A deep learning classifier for digital breast tomosynthesis.

PURPOSE: To develop a computerized detection system for the automatic classification of the presence/absence of mass lesions in digital breast tomosynthesis (DBT) annotated exams, based on a deep convolutional neural network (DCNN). MATERIALS AND METHODS: Three DCNN architectures working at image-level (DBT slice) were compared: two state-of-the-art pre-trained DCNN architectures (AlexNet and VGG19) customized through transfer learning, and one developed from scratch (DBT-DCNN). To evaluate these DCNN-based architectures we analysed their classification performance on two different datasets provided by two hospital radiology departments.DBT slice images were processed following normalization, background correction and data augmentation procedures. The accuracy, sensitivity, and area-under-the-curve (AUC) values were evaluated on both datasets, using receiver operating characteristic curves. A Grad-CAM technique was also implemented providing anindication of the lesion position in the DBT slice. RESULTS: Accuracy, sensitivity and AUC for the investigated DCNN are in-line with the best performance reported in the field. The DBT-DCNN network developed in this work showed an accuracy and a sensitivity of (90% ± 4%) and (96% ± 3%), respectively, with an AUC as good as 0.89 ± 0.04. Ak-fold cross validation test (withk = 4) showed an accuracy of 94.0% ± 0.2%, and a F1-score test provided a value as good as 0.93 ± 0.03. Grad-CAM maps show high activation in correspondence of pixels within the tumour regions. CONCLUSIONS: We developed a deep learning-based framework (DBT-DCNN) to classify DBT images from clinical exams. We investigated also apossible application of the Grad-CAM technique to identify the lesion position.

Ethanol-like effects of thiopental and ketamine in healthy humans.

The gamma-aminobutyric acid-A (GABA(A)) and N-methyl-D-aspartate (NMDA) receptors mediate aspects of the behavioural effects of alcohol. Prior studies reported drugs that block NMDA receptors or facilitate GABA(A) receptor function produce ethanol-like effects in humans. The purpose of this study was to compare the ethanol-related effects of two pharmacological agents with known NMDA and GABA(A) receptor activity. As part of an ongoing, larger study, 28 subjects (age, 21-30) with no personal or family histories of alcoholism were administered subanesthetic doses of the GABA(A) receptor agonist thiopental, the NMDA receptor antagonist, ketamine and placebo on three separate test days. Various ethanol-related measures were administered. At doses of thiopental and ketamine that produced similar levels of sedation and cognitive effects, both agents produced significant ethanol-like effects and subjective intoxication. However, the intensity of the ethanol-like effects of ketamine was greater than that of thiopental. In addition, ketamine produced alterations in perception that were not produced by thiopental. These data provide further support for a model where GABA(A) receptor facilitation may contribute significantly to ethanol effects associated with social drinking, whereas NMDA receptor antagonism may contribute to relatively greater extent to features of ethanol 'intoxication'.

Esmolol infusion during nitroprusside-induced hypotension: impact on hemodynamics, ventricular performance, and venous admixture.

The impact of esmolol infusion on hemodynamics, ventricular performance, venous admixture, sympathoadrenal, and renin-angiotensin system responses during sodium nitroprusside (SNP)-induced hypotension was studied in 11 patients undergoing lymph node dissection during general anesthesia with 60% nitrous oxide and fentanyl. Radial arterial and thermistor-tipped pulmonary catheters were employed for hemodynamic monitoring. Arterial and mixed venous blood gas tensions, arterial plasma renin activity (PRA), and plasma catecholamine levels were measured. Derived hemodynamic parameters and venous admixture (Qs/Qt) data were obtained from standard equations. Transesophageal echocardiography (6 patients) was used to assess left ventricular performance using the relationship between end-systolic wall stress (ESWS) and velocity of circumferential shortening (VCFC). After surgical incision, arterial hypotension was induced with SNP alone. Esmolol was infused at each of the following rates in sequence: 200, 300, and 400 micrograms/kg/min. Each esmolol infusion lasted 20 minutes and the SNP dose was adjusted to maintain MAP at 55 to 60 mm Hg. The mean dose of SNP required to induce hypotension was 5.5 micrograms/kg/min +/- 0.5 SE. Compared to prehypotension values, SNP induced significant increases in Qs/Qt and reductions in PaO2, systemic vascular resistance (SVR), and stroke volume index (SVI). Esmolol infusion caused dose-dependent (highest with 400 micrograms/kg/min) reductions in the SNP requirement, heart rate (HR), SVI, Qs/Qt, and PRA, and also led to significant increases in SVR and left ventricular (LV) internal diameter in diastole as well as systole. Furthermore, esmolol infusion was associated with a dose-dependent downward and leftward shift of the ESWS versus VCFC relationship, implying diminished contractility.(ABSTRACT TRUNCATED AT 250 WORDS)