A Fast Bubble Detection Method in Microtubes Based on Pulsed Ultrasound

In the process of biological microfluidic manipulation, the bubbles generated in the tube will seriously reduce the gauging accuracy. This paper introduces an improving method that can estimate the size of microbubbles in real time. Hence, the measurement data of the liquid volume can be modified according to this method. A microbubble detector based on the pulsed-ultrasound method was studied, including the device structure and the working principle. The assessment formula of the microbubbles in the tube was derived from the simulation results, which adopted the two-phase theory. The digital image processing method was applied to fulfill the microbubble calibration. This detection method was applied to measure the microbubbles in the tube and to modify the flow volume in a timely manner. The results of the experiments showed that this method is effective at improving the microflow gauging accuracy.

A Smart Classification Framework for Enhancing Reliability in Downhole Gas Bubble Sensing

Production logging tools (PLTs) and formation testing, even in logging while drilling (LWD) conditions during underbalanced drilling, are key technologies for assessing the productivity potential of a gas well and therefore to maximize recovery. Gas bubble detection sensors are key components in determining the fluid phases in the reservoir and accurately quantify recoverable reserves, optimize well placement, geosteering and to qualify the production ability of the well. We present here a new nonlinear autoregressive - breakdown artificial intelligence (AI) detection framework for PLT gas bubble detection sensors that categorize in real-time whether and which sensors become unreliable or have broken down during the logging measurements. AI tools allow the automatization of this method that is critical during data quality control of post-drilling PLT, but it is essential when the measurements are performed in LWD as data assessment and processing need to occur in real time. This AI framework was validated on both a training and testing dataset, and exhibited strong classification performance. This method enables accurate real-time breakdown detection for gas bubble detection sensors.

A Doppler ultrasound self-monitoring approach for detection of relevant individual decompression stress in scuba diving

Observing modern decompression protocols alone cannot fully prevent diving injuries especially in repetitive diving. Professional audio Doppler bubble measurements are not available to sports scuba divers. If those non-professionals were able to learn audio Doppler self-assessment for bubble grading, such skill could provide significant information on individual decisions with respect to diving safety. We taught audio Doppler self-assessment of subclavian and precordial probe position to 41 divers in a 45-min standardized, didactically optimized training. Assessment before and after air dives within sports diving limits was made through 684 audio Doppler measurements in dive-site conditions by both trained divers and a medical professional, plus additional 2D-echocardiography reference. In all dives (average maximum depth 22 m; dive time 44 min), 33% of all echocardiography measurements revealed bubbles. The specificity of audio bubble detection in combination of both detection sites was 95%, and sensitivity over all grades was 40%, increasing with higher bubble grades. Dive-site audio-Doppler-grading underestimated echo-derived bubble grades. Bubble detection sensitivity of audio Doppler self-assessments, compared to an experienced examiner, was 62% at subclavian and 73% at precordial position. 6 months after the training and 4.5 months after the last measurement, the achieved Doppler skill level remained stable. Audio Doppler self-assessment can be learned by non-professionals in a single teaching intervention. Despite accurate bubble grading is impossible in dive-site conditions, relevant high bubble grades can be detected by non-professionals. This qualitative information can be important in self-evaluating decompression stress and assessing measures for increased diving safety.

Recognition of overlapping elliptical objects in a binary image

Recognition of overlapping objects is required in many applications in the field of computer vision. Examples include cell segmentation, bubble detection and bloodstain pattern analysis. This paper presents a method to identify overlapping objects by approximating them with ellipses. The method is intended to be applied to complex-shaped regions which are believed to be composed of one or more overlapping objects. The method has two primary steps. First, a pool of candidate ellipses are generated by applying the Euclidean distance transform on a compressed image and the pool is filtered by an overlaying method. Second, the concave points on the contour of the region of interest are extracted by polygon approximation to divide the contour into segments. Then, the optimal ellipses are selected from among the candidates by choosing a minimal subset that best fits the identified segments. We propose the use of the adjusted Rand index, commonly applied in clustering, to compare the fitting result with ground truth. Through a set of computational and optimization efficiencies, we are able to apply our approach in complex images comprised of a number of overlapped regions. Experimental results on a synthetic data set, two types of cell images and bloodstain patterns show superior accuracy and flexibility of our method in ellipse recognition, relative to other methods.

Deep learning-based automated and universal bubble detection and mask extraction in complex two-phase flows

While investigating multiphase flows experimentally, the spatiotemporal variation in the interfacial shape between different phases must be measured to analyze the transport phenomena. For this, numerous image processing techniques have been proposed, showing good performance. However, they require trial-and-error optimization of thresholding parameters, which are not universal for all experimental conditions; thus, their accuracy is highly dependent on human experience, and the overall processing cost is high. Motivated by the remarkable improvements in deep learning-based image processing, we trained the Mask R-CNN to develop an automated bubble detection and mask extraction tool that works universally in gas–liquid two-phase flows. The training dataset was rigorously optimized to improve the model performance and delay overfitting with a finite amount of data. The range of detectable bubble size (particularly smaller bubbles) could be extended using a customized weighted loss function. Validation with different bubbly flows yields promising results, with AP50reaching 98%. Even while testing with bubble-swarm flows not included in the training set, the model detects more than 95% of the bubbles, which is equivalent or superior to conventional image processing methods. The pure processing speed for mask extraction is more than twice as fast as conventional approaches, even without counting the time required for tedious threshold parameter tuning. The present bubble detection and mask extraction tool is available online (https://github.com/ywflow/BubMask).