explored the usefulness of neural networks (NNs) to help in this differentiation. METHODS: CT images of 28 pulmonary nodules, 14 benign and 14 malignant, each having a diameter less than 3 cm were selected. All were sufficiently malignant in appearance to require needle biopsy and surgery. The statistical-multiple object detection and location system (S-MODALS) NN technique developed for automatic target recognition (ATR) was used to differentiate between these benign and malignant nodules. RESULTS: S-MODALS was able to correctly identify all but three benign nodules. S-MODALS classified a nodule as malignant because it looked similar to other malignant nodules. It identified the most similar nodules to display them to the radiologist. The specific features of the nodule that determined its classification were also shown, so that S-MODALS is not simply a "black box" technique but gives insight into the NN diagnostics. CONCLUSION: This initial evaluation of S-MODALS NNs using pulmonary nodules whose CT features were very suspicious for lung cancer demonstrated the potential to reduce the number of biopsies without missing malignant nodules. S-MODALS performed well, but additional optimization of the techniques specifically for CT images would further enhance its performance.

Diagnostic dosages of protirelin (TRH) elevate BP by noncatecholamine mechanisms.

While performing thyroid function tests, we noticed that protirelin (TRH) raised BP, and, therefore, we investigated the effect of diagnostic dosages of protirelin (500 micrograms) on plasma catecholamine levels and cardiovascular function in eight patients one day before, one day after, and four weeks following heart surgery. Mean arterial pressure (MAP), heart rate (HR), plasma norepinephrine (NE), epinephrine (EPI), dopamine (DA), thyroid hormone (triiodothyronine [T3], thyroxine), and thyrotropin (TSH) levels were measured before and after the intravenous injection of protirelin. Protirelin increased MAP transiently from 88 +/- 2 to 103 +/- 3 mm Hg (before surgery), 86 +/- 4 to 102 +/- 4 mm Hg (one day after surgery), and 86 +/- 4 to 104 +/- 5 mm Hg (four weeks after surgery). There were no notable changes in HR or plasma NE, EPI, or DA levels. The T3 and TSH response to protirelin was normal on all three study days. Protirelin raised MAP by an effect on systemic vascular resistance (SVR) rather than an increase in cardiac output. We conclude the following: (1) diagnostic dosages of protirelin transiently elevate MAP and SVR by a noncatecholamine mechanism, (2) clinicians who perform protirelin tests should be aware of protirelin's transient pressor effects.

Measurement of urinary leukocyte esterase activity: a screening test for urinary tract infections.

We evaluated the efficacy of testing for the presence of esterase (an enzyme released only from leukocytes) in the urine as an indicator of the presence of pyuria. We hypothesized that a "dipstick" test for urinary leukocyte esterase activity would be a rapid and simple screening technique for detecting urinary tract infections (UTI). To test our hypothesis we collected fresh urine specimens from 203 patients (148 outpatients, 55 inpatients) with a suspected UTI. Each specimen was divided into three aliquots; one was used for reagent strip testing (for leukocyte esterase, nitrite, and blood), one for microscopy, and one for culture. Of the 203 specimens, 49 showed significant bacteriuria (greater than or equal to 10(5) organisms/mL). The leukocyte esterase test was 100% sensitive (0% false negatives) with a 76% specificity (24% false positives) in predicting significant bacteriuria. Although a positive nitrite reaction was more specific (99% specificity, 0.6% false positives), it was insensitive (27% sensitivity, 73% false negatives). The high sensitivity of reagent strip leukocyte esterase testing for pyuria makes it a valuable screening test that should lead to the elimination of many needless urine cultures and microscopic examinations.

Increased circulating plasma norepinephrine concentrations in noncardiac causes of pulmonary hypertension.

Sympathetic nervous system (SNS) activity and circulating norepinephrine (NE) levels may play roles in the elevated pulmonary vascular resistance (PVR) found in patients with pulmonary hypertension. To study the relationship between plasma NE levels, pulmonary NE metabolism, and pulmonary hypertension, we studied 9 patients, suspected of having noncardiac causes of pulmonary hypertension, before and after vasodilator therapy with phentolamine, nitroglycerin, isoproterenol, or hydralazine. Patients were admitted to the ICU and studied using pulmonary artery thermodilution catheters. Seven of 9 patients had noncardiogenic pulmonary hypertension whereas 2 patients did not have hemodynamic evidence of pulmonary hypertension. Simultaneous pulmonary and radial artery samples were analyzed for plasma NE and epinephrine (EPI) content before and after various vasodilator agents. Baseline pulmonary artery (PA) and radial artery NE concentrations correlated (r = .72) with PVR, PA pressures, and PA minus pulmonary capillary wedge (WP) pressures and were increased compared to the 2 patients with normal pulmonary pressures (p less than .01). The normal pulmonary extraction of circulating NE was absent in the 7 patients with pulmonary hypertension. PVR decreased significantly in all 7 patients with each vasodilator (p less than .05); however, the decrease was independent of any change in plasma NE concentration and therapy had no effect on the pulmonary extraction of circulating NE. These data indicate that elevations in plasma NE are coincident with the presence of noncardiogenic pulmonary hypertension and that acute pharmacologic reduction of PVR does not normalize the loss of pulmonary NE metabolism.

Sympathetic nervous system "switch off" with severe hypothermia.

Hypothermia occurs frequently in the critically ill patient, yet little is known about the endogenous catecholamine response to this stress. To study this problem, we measured heart rate (HR), mean arterial blood pressure (MAP), and plasma levels of norepinephrine (NE) and epinephrine (Epi) in subhuman primates (baboons) during progressive hypothermia from 37 degrees to 29 degrees C and then during rewarming to 37 degrees C. As the core temperature decreased from 37 degrees to 33 degrees C, HR and MAP increased significantly (p less than 0.05), but as core temperature further decreased from 33 degrees to 29 degrees C, the HR and MAP fell to prehypothermic levels. Plasma concentrations of NE and Epi increased significantly (p less than 0.01) as core temperature fell from 37 degrees to 31 degrees C, but as core temperature dropped from 31 degrees to 29 degrees C, plasma NE and Epi levels decreased towards prehypothermic concentrations. These findings indicate that the sympathetic nervous system (SNS) responds quickly to hypothermia but may be "switched off" at a threshold temperature of about 29 degrees C. We speculate that hypotensive patients with temperatures less than or equal to 29 degrees C may benefit from infusions of exogenous catecholamines, especially if there have been only minimal benefits achieved with conventional therapy such as fluids, and an increase in ambient temperature.

Hypocalcemia in critically ill patients.

Hypocalcemia is an important metabolic problem in critical care medicine. To determine the frequency of this problem and the patient subsets at risk, a retrospective study of a large series of ICU patients was performed. During the study period, 259 patients were admitted to the ICU, of whom 210 (81%) had a serum calcium (Ca++) measured. Of these 210 patients, 135 (64%) were hypocalcemic (serum Ca++ less than 8.5 mg/dl) and 75 (36%) were normocalcemic. Serum albumin concentration was less than 3.5 g/dl in 70% of the hypocalcemic patients who hd albumin measured, suggesting that the ionized Ca++ concentration may have been normal in many of these patients. On the other hand, 32% of the hypocalcemic patients were alkalotic (pH greater than or equal to 7.45) which indicates that ionized Ca++ levels may have been low because Ca++ binding to protein increases with alkalosis. Gastrointestinal bleeders and postabdominal surgery patients were more likely to have low total serum Ca++ whereas cardiac and neurosurgical patients were more likely to have a normal total serum Ca++ (p less than 0.05). Ionized Ca++ was calculated in 36 of the normocalcemic and 80 of the hypocalcemic patients. The hypocalcemic group had significantly lower ionized Ca++ levels when compared to those of the normocalcemic group (p less than 0.001). Patients with low serum Ca++ values spent a longer time in the ICU (p less than 0.01), had an increased incidence of renal failure and sepsis (p less than 0.01), had an increased mortality rate (p less than 0.001), and received a greater number of blood transfusions (p less than 0.001) than patients in the normocalcemic group. It is concluded that: (1) hypocalcemia is a frequent finding in critically ill patients; (2) determining ionized Ca++ levels is useful because many ICU patients have alterations in both arterial pH and serum albumin levels; (3) hypoalbuminemia, sepsis, red cell transfusions, and renal failure are predisposing factors for hypocalcemia; and (4) hypocalcemic patients do less well clinically than normocalcepatientsmic patients.

Marked stress hyperglycemia in a child.

A 16-month-old child is presented who developed severe "stress hyperglycemia" with the onset of a respiratory distress syndrome. The serum glucose rose from 74 mg/dl to 520 mg/dl and returned to 52 mg/dl all within 24 h. These variations in serum glucose were paralleled by a similar marked increase in plasma cortisol concentration (from 41 to 90 to 36 micrograms/dl). The patient was receiving no iv dextrose and required no exogenous insulin. These data are consistent with a diagnosis of stress hyperglycemia. This report is unique in that, to our knowledge, stress hyperglycemia to levels greater than 350 mg/dl have not been previously reported in either adults or children.

Plasma, urine, and CSF catecholamine concentrations during and after ketamine anesthesia.

Ketamine has been reported to increase plasma catecholamine concentrations. Prior investigations have only studied plasma catecholamine levels for short periods after iv ketamine. Because ketamine is one of the most frequently used anesthetic agents in critical care research, we evaluated ketamine's effect on catecholamines over a longer period of time. Plasma, urine, and CSF epinephrine (E) and norepinephrine (NE) concentrations were serially measured during a 2-h ketamine infusion and a subsequent 2-h "wake-up" period. No changes in heart rate, mean arterial blood pressure or urine, plasma, and CSF NE concentrations were noted during the 4-h study period, whereas there were significant (p less than 0.005) increases in urine, plasma, and CSF E levels during ketamine infusion but not during the wake-up period. An unexpected finding was that the baboons have very high basal plasma E levels versus those in humans. It is concluded that ketamine is a useful anesthetic agent for critical care research involving measurements of sympathetic nervous system activity. The interesting observation of high plasma levels in the baboon warrants further investigation.

Bedside blood glucose determinations in critical care medicine: a comparative analysis of two techniques.

The accuracy of a new reagent test strip (Chemstrip BG) for the quantitative estimation of whole blood glucose concentration was compared to that of a widely used test strip (Dextrostix). To perform the comparative study, venous blood specimens were obtained from 133 patients. Interpretations of reagent strip glucose readings were made by three observers who separately recorded their impressions from the two strips on each patient. These estimated levels were compared to quantitative (glucose oxidase) concentrations measured from the same specimen. Although there were significant correlations (p less than 0.001) between both reagent strips and the measured values, the Chemstrip BG results more closely approximately measured values for the range of glucose concentrations encountered (10-600 mg/dl). In addition, Chemstrip BG was easier to read and measured a wider range of values than Dextrostix and, unlike Dextrostix, Chemstrip BG strips could be stored for review at a later time. This new glucose test strip accurately and rapidly determines whole blood glucose concentration at the bedside.