Brachial artery vasomotion and transducer pressure effect on measurements by active contour segmentation on ultrasound

MallesNet: A multi-object assistance based network for brachial plexus segmentation in ultrasound images

Ultrasound-guided injection is widely used to help anesthesiologists perform anesthesia in peripheral nerve blockade (PNB). However, it is a daunting task to accurately identify nerve structure in ultrasound images even for the experienced anesthesiologists. In this paper, a Multi-object assistance based Brachial Plexus Segmentation Network, named MallesNet, is proposed to improve the nerve segmentation performance in ultrasound image with the assistance of simultaneously segmenting its surrounding anatomical structures (e.g., muscle, vein, and artery). The MallesNet is designed by following the framework of Mask R-CNN to implement the multi object identification and segmentation. Moreover, a spatial local contrast feature (SLCF) extraction module is proposed to compute contrast features at different scales to effectively obtain useful features for small objects. And the self-attention gate (SAG) is also utilized to capture the spatial relationships in different channels and further re-weight the channels in feature maps by following the design of non-local operation and channel attention. Furthermore, the upsampling mechanism in original Feature Pyramid Network (FPN) is improved by adopting the transpose convolution and skip concatenation to fine-tune the feature maps. The Ultrasound Brachial Plexus Dataset (UBPD) is also proposed to support the research on brachial plexus segmentation, which consists of 1055 ultrasound images with four objects (i.e., nerve, artery, vein and muscle) and their corresponding label masks. Extensive experimental results using UBPD dataset demonstrate that MallesNet can achieve a better segmentation performance on nerves structure and also on surrounding structures in comparison to other competing approaches.

Bio-Impedance Sensor for Real-Time Artery Diameter Waveform Assessment

The real-time artery diameter waveform assessment during cardio cycle can allow the measurement of beat-to-beat pressure change and the long-term blood pressure monitoring. The aim of this study is to develop a self-calibrated bio-impedance-based sensor, which can provide regular measurement of the blood-pressure-dependence time variable parameters such as the artery diameter waveform and the elasticity. This paper proposes an algorithm based on analytical models which need prior geometrical and physiological patient parameters for more appropriate electrode system selection and hence location to provide accurate blood pressure measurement. As a result of this study, the red cell orientation effect contribution was estimated and removed from the bio-impedance signal obtained from the artery to keep monitoring the diameter waveform correspondence to the change of blood pressure.

Boundary Restored Network for Subpleural Pulmonary Lesion Segmentation on Ultrasound Images at Local and Global Scales

To evaluate the application of machine learning for the detection of subpleural pulmonary lesions (SPLs) in ultrasound (US) scans, we propose a novel boundary-restored network (BRN) for automated SPL segmentation to avoid issues associated with manual SPL segmentation (subjectivity, manual segmentation errors, and high time consumption). In total, 1612 ultrasound slices from 255 patients in which SPLs were visually present were exported. The segmentation performance of the neural network based on the Dice similarity coefficient (DSC), Matthews correlation coefficient (MCC), Jaccard similarity metric (Jaccard), Average Symmetric Surface Distance (ASSD), and Maximum symmetric surface distance (MSSD) was assessed. Our dual-stage boundary-restored network (BRN) outperformed existing segmentation methods (U-Net and a fully convolutional network (FCN)) for the segmentation accuracy parameters including DSC (83.45 ± 16.60%), MCC (0.8330 ± 0.1626), Jaccard (0.7391 ± 0.1770), ASSD (5.68 ± 2.70 mm), and MSSD (15.61 ± 6.07 mm). It also outperformed the original BRN in terms of the DSC by almost 5%. Our results suggest that deep learning algorithms aid fully automated SPL segmentation in patients with SPLs. Further improvement of this technology might improve the specificity of lung cancer screening efforts and could lead to new applications of lung US imaging.

Brachial flow-mediated dilation by continuous monitoring of arterial cross-section with ultrasound imaging

Objective

Impairment of flow-mediated dilation of the brachial artery is a marker of endothelial dysfunction and often predisposes atherosclerosis and cardiovascular events. In this study, we propose a user-guided automated approach for monitoring arterial cross-section during hyperemic response to improve reproducibility and sensitivity of flow-mediated dilation.

Material and methods

Ultrasound imaging of the brachial artery was performed in 11 volunteers in cross-sectional and in 5 volunteers in longitudinal view. During each examination, images were recorded continuously before and after inducing ischemia. Time-dilation curves of the brachial lumen cross-section were measured by user-guided automated segmentation of brachial images with the feed-forward active contour (FFAC) algorithm. %FMD was determined by the ratio of peak dilation to the baseline value. Each measurement was repeated twice in two sessions 1 h apart on the same arm to evaluate the reproducibility of the measurements. The intra-subject variation in flow-mediated dilation between two sessions (subject-specific) and inter-group variation in flow-mediated dilation with all the subjects within a session grouped together (group-specific) were measured for FFAC. The FFAC measurements were compared with the conventional diameter measurements made using echo tracking in longitudinal views.

Results

Flow-mediated dilation values for cross-sectional area were greater than those measured by diameter dilation: 33.1% for cross-sectional area compared to 22.5% for diameter. Group-specific flow-mediated dilation measurements for cross-sectional area were highly reproducible: 33.2% vs. 33.0% (p > 0.05) with coefficient of variation CV of 0.4%. The group-specific flow-mediated dilations measured by echo tracking for the two sessions were 21.1 vs. 23.9% with CV of 9%. Subject-specific CV for cross-sectional area by FFAC was 10% ± 2% versus 24% ± 10% for the conventional approach. Using correlation as a metric of evaluation also showed better performance for cross-sectional imaging: correlation coefficient, R, between two sessions for cross-sectional area was 0.92 versus 0.72 for the conventional approach based on diameter measurements.

Conclusion

Peak dilation area measured by continuous automated monitoring of cross-sectional area of the brachial artery provides more reproducible and higher-sensitivity measurement of flow-mediated dilation compared to the conventional approach of using vascular diameter measured using longitudinal imaging.

Comparison of hemodialysis arteriovenous fistula blood flow rates measured by Doppler ultrasound and phase-contrast magnetic resonance imaging

OBJECTIVE

The objective of this study was to compare blood flow rates measured by Doppler ultrasound (DUS) and phase-contrast magnetic resonance imaging (MRI) in patients having a hemodialysis arteriovenous fistula (AVF) and to identify scenarios in which there was significant discordance between these two approaches.

METHODS

Blood flow rates in the proximal artery (PA) and draining vein (DV) of newly created upper extremity AVFs were measured and compared using DUS and phase-contrast MRI at 1 day, 6 weeks, and 6 months postoperatively.

RESULTS

Blood flow rates in the PA measured by DUS (1155 ± 907 mL/min, mean ± standard deviation) and by MRI (1170 ± 657 mL/min) were not statistically different (P = .812) based on 78 data pairs from 49 patients. DV DUS flow (1277 ± 995 mL/min) and MRI flow (1130 ± 655 mL/min) were also not statistically different (P = .071) based on 64 data pairs. In both PA and DV, the two methods substantially agreed with each other (Cohen κ: PA, 0.66; DV, 0.67) when flow rates were put into four clinically relevant categories (<300, 300-599, 600-1499, and ≥1500 mL/min). The Bland-Altman analyses of DUS and MRI flow identified six and four outliers for PA and DV, respectively. Seven outliers had higher DUS than MRI flow, with all DUS scan sites having a large lumen or significant local curvature; the other three had lower DUS flow, partly due to an underestimation of lumen diameter by DUS.

CONCLUSIONS

DUS and MRI flow rates are generally comparable in both PA and DV. When DUS is used for flow measurements, careful attention to accurate lumen diameter measurements is needed and scan sites with marked curvature should be avoided. Our result may improve the accuracy of DUS-measured AVF blood flow rate.

Machine learning for medical ultrasound: status, methods, and future opportunities

Ultrasound (US) imaging is the most commonly performed cross-sectional diagnostic imaging modality in the practice of medicine. It is low-cost, non-ionizing, portable, and capable of real-time image acquisition and display. US is a rapidly evolving technology with significant challenges and opportunities. Challenges include high inter- and intra-operator variability and limited image quality control. Tremendous opportunities have arisen in the last decade as a result of exponential growth in available computational power coupled with progressive miniaturization of US devices. As US devices become smaller, enhanced computational capability can contribute significantly to decreasing variability through advanced image processing. In this paper, we review leading machine learning (ML) approaches and research directions in US, with an emphasis on recent ML advances. We also present our outlook on future opportunities for ML techniques to further improve clinical workflow and US-based disease diagnosis and characterization.

Feed-forward active contour analysis for improved brachial artery reactivity testing

The object of this study was to utilize a novel feed-forward active contour (FFAC) algorithm to find a reproducible technique for analysis of brachial artery reactivity. Flow-mediated dilation (FMD) is an important marker of vascular endothelial function but has not been adopted for widespread clinical use given its technical limitations, including inter-observer variability and differences in technique across clinical sites. We developed a novel FFAC algorithm with the goal of validating a more reliable standard. Forty-six healthy volunteers underwent FMD measurement according to the standard technique. Ultrasound videos lasting 5-10 seconds each were obtained pre-cuff inflation and at minutes 1 through 5 post-cuff deflation in longitudinal and transverse views. Automated segmentation using the FFAC algorithm with initial boundary definition from three different observers was used to analyze the images to measure diameter/cross-sectional area over the cardiac cycle. The %FMD was calculated for average, minimum, and maximum diameters/areas. Using the FFAC algorithm, the population-specific coefficient of variation (CV) at end-diastole was 3.24% for transverse compared to 9.96% for longitudinal measurements; the subject-specific CV was 15.03% compared to 57.41%, respectively. For longitudinal measurements made via the conventional method, the population-specific CV was 4.77% and subject-specific CV was 117.79%. The intraclass correlation coefficient (ICC) for transverse measurements was 0.97 (95% CI: 0.95-0.98) compared to 0.90 (95% CI: 0.84-0.94) for longitudinal measurements with FFAC and 0.72 (95% CI: 0.51-0.84) for conventional measurements. In conclusion, transverse views using the novel FFAC method provide less inter-observer variability than traditional longitudinal views. Improved reproducibility may allow adoption of FMD testing in a clinical setting. The FFAC algorithm is a robust technique that should be evaluated further for its ability to replace the more limited conventional technique for measurement of FMD.

Elevated C-reactive protein levels and enhanced high frequency vasomotion in patients with ischemic heart disease during brachial flow-mediated dilation

PURPOSE

The physiological role of vasomotion, rhythmic oscillations in vascular tone or diameter, and its underlying mechanisms are unknown. We investigated the characteristics of brachial artery vasomotion in patients with ischemic heart disease (IHD).

METHODS

We performed a retrospective study of 37 patients with IHD. Endothelial function was assessed using flow-mediated dilation (FMD), and power spectral analysis of brachial artery diameter oscillations during FMD was performed. Frequency-domain components were calculated by integrating the power spectrums in three frequency bands (in ms2) using the MemCalc (GMS, Tokyo, Japan): very-low frequency (VLF), 0.003-0.04 Hz; low frequency (LF), 0.04-0.15 Hz; and high frequency (HF), 0.15-0.4 Hz. Total spectral power (TP) was calculated as the sum of all frequency bands, and each spectral component was normalized against TP.

RESULTS

Data revealed that HF/TP closely correlated with FMD (r = -0.33, p = 0.04), whereas VLF/TP and LF/TP did not. We also explored the relationship between elevated C-reactive protein (CRP) levels and vasomotion. HF/TP was significantly increased in subjects with high CRP levels (CRP;>0.08 mg/dL) compared with subjects with low CRP levels (0.052±0.026 versus 0.035±0.022, p<0.05). The HF/TP value closely correlated with CRP (r = 0.24, p = 0.04), whereas the value of FMD did not (r = 0.023, p = 0.84). In addition, elevated CRP levels significantly increased the value of HF/TP after adjustment for FMD and blood pressure (β = 0.33, p<0.05).

CONCLUSION

The HF component of brachial artery diameter oscillation during FMD measurement correlated well with FMD and increased in the presence of elevated CRP levels in subjects with IHD.