Search

Widespread adoption of optical diagnosis of colorectal neoplasia is prevented by suboptimal endoscopist performance and lack of standardized training and competence evaluation. We aimed to assess diagnostic accuracy of endoscopists in optical diagnosis of colorectal neoplasia in the framework of artificial intelligence (AI) validation studies. Literature searches of databases (PubMed/MEDLINE, EMBASE, Scopus) up to April 2022 were performed to identify articles evaluating accuracy of individual endoscopists in performing optical diagnosis of colorectal neoplasia within studies validating AI against a histologically verified ground-truth. The main outcomes were endoscopists' pooled sensitivity, specificity, positive and negative predictive value (PPV/NPV), positive and negative likelihood ratio (LR) and area under the curve (AUC for sROC) for predicting adenomas versus non-adenomas. Six studies with 67 endoscopists and 2085 (IQR: 115–243,5) patients were evaluated. Pooled sensitivity and specificity for adenomatous histology was respectively 84.5% (95% CI 80.3%–88%) and 83% (95% CI 79.6%–85.9%), corresponding to a PPV, NPV, LR+, LR− of 89.5% (95% CI 87.1%–91.5%), 75.7% (95% CI 70.1%–80.7%), 5 (95% CI 3.9%–6.2%) and 0.19 (95% CI 0.14%–0.25%). The AUC was 0.82 (CI 0.76–0.90). Expert endoscopists showed a higher sensitivity than non-experts (90.5%, [95% CI 87.6%–92.7%] vs. 75.5%, [95% CI 66.5%–82.7%], p < 0.001), and Eastern endoscopists showed a higher sensitivity than Western (85%, [95% CI 80.5%–88.6%] vs. 75.8%, [95% CI 70.2%–80.6%]). Quality was graded high for 3 studies and low for 3 studies. We show that human accuracy for diagnosis of colorectal neoplasia in the setting of AI studies is suboptimal. Educational interventions could benefit by AI validation settings which seem a feasible framework for competence assessment.

A comparison between ultra-fast X-ray CT and a wire-mesh sensor is presented. The measurements were carried out in a vertical pipe of 42 mm inner diameter, which was supplied with an air-water mixture. Both gas and liquid superficial velocities were varied. The X-ray CT delivered 263 frames per second, while the wire-mesh sensor was operated at a frequency four times higher. Two different gas injectors were used: 4 orifices of 5 mm diameter for creating large bubbles and gas plugs and a sintered plate with a pore size of 100 µm for generating a bubbly flow. It was found that the wire-mesh sensor has a significantly higher resolution than the X-ray CT. Small bubbles, which are clearly shown by the wire-mesh sensor, cannot be found in the CT images, because they cross the measuring plane before a complete scan can be performed. This causes artifacts in the reconstructed images, instead. Furthermore, there are large deviations between the quantitative information contained in the reconstructed tomographic 2D distributions and the gas fractions measured by the sensor, while the agreement is very good, when the gas fraction is obtained by a direct evaluation of the X-ray attenuation along the available through-transmission chords of the tomography setup. This shows that there are still potentials for an improvement of the image reconstruction method. Concerning the wire-mesh sensor it was found, that the gas fraction inside large bubbles is slightly underestimated. Furthermore, a significant distortion of large Taylor bubbles by the sensor was found for small liquid velocities up to 0.24 m/s. This effect vanished with growing superficial water velocity.

Mesoscale structure of 1-propanol aqueous solutions with propanol mole fraction xp ranging from 0.1 to 0.33 has been studied by means of small angle neutron scattering (SANS) and large-scale reverse Monte Carlo (RMC) technique. Analysis of the SANS intensities in terms of a fractal model shows that the fractal dimension df of mesoscale structure of the solution is about 1.8–1.9 for water-rich solution and about 1.5 for propanol-rich solution. Percolation analysis on the RMC results reveals that the water molecules and the propanol molecules cluster, respectively, as a mass fractal, the dimension dM of which is about 2.3–2.5 for both clusters for water-rich solution. Furthermore, the distribution of the cluster size is expressed by a simple power law with an exponent τ of about 1.35–1.5 for the propanol clusters and 1.05–1.2 for the water clusters. These results imply that the current solution is characterized by polydisperse mass fractals. In fact, a theoretical relation for polydisperse system of mass fractals, df=dM(2-τ), holds well in the current solution. The characteristic change in df from 1.8–1.9 to 1.5 described above is attributed to the crossover between the water-rich regime and the propanol-rich regime. Most of the water molecules and the propanol molecules are located on the interface between clusters, and the water molecules form thin layers of about 10 Å thick irrespective of 1-propanol content studied.

Void fraction distributions of air-water two-phase flow in 3 different diameter vertical tubes (0.042 m, 0.1 m, and 0.2 m) were measured by 16x16 wire-mesh sensors with a sampling frequency of 1 kHz. Also, predictions under the same flow conditions were made by a 3D extended two-fluid model, generating a time-series of cross-section averaged void fraction at the same frequency with the experiment. Fluctuation characteristics from experiments and simulations were analysed in view of the probability density function, frequency spectrum, and Trajectory Parallel Measures (TPM). The analyses of TPM showed that dynamics is well-established for large nunnles whose fiameter is on the order of Taylor wave length in an air-water system, although small bubble void fraction signal is almost random in dynamical point of view. It is shown that the fluctuation component in void fraction time series can be used as one of criteria for validation of numerical models.

Void fraction distributions of air-water two-phase flow in 3 different diameter vertical tubes (0.042 m, 0.1 m, and 0.2 m) were measured by 16x16 wire-mesh sensors with a sampling frequency of 1 kHz. Also, predictions under the same flow conditions were made by a 3D extended two-fluid model, generating a time-series of cross-section averaged void fraction at the same frequency with the experiment. Fluctuation characteristics from experiments and simulations were analysed in view of the probability density function, frequency spectrum, and Trajectory Parallel Measures (TPM). The analyses of TPM showed that dynamics is well-established for large nunnles whose fiameter is on the order of Taylor wave length in an air-water system, although small bubble void fraction signal is almost random in dynamical point of view. It is shown that the fluctuation component in void fraction time series can be used as one of criteria for validation of numerical models.

Void fraction distributions of air-water two-phase flow in 3 different diameter vertical tubes (0.042 m, 0.1 m, and 0.2 m) were measured by 16x16 wire-mesh sensors with a sampling frequency of 1 kHz. Also, predictions under the same flow conditions were made by a 3D extended two-fluid model, generating a time-series of cross-section averaged void fraction at the same frequency with the experiment. Fluctuation characteristics from experiments and simulations were analysed in view of the probability density function, frequency spectrum, and Trajectory Parallel Measures (TPM). The analyses of TPM showed that dynamics is well-established for large nunnles whose fiameter is on the order of Taylor wave length in an air-water system, although small bubble void fraction signal is almost random in dynamical point of view. It is shown that the fluctuation component in void fraction time series can be used as one of criteria for validation of numerical models.

Void fraction distributions of air-water two-phase flow in 3 different diameter vertical tubes (0.042 m, 0.1 m, and 0.2 m) were measured by 16x16 wire-mesh sensors with a sampling frequency of 1 kHz. Also, predictions under the same flow conditions were made by a 3D extended two-fluid model, generating a time-series of cross-section averaged void fraction at the same frequency with the experiment. Fluctuation characteristics from experiments and simulations were analysed in view of the probability density function, frequency spectrum, and Trajectory Parallel Measures (TPM). The analyses of TPM showed that dynamics is well-established for large nunnles whose fiameter is on the order of Taylor wave length in an air-water system, although small bubble void fraction signal is almost random in dynamical point of view. It is shown that the fluctuation component in void fraction time series can be used as one of criteria for validation of numerical models.

The present paper describes the concept of a fast scanning X-ray tomograph, the hardware development, and measurement results of gas-liquid two-phase flow in a vertical pipe. The device uses 18 pulsed X-ray sources activated in a successive order. In this way, a complete set of 18 independent projections of the object is obtained within 38 ms, i.e. the measuring rate is about 250 frames per second. Finally, to evaluate the measurement capability of the fast X-ray CT, a wire-mesh sensor was installed in the flow loop and both systems were operated for the same two-phase flow simultaneously. Comparison of the time series of the cross section averaged void fraction from both systems showed sufficient agreement for slug flow at large void fractions, while the fast CT underestimated the void fraction of bubbly flow especially in low void fraction range. For the wire-mesh sensor, coerced deformation of slug bubble interface was found. Further hardware improvement is in progress to achieve better resolution with the fast X-ray CT scanner.

Small angle neutron scattering (SANS) measurements have been carried out over a wide range of Q from 0.003 to 7 Å-1 on the 1-propanol-water solution with KCl and that without KCl in relation to the salt-induced reentrant phase-separation phenomena. The Q dependence of the SANS intensity for both solutions is characterized by the same fractal structure with the fractal dimension d[sub f] of 1.9. The only difference between the two solutions is the value of correlation length ξ. In the solution without KCl the value of ξ is rather small, ranging from 7.4 to 6.2 Å depending on temperature, while in the solution with KCl it becomes very large, from 48 to 355 Å or more depending on temperature, and diverges toward the phase-separation temperatures in a similar way for both the upper and lower one-phase regions. No significant differences have been found on the growing structure of fluctuation in the upper and lower one-phase regions.