Deep Learning Based Prediction of Pulmonary Hypertension in Newborns Using Echocardiograms.

Pulmonary hypertension (PH) in newborns and infants is a complex condition associated with several pulmonary, cardiac, and systemic diseases contributing to morbidity and mortality. Thus, accurate and early detection of PH and the classification of its severity is crucial for appropriate and successful management. Using echocardiography, the primary diagnostic tool in pediatrics, human assessment is both time-consuming and expertise-demanding, raising the need for an automated approach. Little effort has been directed towards automatic assessment of PH using echocardiography, and the few proposed methods only focus on binary PH classification on the adult population. In this work, we present an explainable multi-view video-based deep learning approach to predict and classify the severity of PH for a cohort of 270 newborns using echocardiograms. We use spatio-temporal convolutional architectures for the prediction of PH from each view, and aggregate the predictions of the different views using majority voting. Our results show a mean F1-score of 0.84 for severity prediction and 0.92 for binary detection using 10-fold cross-validation and 0.63 for severity prediction and 0.78 for binary detection on the held-out test set. We complement our predictions with saliency maps and show that the learned model focuses on clinically relevant cardiac structures, motivating its usage in clinical practice. To the best of our knowledge, this is the first work for an automated assessment of PH in newborns using echocardiograms.

Interpretable and intervenable ultrasonography-based machine learning models for pediatric appendicitis.

Appendicitis is among the most frequent reasons for pediatric abdominal surgeries. Previous decision support systems for appendicitis have focused on clinical, laboratory, scoring, and computed tomography data and have ignored abdominal ultrasound, despite its noninvasive nature and widespread availability. In this work, we present interpretable machine learning models for predicting the diagnosis, management and severity of suspected appendicitis using ultrasound images. Our approach utilizes concept bottleneck models (CBM) that facilitate interpretation and interaction with high-level concepts understandable to clinicians. Furthermore, we extend CBMs to prediction problems with multiple views and incomplete concept sets. Our models were trained on a dataset comprising 579 pediatric patients with 1709 ultrasound images accompanied by clinical and laboratory data. Results show that our proposed method enables clinicians to utilize a human-understandable and intervenable predictive model without compromising performance or requiring time-consuming image annotation when deployed. For predicting the diagnosis, the extended multiview CBM attained an AUROC of 0.80 and an AUPR of 0.92, performing comparably to similar black-box neural networks trained and tested on the same dataset.

Molecularly imprinted electrochemical sensor for the selective and sensitive determination of octreotide in cancer patient plasma sample.

In this study, a molecularly imprinted polymer film (P (ANI)@MIP) on the electrode surface was fabricated using aniline as a functional monomer and octreotide (OC) as a template molecule. The developed P (ANI)@MIP was electrochemically electropolymerized on a glassy carbon electrode (GCE) surface. Each step of MIP production was evaluated by viewing the [Fe (CN)6]3-/4- signal obtained using cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). The P (ANI)@MIP film layer was studied with a scanning electron microscope (SEM), Raman, and contact angle measurements. The parameters consisting of monomer, template ratio, cycle number, removal solution, removal time, and rebinding time were optimized to obtain the best electrochemical sensor. The developed method was validated in line with ICH guidelines. The linear range, LOD, and LOQ were found as 10-80 fM, 0.801 fM, and 2.670 fM, respectively. The selectivity of the method was tested with the response of somatostatin and lanreotide from the same growth hormone family by comparing the OC response. The developed P (ANI)@MIP/GCE sensor is the first reported method for electrochemical analysis of OC. The P (ANI)@MIP/GCE sensor exhibited high sensitivity and selectivity for OC. The novel MIP sensor was used to determine OC in cancer patient plasma samples. The concentration of OC in cancer patients varied between 8.98 ng/mL and 10.10 ng/mL.

GC-MS analysis of seven metabolites for the screening of pregnant women with Down Syndrome fetuses.

Down Syndrome is a genetic disorder caused by the presence of all or part of a third copy of chromosome 21. Metabolomics is identification and quantification of small-molecule metabolites (molecular weight <1000 Da) in tissues, cells and physiological fluids within a certain period time. Metabolites are intermediate products of various types of biochemical reactions that participate in bonding metabolic pathways. In this study, metabolites such as 2-Hydroxybutyric acid, 3-Hydroxybutyric acid, ß-Hydroxyisovaleric acid, Uracil, Glutamic acid, Maltose and Melezitose were chosen as the possible determinants/markers for the prenatal screening of Down Syndrome. Quantitative analysis of the metabolites conducted by GCMS method using 5 % phenyl / 95 % dimethylpolysiloxane (30 m ×0.25 mm, 0.25 µm film thickness) capillary column. The oven temperature was held constant at 60 °C for 1 min and ramped at 10 °C /min to 200 °C then ramped at 30 °C/min to 320 °C and hold for 6 min before cool-down, as helium mobile phase and flow rate of 2.8 mL/min and adding Myristic acid-d27 as an internal standard. Our method was validated by parameters of system suitability, stability, linearity, sensitivity, accuracy, precision, selectivity, robustness and ruggedness. The developed and validated method was applied to plasma samples taken from pregnant women with Down Syndrome (study group) and euploid fetuses (healthy group). The levels of these seven metabolites are statistically different (p < 0.05 for all) between the groups. It can be concluded that these relevant metabolites might be used for the prenatal screening of Down Syndrome.

Metabolic Infrastructure of Pregnant Women With Trisomy 21 Fetuses; Metabolomic Analysis.

We aimed to configure impaired/altered metabolomic profiles of pregnant women carrying Down syndrome (DS) fetuses. The study involved 21 and 32 pregnant women with DS and euploid fetuses, respectively, as determined by prenatal screening and diagnosis as part of an antenatal care program. Metabolomic analyses were carried out using gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-qTOF-MS) methods. A total of 95 metabolites were identified. GC-MS analysis indicated that levels of 2-hydroxybutyric acid, benzoic acid, nonanoic acid, 3-hydroxybutyric acid, and 2-ketoisocaproic acid were increased in the DS group, where beta-alanine, threonic acid, oxalic acid, alpha-tocopherol, uracil, 2-piperidone, and creatinine were decreased. However, LC-qTOF-MS analysis showed that lipid-related metabolites were decreased in women carrying DS fetuses, whereas creatine, N4-phosphoagmatine, citrate, 2,5-dioxopentanoate, 2-furoate, pyruvate, and fructose levels were increased. Pathway analysis was also performed using metabolites whose levels were significantly altered (p<0.05) between the groups, and the findings indicated that the biosynthesis pathways of aminoacyl-tRNA and "valine-leucine-isoleucine", and metabolism pathways of "glycine-serine-threonine", nitrogen, "alanine-aspartate-glutamate", propanoate, glycerophospholipid, cysteine, methionine, and phenylalanine were significantly altered. Our findings indicate a special type of metabolic status/syndrome in pregnant women with Down syndrome fetuses. It could be speculated that altered metabolic status might influence both gametogenesis and embryogenesis. Down syndrome is a complex genetic disorder that is important to detect prenatally, but may also be prevented by taking necessary precautions prior to pregnancy.

Spatial domain reconstruction for imaging speed-of-sound with pulse-echo ultrasound: simulation and in vivo study.

Despite many uses of ultrasound, some pathologies such as breast cancer still cannot reliably be diagnosed in either conventional B-mode ultrasound imaging nor with more recent ultrasound elastography methods. Speed-of-sound (SoS) is a quantitative imaging biomarker, which is sensitive to structural changes due to pathology, and hence could facilitate diagnosis. Full-angle ultrasound computed tomography (USCT) was proposed to obtain spatially-resolved SoS images, however, its water-bath setup involves practical limitations. To increase clinical utility and for widespread use, recently, a limited-angle Fourier-domain SoS reconstruction was proposed, however, it suffers from significant image reconstruction artifacts. In this work, we present a SoS reconstruction strategy, where the forward problem is formulated using differential time-of-flight measurements based on apparent displacements along different ultrasound wave propagation paths, and the inverse problem is solved in spatial-domain using a proposed total-variation scheme with spatially-varying anisotropic weighting to compensate for geometric bias from limited angle imaging setup. This is shown to be robust to missing displacement data and easily allow for incorporating any prior geometric information. In numerical simulations, SoS values in inclusions are accurately reconstructed with 90% accuracy up to a noise level of 50%. With respect to Fourier-domain reconstruction, our proposed method improved contrast ratio from 0.37 to 0.67 for even high noise levels such as 50%. Numerical full-wave simulation and our preliminary in vivo results illustrate the clinical applicability of our method in a breast cancer imaging setting. Our proposed method requires single-sided access to the tissue and can be implemented as an add-on to conventional ultrasound equipment, applicable to a range of transducers and applications.

Inverse Problem of Ultrasound Beamforming With Sparsity Constraints and Regularization.

Ultrasound (US) beamforming is the process of reconstructing an image from acquired echo traces on several transducer elements. Typical beamforming approaches, such as delay-and-sum, perform simple projection operations, while techniques using statistical information also exist, e.g., adaptive, phase coherence, delay-multiply-and-sum, and sparse coding approaches. Inspired by the feasibility and success of inverse problem (IP) formulations in several image reconstruction problems, such as computed tomography, we herein devise an IP approach for US beamforming. We define a linear forward model for the synthesis of the beamformed image, and solve its IP thanks to several intuitive and physics-based constraints and regularization terms proposed. These reflect the prior knowledge about the spectra of carrier signal and spatial coherence of modulated signal. These constraints admit effective formulation through sparse representations. Our proposed method was evaluated for plane-wave imaging (PWI) that transmits unfocused waves, enabling high frame rates with large field of view at the expense of much lower image quality with conventional beamforming techniques. Results are evaluated in numerical simulations, as well as tissue-mimicking phantoms and in vivo data provided by PWI challenge in medical US. The best results achieved by our proposed techniques are 0.39-mm full-width at half-maximum for spatial resolution and 16.3-dB contrast-to-noise ratio, using a single plane-wave transmit.

Robust motion tracking in liver from 2D ultrasound images using supporters.

PURPOSE: Effectiveness of image-guided radiation therapy with precise dose delivery depends highly on accurate target localization, which may involve motion during treatment due to, e.g., breathing and drift. Therefore, it is important to track the motion and adjust the radiation delivery accordingly. Tracking generally requires reliable target appearance and image features, whereas in ultrasound imaging acoustic shadowing and other artifacts may degrade the visibility of a target, leading to substantial tracking errors. To minimize such errors, we propose a method based on so-called supporters, a computer vision tracking technique. This allows us to leverage information from surrounding motion for improving robustness of motion tracking on 2D ultrasound image sequences of the liver. METHODS: Image features, potentially useful for predicting the target positions, are individually tracked, and a supporter model capturing the coupling of motion between these features and the target is learned on-line. This model is then applied to predict the target position, when the target cannot be otherwise tracked reliably. RESULTS: The proposed method was evaluated using the Challenge on Liver Ultrasound Tracking (CLUST)-2015 dataset. Leave-one-out cross-validation was performed on the training set of 24 2D image sequences of each 1-5 min. The method was then applied on the test set (24 2D sequences), where the results were evaluated by the challenge organizers, yielding 1.04 mm mean and 2.26 mm 95%ile tracking error for all targets. We also devised a simulation framework to emulate acoustic shadowing artifacts from the ribs, which showed effective tracking despite the shadows. CONCLUSIONS: Results support the feasibility and demonstrate the advantages of using supporters. The proposed method improves its baseline tracker, which uses optic flow and elliptic vessel models, and yields the state-of-the-art real-time tracking solution for the CLUST challenge.

Compliance boundary conditions for simulating deformations in a limited target region.

Patient-specific models in medical procedures are often limited to a relatively small region of interest due either to computational concerns or to the fact that only a part of anatomy could be observed in the input medical images. Thus, for deformable planning or training simulations, boundary conditions at the borders of such models are necessitated. Zero-displacement or -force constraints at outer boundaries are commonly used, with the assumption that the selected region is large enough to minimize effects on the deformable behavior inside the region of interest. This may, however, still result in errors and does require superfluous elements to extend models. In this work, a mixed boundary condition type, called compliance boundary condition, is proposed to constrain model boundaries. Different techniques to define and estimate these boundary constraints are studied with simulation experiments. Results are presented for palpation on 2D and 3D phantoms and needle insertion to a male pelvic anatomical model.