Deep Learning for FAST Quality Assessment.

OBJECTIVES: To determine the feasibility of using a deep learning (DL) algorithm to assess the quality of focused assessment with sonography in trauma (FAST) exams. METHODS: Our dataset consists of 441 FAST exams, classified as good-quality or poor-quality, with 3161 videos. We first used convolutional neural networks (CNNs), pretrained on the Imagenet dataset and fine-tuned on the FAST dataset. Second, we trained a CNN autoencoder to compress FAST images, with a 20-1 compression ratio. The compressed codes were input to a two-layer classifier network. To train the networks, each video was labeled with the quality of the exam, and the frames were labeled with the quality of the video. For inference, a video was classified as poor-quality if half the frames were classified as poor-quality by the network, and an exam was classified as poor-quality if half the videos were classified as poor-quality. RESULTS: The results with the encoder-classifier networks were much better than the transfer learning results with CNNs. This was primarily because the Imagenet dataset is not a good match for the ultrasound quality assessment problem. The DL models produced video sensitivities and specificities of 99% and 98% on held-out test sets. CONCLUSIONS: Using an autoencoder to compress FAST images is a very effective way to obtain features that can be used to predict exam quality. These features are more suitable than those obtained from CNNs pretrained on Imagenet.

Exact time-dependent analytical solutions for entropy production rate in a system operating in a heat bath in which temperature varies linearly in space.

The nonequilibrium thermodynamics feature of a Brownian motor is investigated by obtaining exact time-dependent solutions. This in turn enables us to investigate not only the long time property (steady state) but also the short time the behavior of the system. The general expressions for the free energy, entropy production e[over ̇]_{p}(t) as well as entropy extraction h[over ̇]_{d}(t) rates are derived for a system that is genuinely driven out of equilibrium by time-independent force as well as by spatially varying thermal background. We show that for a system that operates between hot and cold reservoirs, most of the thermodynamics quantities approach a nonequilibrium steady state in the long time limit. The change in free energy becomes minimal at a steady state. However, for a system that operates in a heat bath where its temperature varies linearly in space, the entropy production and extraction rates approach a nonequilibrium steady state while the change in free energy varies linearly in space. This reveals that unlike systems at equilibrium, when systems are driven out of equilibrium, their free energy may not be minimized. The thermodynamic properties of a system that operates between the hot and cold baths are further compared and contrasted with a system that operates in a heat bath where its temperature varies linearly in space along with the reaction coordinate. We show that the entropy, entropy production, and extraction rates are considerably larger for the linearly varying temperature case than a system that operates between the hot and cold baths revealing such systems are inherently irreversible. For both cases, in the presence of load or when a distinct temperature difference is retained, the entropy S(t) monotonously increases with time and saturates to a constant value as t further steps up. The entropy production rate e[over ̇]_{p} decreases in time and at steady state, e[over ̇]_{p}=h[over ̇]_{d}>0, which agrees with the results shown in M. Asfaw's [Phys. Rev. E 89, 012143 (2014)1539-375510.1103/PhysRevE.89.012143; Phys. Rev. E 92, 032126 (2015)10.1103/PhysRevE.92.032126]. Moreover, the velocity, as well as the efficiency of the system that operates between the hot and cold baths, are also collated and contrasted with a system that operates in a heat bath where its temperature varies linearly in space along with the reaction coordinate. A system that operates between the hot and cold baths has significantly lower velocity but a higher efficiency in comparison with a linearly varying temperature case.

Effect of viscous friction on entropy, entropy production, and entropy extraction rates in underdamped and overdamped media.

Considering viscous friction that varies spatially and temporally, the general expressions for entropy production, free energy, and entropy extraction rates are derived to a Brownian particle that walks in overdamped and underdamped media. Via the well known stochastic approaches to underdamped and overdamped media, the thermodynamic expressions are first derived at a trajectory level then generalized to an ensemble level. To study the nonequilibrium thermodynamic features of a Brownian particle that hops in a medium where its viscosity varies on time, a Brownian particle that walks on a periodic isothermal medium (in the presence or absence of load) is considered. The exact analytical results depict that in the absence of load f=0, the entropy production rate e[over ̇]_{p} approaches the entropy extraction rate h[over ̇]_{d}=0. This is reasonable since any system which is in contact with a uniform temperature should obey the detail balance condition in a long time limit. In the presence of load and when the viscous friction decreases either spatially or temporally, the entropy S(t) monotonously increases with time and saturates to a constant value as t further steps up. The entropy production rate e[over ̇]_{p} decreases in time and at steady state (in the presence of load) e[over ̇]_{p}=h[over ̇]_{d}>0. On the contrary, when the viscous friction increases either spatially or temporally, the rate of entropy production as well as the rate of entropy extraction monotonously steps up showing that such systems are inherently irreversible. Furthermore, considering a spatially varying viscosity, the nonequilibrium thermodynamic features of a Brownian particle that hops in a ratchet potential with load is explored. In this case, the direction of the particle velocity is dictated by the magnitude of the external load of f. Far from the stall load, e[over ̇]_{p}=h[over ̇]_{d}>0 and at stall force e[over ̇]_{p}=h[over ̇]_{d}=0 revealing the system is reversible at this particular choice of parameter. In the absence of load, e[over ̇]_{p}=h[over ̇]_{d}>0 as long as a distinct temperature difference is retained between the hot and cold baths. Moreover, considering a multiplicative noise, we explore the thermodynamic features of the model system.

Sedimentation rate of erythrocyte from physics prospective.

An erythrocytes sedimentation rate (ESR) measures how fast a blood sample sediments along a test tube in one hour in a clinical laboratory. Since elevated level of ESR is associated with inflammatory diseases, ESR is one of the routine hematology test in a clinical laboratory. In this paper, the physics of erythrocyte (RBC) sedimentation rate as well as the dynamics of the RBC is explored by modeling the dynamics of the cells as the motion of Brownian particle moving in a viscous medium. The viscous friction of blood γ is considered to decrease as the temperature of the medium increases. The results obtained in this work show that the ESR increases as the number of red blood cells (that bind together in the sedimentation process) steps up. The room temperature also affects the sedimentation rate. As the room temperature rises up, the ESR steps up. Furthermore the dynamics of the RBC along a Westergren pipet that is held in an upright position is explored. The exact analytic result depicts that the velocity of cells increases as the number of cells that form rouleaux steps up. Since our study is performed by considering physiological parameters, the results obtained in this work not only can be justified experimentally but also helps to understand most hematological experiments that are conducted in vitro.

Entropy production and entropy extraction rates for a Brownian particle that walks in underdamped medium.

The expressions for entropy production, free energy, and entropy extraction rates are derived for a Brownian particle that walks in an underdamped medium. Our analysis indicates that as long as the system is driven out of equilibrium, it constantly produces entropy at the same time it extracts entropy out of the system. At steady state, the rate of entropy production e[over ̇]_{p} balances the rate of entropy extraction h[over ̇]_{d}. At equilibrium both entropy production and extraction rates become zero. The entropy production and entropy extraction rates are also sensitive to time. As time progresses, both entropy production and extraction rates increase in time and saturate to constant values. Moreover, employing microscopic stochastic approach, several thermodynamic relations for different model systems are explored analytically and via numerical simulations by considering a Brownian particle that moves in overdamped medium. Our analysis indicates that the results obtained for underdamped cases quantitatively agree with overdamped cases at steady state. The fluctuation theorem is also discussed.

Free energy and entropy production rate for a Brownian particle that walks on overdamped medium.

We derive general expressions for the free energy, entropy production, and entropy extraction rates for a Brownian particle that walks in a viscous medium where the dynamics of its motion is governed by the Langevin equation. It is shown that, when the system is out of equilibrium, it constantly produces entropy and at the same time extracts entropy out of the system. Its entropy production and extraction rates decrease in time and saturate to a constant value. In the long-time limit, the rate of entropy production balances the rate of entropy extraction and, at equilibrium, both entropy production and extraction rates become zero. Moreover, considering different model systems, not only do we investigate how various thermodynamic quantities behave in time but also we discuss the fluctuation theorem in detail.

Exact analytical thermodynamic expressions for a Brownian heat engine.

The nonequilibrium thermodynamics feature of a Brownian motor operating between two different heat baths is explored as a function of time t. Using the Gibbs entropy and Schnakenberg microscopic stochastic approach, we find exact closed form expressions for the free energy, the rate of entropy production, and the rate of entropy flow from the system to the outside. We show that when the system is out of equilibrium, it constantly produces entropy and at the same time extracts entropy out of the system. Its entropy production and extraction rates decrease in time and saturate to a constant value. In the long time limit, the rate of entropy production balances the rate of entropy extraction, and at equilibrium both entropy production and extraction rates become zero. Furthermore, via the present model, many thermodynamic theories can be checked.

Upper gastrointestinal endoscopy: a review of 10,000 cases.

A total of 10,000 patients underwent upper gastrointestinal endosopy examination between August 1979 and October 1994 at Tikur Anbessa Hospital, Addis Ababa. The major indications were dyspepsia (59.4%), upper gastrointestinal bleeding (18%) and liver disease (10.8%). The other indications include dysphagia (2.2%), gastric outlet obstruction (2.1%), postoperative dyspeptic symptoms (1.9%), weight loss and/or anemia (1.4%), epigastric mass (0.6%) and odynophagia 0.2%. The mean age of the patients and their sex ratio was 36 years and 2:1, respectively. Twenty eight percent of the patients had normal findings. The commonest abnormal findings include duodenal ulcer (41%), esophageal varices (9%), acute gastritis (6%), duodenitis (3.4%), and reflux esophagitis (2.3%). Benign gastric ulcer was rare. The ratio of duodenal ulcer to gastric ulcer was 19.1%. Duodenal ulcer (45.6%), esophageal varices (15.6) and acute gastritis (5.7%) were found to be the commonest causes of upper gastrointestinal bleeding. The endoscopy or histology diagnosis of cancer in both the esophagus and stomach was 2.8% and 1.3%, respectively. The agreement between endoscopy and histology in the diagnosis of esophageal and gastric cancer was 80%. There was no major complication related to endoscopy or premeditation. Endoscopy is a fairly accurate and safe procedure and therefore should be available and applied widely for the diagnosis of upper gastrointestinal diseases in Ethiopia.