Fatemi, Mostafa

Background: For COVID-19 lung severity, segmentation of lungs on computed tomography (CT) is the first crucial step. Current deep learning (DL)-based Artificial Intelligence (AI) models have a bias in the training stage of segmentation because only one set of ground truth (GT) annotations are evaluated. We propose a robust and stable inter-variability analysis of CT lung segmentation in COVID-19 to avoid the effect of bias. Methodology: The proposed inter-variability study consists of two GT tracers for lung segmentation on chest CT. Three AI models, PSP Net, VGG-SegNet, and ResNet-SegNet, were trained using GT annotations. We hypothesized that if AI models are trained on the GT tracings from multiple experience levels, and if the AI performance on the test data between these AI models is within the 5% range, one can consider such an AI model robust and unbiased. The K5 protocol (training to testing: 80%:20%) was adapted. Ten kinds of metrics were used for performance evaluation. Results: The database consisted of 5000 CT chest images from 72 COVID-19-infected patients. By computing the coefficient of correlations (CC) between the output of the two AI models trained corresponding to the two GT tracers, computing their differences in their CC, and repeating the process for all three AI-models, we show the differences as 0%, 0.51%, and 2.04% (all < 5%), thereby validating the hypothesis. The performance was comparable; however, it had the following order: ResNet-SegNet > PSP Net > VGG-SegNet. Conclusions: The AI models were clinically robust and stable during the inter-variability analysis on the CT lung segmentation on COVID-19 patients.

Abstract Objective . Detrusor overactivity (DO) is a urodynamic observation characterized by fluctuations in detrusor pressure ( P det ) of the bladder. Although detecting DO is important for the management of bladder symptoms, the invasive nature of urodynamic studies (UDS) makes it a source of discomfort and morbidity for patients. Ultrasound bladder vibrometry (UBV) could provide a direct and noninvasive means of detecting DO, due to its sensitivity to changes in elasticity and load in the bladder wall. In this study, we investigated the feasibility and applying UBV toward detecting DO. Approach . UBV and urodynamic study (UDS) measurements were collected in 76 neurogenic bladder patients (23 with DO). Timestamped group velocity squared ( c g 2 ) data series were collected from UBV measurements. Concurrent P det data series were identically analyzed for comparison and validation. A processing approach is developed to separate transient fluctuations in the data series from the larger trend of the data and a DO index is proposed for characterizing the transient peaks observed in the data. Main Results. Applying the DO index as a classifier for DO produced sensitivities and specificities of 0.70 and 0.75 for c g 2 data series and 0.70 and 0.83 for P det data series respectively. Significance . It was found that DO can be feasibly detected from data series of timestamped UBV measurements. Collectively, these initial results are promising, and further refinement to the UBV measurement process is likely to improve and clarify its capabilities for noninvasive detection of DO.

Background: COVID-19 lung segmentation using Computed Tomography (CT) scans is important for the diagnosis of lung severity. The process of automated lung segmentation is challenging due to (a) CT radiation dosage and (b) ground-glass opacities caused by COVID-19. The lung segmentation methodologies proposed in 2020 were semi- or automated but not reliable, accurate, and user-friendly. The proposed study presents a COVID Lung Image Analysis System (COVLIAS 1.0, AtheroPoint™, Roseville, CA, USA) consisting of hybrid deep learning (HDL) models for lung segmentation. Methodology: The COVLIAS 1.0 consists of three methods based on solo deep learning (SDL) or hybrid deep learning (HDL). SegNet is proposed in the SDL category while VGG-SegNet and ResNet-SegNet are designed under the HDL paradigm. The three proposed AI approaches were benchmarked against the National Institute of Health (NIH)-based conventional segmentation model using fuzzy-connectedness. A cross-validation protocol with a 40:60 ratio between training and testing was designed, with 10% validation data. The ground truth (GT) was manually traced by a radiologist trained personnel. For performance evaluation, nine different criteria were selected to perform the evaluation of SDL or HDL lung segmentation regions and lungs long axis against GT. Results: Using the database of 5000 chest CT images (from 72 patients), COVLIAS 1.0 yielded AUC of ~0.96, ~0.97, ~0.98, and ~0.96 (p-value < 0.001), respectively within 5% range of GT area, for SegNet, VGG-SegNet, ResNet-SegNet, and NIH. The mean Figure of Merit using four models (left and right lung) was above 94%. On benchmarking against the National Institute of Health (NIH) segmentation method, the proposed model demonstrated a 58% and 44% improvement in ResNet-SegNet, 52% and 36% improvement in VGG-SegNet for lung area, and lung long axis, respectively. The PE statistics performance was in the following order: ResNet-SegNet > VGG-SegNet > NIH > SegNet. The HDL runs in <1 s on test data per image. Conclusions: The COVLIAS 1.0 system can be applied in real-time for radiology-based clinical settings.

PurposeContrast‐free visualization of microvascular blood flow (MBF) using ultrasound can play a valuable role in diagnosis and detection of diseases. In this study, we demonstrate the importance of quantifying ensemble coherence for robust MBF imaging. We propose a novel approach to quantify ensemble coherence by estimating the local spatiotemporal correlation (LSTC) image, and evaluate its efficacy through simulation and in vivo studies.MethodsThe in vivo patient studies included three volunteers with a suspicious breast tumor, 15 volunteers with a suspicious thyroid tumor, and two healthy volunteers for renal MBF imaging. The breast data displayed negligible prior motion and were used for simulation analysis involving synthetically induced motion, to assess its impact on ensemble coherency and motion artifacts in MBF images. The in vivo thyroid data involved complex physiological motion due to its proximity to the pulsating carotid artery, which was used to assess the in vivo efficacy of the proposed technique. Further, in vivo renal MBF images demonstrated the feasibility of using the proposed ensemble coherence metric for curved array‐based MBF imaging involving phase conversion. All ultrasound data were acquired at high imaging frame rates and the tissue signal was suppressed using spatiotemporal clutter filtering. Thyroid tissue motion was estimated using two‐dimensional normalized cross correlation‐based speckle tracking, which was subsequently used for ensemble motion correction. The coherence of the MBF image was quantified based on Casorati correlation of the Doppler ensemble.ResultsThe simulation results demonstrated that an increase in ensemble motion corresponded with a decrease in ensemble coherency, which reciprocally degraded the MBF images. Further the data acquired from breast tumors demonstrated higher ensemble coherency than that from thyroid tumors. Motion correction improved the coherence of the thyroid MBF images, which substantially improved its visualization. The proposed coherence metrics were also useful in assessing the ensemble coherence for renal MBF imaging. The results also demonstrated that the proposed coherence metric can be reliably estimated from downsampled ensembles (by up to 90), thus allowing improved computational efficiency for potential applications in real‐time MBF imaging.ConclusionsThis pilot study demonstrates the importance of assessing ensemble coherency in contrast‐free MBF imaging. The proposed LSTC image quantified coherence of the Doppler ensemble for robust MBF imaging. The results obtained from this pilot study are promising, and warrant further development and in vivo validation.

Abstract In this study, we explored the feasibility of using functional ultrasound (fUS) imaging to visualize cerebral activation associated with thalamic deep brain stimulation (DBS), in rodents. The ventrolateral (VL) thalamus was stimulated using electrical pulses of low and high frequencies of 10 and 100 Hz, respectively, and multiple voltages (1–7 V) and pulse widths (50–1500 μ s). The fUS imaging demonstrated DBS-evoked activation of cerebral cortex based on changes of cerebral blood volume, specifically at the primary motor cortex (PMC). Low frequency stimulation (LFS) demonstrated significantly higher PMC activation compared to higher frequency stimulation (HFS), at intensities (5–7 V). Whereas, at lower intensities (1–3 V), only HFS demonstrated visible PMC activation. Further, LFS-evoked cerebral activation was was primarily located at the PMC. Our data presents the functionality and feasibility of fUS imaging as an investigational tool to identify brain areas associated with DBS. This preliminary study is an important stepping stone towards conducting real-time functional ultrasound imaging of DBS in awake and behaving animal models, which is of significant interest to the community for studying motor-related disorders.

Ultrasound measurements of detrusor muscle thickness have been proposed as a diagnostic biomarker in patients with bladder overactivity and voiding dysfunction. In this study, we present an approach based on deep learning (DL) and dynamic programming (DP) to segment the bladder sac and measure the detrusor muscle thickness from transabdominal 2D B-mode ultrasound images. To assess the performance of our method, we compared the results of automated methods to the manually obtained reference bladder segmentations and wall thickness measurements of 80 images obtained from 11 volunteers. It takes less than a second to segment the bladder from a 2D B-mode image for the DL method. The average Dice index for the bladder segmentation is 0.93 ± 0.04 mm, and the average root-mean-square-error and standard deviation for wall thickness measurement are 0.7 ± 0.2 mm, which is comparable to the manual ground truth. The proposed fully automated and fast method could be a useful tool for segmentation and wall thickness measurement of the bladder from transabdominal B-mode images. The computation speed and accuracy of the proposed method will enable adaptive adjustment of the ultrasound focus point, and continuous assessment of the bladder wall during the filling and voiding process of the bladder.

AbstractBackgroundShear wave elastography (SWE) shows promise in peripheral neuropathy evaluation but has potential limitations due to tissue size and heterogeneity. We tested SWE sensitivity to elasticity change and the effect of probe position in a median nerve cadaver model.MethodsTen specimens were used to measure median nerve elasticity under increasing loads using SWE and indentation. Measurements were compared using repeated‐measures analysis of variance.ResultsIndentation and SWE‐based longitudinal nerve elasticity increased with tensile loading (P< .01), showing a similar relationship. Acquisition in a transverse plane showed lower values compared with longitudinal measurements, mostly under higher loads (P= .03), as did postdissection elasticity (P= .02). Elasticity did not change when measured proximal to the carpal tunnel.ConclusionsLongitudinal SWE is sensitive to changes in median nerve elasticity. Measuring elasticity of peripheral nerves noninvasively could elucidate intra‐neural pathology related to compression neuropathies, and proof to be of added value as a diagnostic or prognostic tool.

Sparse array (SA) is an approach to reduce the number of system channels. However, SA suffers from grating lobe (GL) artefacts due to the sparsity of array aperture resulting in degradation of the ultrasound image quality. Based on a given or known data sets of radio frequency (RF) echo acquired from active elements of an array, RF echo data in unknown and/or inactive elements of array can be created virtually and used to suppress the GL artefact in SA. This letter presents gap-filling (GF) approaches to generate channel data, which are not physically acquired. It is demonstrated that the proposed GF technique can reduce the artefacts of SA by filling the gaps in the array aperture. Simulation results show that the GF technique can suppress the GL level of SA by up to 16 dB. Also, the GF technique can improve the image quality of fully sampled arrays with small number of active elements.

Vibrational characteristics of bone are directly dependent on its physical properties. In this study, a vibrational method for bone evaluation is introduced. We propose a new type of quantitative vibro-acoustic method based on the acoustic radiation force of ultrasound for bone characterization in persons with fracture. Using this method, we excited the clavicle or ulna by an ultrasound radiation force pulse which induces vibrations in the bone, resulting in an acoustic wave that is measured by a hydrophone placed on the skin. The acoustic signals were used for wave velocity estimation based on a cross-correlation technique. To further separate different vibration characteristics, we adopted a variational mode decomposition technique to decompose the received signal into an ensemble of band-limited intrinsic mode functions, allowing analysis of the acoustic signals by their constitutive components. This prospective study included 15 patients: 12 with clavicle fractures and three with ulna fractures. Contralateral intact bones were used as controls. Statistical analysis demonstrated that fractured bones can be differentiated from intact ones with a detection probability of 80%. Additionally, we introduce a “healing factor” to quantify the bone healing progress which successfully tracked the progress of healing in 80% of the clavicle fractures in the study.