A wavelet neural operator based elastography for localization and quantification of tumors.

BACKGROUND AND OBJECTIVES: The application of intelligent imaging techniques and deep learning in the field of computer-aided diagnosis and medical imaging have improved and accelerated the early diagnosis of many diseases. Elastography is an imaging modality where an inverse problem is solved to extract the elastic properties of tissues and subsequently mapped to anatomical images for diagnostic purposes. In the present work, we propose a wavelet neural operator-based approach for correctly learning the non-linear mapping of elastic properties directly from measured displacement field data. METHODS: The proposed framework learns the underlying operator behind the elastic mapping and thus can map any displacement data from a family to the elastic properties. The displacement fields are first uplifted to a high-dimensional space using a fully connected neural network. On the lifted data, certain iterations are performed using wavelet neural blocks. In each wavelet neural block, the lifted data are decomposed into low, and high-frequency components using wavelet decomposition. To learn the most relevant patterns and structural information from the input, the neural network kernels are directly convoluted with the outputs of the wavelet decomposition. Thereafter the elasticity field is reconstructed from the outputs from convolution. The mapping between the displacement and the elasticity using wavelets is unique and remains stable during training. RESULTS: The proposed framework is tested on several artificially fabricated numerical examples, including a benign-cum-malignant tumor prediction problem. The trained model was also tested on real Ultrasound-based elastography data to demonstrate the applicability of the proposed scheme in clinical usage. The proposed framework reproduces the highly accurate elasticity field directly from the displacement inputs. CONCLUSIONS: The proposed framework circumvents different data pre-processing and intermediate steps utilized in traditional methods, hence providing an accurate elasticity map. The computationally efficient framework requires fewer epochs for training, which bodes well for its clinical usability for real-time predictions. The weights and biases from pre-trained models can also be employed for transfer learning, which reduces the effective training time with random initialization.

Source Aware Deep Learning Framework for Hand Kinematic Reconstruction Using EEG Signal.

The ability to reconstruct the kinematic parameters of hand movement using noninvasive electroencephalography (EEG) is essential for strength and endurance augmentation using exoskeleton/exosuit. For system development, the conventional classification-based brain-computer interface (BCI) controls external devices by providing discrete control signals to the actuator. A continuous kinematic reconstruction from EEG signal is better suited for practical BCI applications. The state-of-the-art multivariable linear regression (mLR) method provides a continuous estimate of hand kinematics, achieving a maximum correlation of up to 0.67 between the measured and the estimated hand trajectory. In this work, three novel source aware deep learning models are proposed for motion trajectory prediction (MTP). In particular, multilayer perceptron (MLP), convolutional neural network-long short-term memory (CNN-LSTM), and wavelet packet decomposition (WPD) for CNN-LSTM are presented. In addition, novelty in the work includes the utilization of brain source localization (BSL) [using standardized low-resolution brain electromagnetic tomography (sLORETA)] for the reliable decoding of motor intention. The information is utilized for channel selection and accurate EEG time segment selection. The performance of the proposed models is compared with the traditionally utilized mLR technique on the reach, grasp, and lift (GAL) dataset. The effectiveness of the proposed framework is established using the Pearson correlation coefficient (PCC) and trajectory analysis. A significant improvement in the correlation coefficient is observed when compared with the state-of-the-art mLR model. Our work bridges the gap between the control and the actuator block, enabling real-time BCI implementation.

Mechanical Characterization and Standardization of Silicon Scalp and Dura Surrogates for Neurosurgical Simulation.

BACKGROUND: Simulation-based neurosurgical training allows the development of surgical skills outside the operating room. However, the use of nonstandardized materials and poor haptic feedback remain the primary limitations of the surgical simulators. Therefore, this work proposes a comprehensive scheme for scalp and dura surrogate synthesis and their standardization for neurosurgical training. METHODS: Eight different variants of silicone-based scalp (S1-S8) and dura (D1-D8) surrogates were synthesized. The samples were evaluated by 26 neurosurgeons. They provided their feedback in a Likert scale questionnaire. Kruskal-Wallis test with Dunn multiple comparisons was used for statistical analysis of surgeons' scores. The samples were mechanically characterized using Shore A hardness and dynamic nanoindentation testing. RESULTS: The highest mean Likert score values were obtained for S3 scalp and D8 dura variants. The comparison of S3 and D8 with the rest of the variants in the respective groups was statistically significant in 21 of 28 instances. The average Shore A hardness and storage modulus of the S3 variant were 21.9 DU and 505.3 kPa, respectively. The corresponding values for the D8 variant were 32.5 DU and 632 kPa, respectively. CONCLUSIONS: This study proposes a method for the synthesis, evaluation, and standardization of scalp and dura surrogates. The study achieved standardized silicone compositions along with a recommendable range of Shore hardness and viscoelastic moduli values for the scalp and dura surrogates. This work can be extended for the standardization of surrogates for other tissues involved in neurosurgical simulators.

An Integrated Dynamic Closed Loop Simulation Platform for Elbow Flexion Augmentation Using an Upper Limb Exosuit Model.

Wearable robotic devices are designed to assist, enhance or restore human muscle performance. Understanding how a wearable robotic device changes human biomechanics through complex interaction is important to guide its proper design, parametric optimization and functional success. The present work develops a human-machine-interaction simulation platform for closed loop dynamic analysis with feedback control and to study the effect of soft-robotic wearables on human physiology. The proposed simulation platform incorporates Computed Muscle Control (CMC) algorithm and is implemented using the MATLAB -OpenSim interface. The framework is generic and will allow incorporation of any advanced control strategy for the wearable devices. As a demonstration, a Gravity Compensation (GC) controller has been implemented on the wearable device and the resulting decrease in the joint moments, muscle activations and metabolic costs during a simple repetitive load lifting task with two different speeds is investigated.

A Poly-vinyl Alcohol (PVA)-based phantom and training tool for use in simulated Transrectal Ultrasound (TRUS) guided prostate needle biopsy procedures.

Trans-rectal ultrasound-guided needle biopsy is a well-established diagnosis technique for prostate cancer. To enhance the needle manoeuvring skills under ultrasound (US) guidance, it is preferable to train medical practitioners in needle biopsy on tissue-mimicking phantoms. This phantom should mimic the morphology as well as mechanical and acoustic properties of the human male pelvic region to provide a surgical experience and feedback. In this study, polyvinyl alcohol (PVA) was used and evaluated for prostate phantom development, that is stiffness tunable, US-compatible and durable phantom material. Three samples, each with 5%, 10%, and 15% concentration of PVA material, were prepared, and their mechanical and shrinkage characteristics were investigated. The anatomy of male pelvic region was used to develop an anatomically correct phantom. Later US-guided needle biopsy was performed on the phantom. The range of elastic moduli of the PVA samples was 2∼146 kPa. Their elastic moduli and volumes were found to remain statistically close from seventh to eighth freeze-thaw cycle (p>0.05). Initial US scans of the phantom resulted in satisfactory B-mode images, with a clear distinction between the prostate and its surrounding organs. This study demonstrated the applicability of PVA hydrogel as a phantom material for training in US-guided needle biopsy.

Comparing Migratory and Mechanical Properties of Human Bone Marrow-Derived Mesenchymal Stem Cells with Colon Cancer Cells In Vitro.

BACKGROUND: Colon cancer cells can migrate and metastasize by undergoing epithelial-to-mesenchymal transition (EMT). Mesenchymal stem cells (MSCs) are non-cancerous, multipotent adult stem cells, which can also migrate. In this study, we wanted to compare the biological, physical, and functional properties of these migratory cells. MATERIALS AND METHODS: HT-29 and HCT-116, two human colon carcinoma cell lines, represent less aggressive and more aggressive cancer cells, respectively. MSCs were isolated from human bone marrow. After confirming the identity of all the cell types, they were evaluated for E-cadherin, ß1-integrin, Vimentin, ZEB-1, ß-catenin, and 18S rRNA using Q-PCR. MMP-2 and MMP-9 activity were evaluated using gelatin zymography. Functional tests like wound healing assay, migration assay, and invasion assay were also done. Biomechanical properties like cell stiffness and non-specific adhesion (between indenter probe and cell membrane) were evaluated through nanoindentation using atomic force microscopy (AFM). RESULTS: Expression of EMT and stem cell markers showed typical expression patterns for HT-29, HCT-116, and MSCs. Functional tests showed that MSCs migrated faster than malignant cells. MMP-2 and MMP-9 activity reinforced this behavior. Interestingly, the migration/invasion capacity of MSCs was comparable to aggressive HCT-116, and more than HT-29. MSCs also showed the maximum cell stiffness and non-specific cell-probe adhesions, followed by HCT116 and HT29 cells. CONCLUSIONS: Our findings indicate that the migratory properties of MSCs is comparable or even greater than that of cancer cells and despite their high migration potential, they also have the maximum stiffness.

Biomechanical modelling and computer aided simulation of deep brain retraction in neurosurgery.

BACKGROUND AND OBJECTIVES: Surgical simulators are widely used to promote faster and safer surgical training. They not only provide a platform for enhancing surgical skills but also minimize risks to the patient's safety, operation theatre usage, and financial expenditure. Retracting the soft brain tissue is an unavoidable procedure during any surgery to access the lesioned tissue deep within the brain. Excessive retraction often results in damaging the brain tissue, thus requiring advanced skills and prior training using virtual platforms. Such surgical simulation platforms require an anatomically correct computational model that can accurately predict the brain deformation in real-time. METHODS: In this study, we present a 3D finite element brain model reconstructed from MRI dataset. The model incorporates precisely the anatomy and geometrical features of the canine brain. The brain model has been used to formulate and solve a quasi-static boundary value problem for brain deformation during brain retraction. The visco-hyperelastic framework within the theory of non-linear elasticity has been used to set up the boundary value problem. Consequently, the derived non-linear field equations have been solved using finite element solver ABAQUS. RESULTS: The retraction simulations have been performed for two scenarios: retraction pressure in the brain and forces required to perform the surgery. The brain was retracted by 5 mm and retained at that position for 30 minutes, during which the retraction pressure attenuates to 36% of its peak value. Both the model predictions as well as experimental observations on canine brain indicate that brain retraction up to 30 minutes did not cause any significant risk of induced damage. Also, the peak retraction pressure level indicates that intermittent retraction is a safer procedure as compared to the continuous retraction, for the same extent of retraction. CONCLUSIONS: The results of the present study indicate the potential of a visco-hyperelastic framework for simulating deep brain retraction effectively. The simulations were able to capture the dominant characteristics of brain tissue undergoing retraction. The developed platform could serve as a basis for the development of a detailed model in the future that can eventually be used for effective preoperative planning and training purposes.

Robust motion artefact resistant circuit for calculation of Mean Arterial Pressure from pulse transit time.

Cuff-less and non-invasive methods of Blood Pressure (BP) monitoring have faced a lot of challenges like stability, noise, motion artefact and requirement for calibration. These factors are the major reasons why such devices do not get approval from the medical community easily. One such method is calculating Blood Pressure indirectly from pulse transit time (PTT) obtained from electrocardiogram (ECG) and Photoplethysmogram (PPG). In this paper we have proposed two novel analog signal conditioning circuits for ECG and PPG that increase stability, remove motion artefacts, remove the sinusoidal wavering of the ECG baseline due to respiration and provide consistent digital pulses corresponding to blood pulses/heart-beat. We have combined these two systems to obtain the PTT and then correlated it with the Mean Arterial Pressure (MAP). The aim was to perform major part of the processing in analog domain to decrease processing load over microcontroller so as to reduce cost and make it simple and robust. We have found from our experiments that the proposed circuits can calculate the Heart Rate (HR) with a maximum error of ~3.0% and MAP with a maximum error of ~2.4% at rest and ~4.6% in motion.

Simultaneous Analysis of Elastic and Nonspecific Adhesive Properties of Thin Sample and Biological Cell Considering Bottom Substrate Effect.

A new asymptotically correct contact model has been developed for conical tip based atomic force microscopy (AFM) nanoindentation. This new model provides both elastic and nonspecific adhesion properties of cells and soft gels by taking sample thickness at the point of indentation and its depth of indentation into consideration. The bottom substrate effect (BSE) is the most common source of error in the study of "AFM force maps" of the cellular sample. The present model incorporates an asymptotically correct correction term as a function of depth of indentation to eliminate the substrate effect in the analysis. Later, the model is extended to analyze the unloading portion of the indentation curve to extract the stiffness and adhesive properties simultaneously. A comparative study of the estimated material properties using other established contact models shows that the provided corrections effectively curb the errors coming from infinite thickness assumption. Nonspecific adhesive nature of a cell is represented in terms of adhesion parameter (γa) based on the "work of adhesion," this is an alternative to the peak value of tip-sample attractive (negative) force commonly used as representative adhesion measurement. The simple analytical expression of the model can help in estimating more realistic and accurate biomechanical properties of cells from atomic force microscopy based indentation technique.

Computational study to investigate effect of tonometer geometry and patient-specific variability on radial artery tonometry.

Tonometry-based devices are valuable method for vascular function assessment and for measurement of blood pressure. However current design and calibration methods rely on simple models, neglecting key geometrical features, and anthropometric and property variability among patients. Understanding impact of these influences on tonometer measurement is thus essential for improving outcomes of current devices, and for proposing improved design. Towards this goal, we present a realistic computational model for tissue-device interaction using complete wrist section with hyperelastic material and frictional contact. Three different tonometry geometries were considered including a new design, and patient-specific influences incorporated via anthropometric and age-dependent tissue stiffness variations. The results indicated that the new design showed stable surface contact stress with minimum influence of the parameters analyzed. The computational predictions were validated with experimental data from a prototype based on the new design. Finally, we showed that the underlying mechanics of vascular unloading in tonometry to be fundamentally different from that of oscillatory method. Due to directional loading in tonometry, pulse amplitude maxima was observed to occur at a significantly lower compression level (around 31%) than previously reported, which can impact blood pressure calibration approaches based on maximum pulse pressure recordings.

Modeling electrically active viscoelastic membranes.

The membrane protein prestin is native to the cochlear outer hair cell that is crucial to the ear's amplification and frequency selectivity throughout the whole acoustic frequency range. The outer hair cell exhibits interrelated dimensional changes, force generation, and electric charge transfer. Cells transfected with prestin acquire unique active properties similar to those in the native cell that have also been useful in understanding the process. Here we propose a model describing the major electromechanical features of such active membranes. The model derived from thermodynamic principles is in the form of integral relationships between the history of voltage and membrane resultants as independent variables and the charge density and strains as dependent variables. The proposed model is applied to the analysis of an active force produced by the outer hair cell in response to a harmonic electric field. Our analysis reveals the mechanism of the outer hair cell active (isometric) force having an almost constant amplitude and phase up to 80 kHz. We found that the frequency-invariance of the force is a result of interplay between the electrical filtering associated with prestin and power law viscoelasticity of the surrounding membrane. Paradoxically, the membrane viscoelasticity boosts the force balancing the electrical filtering effect. We also consider various modes of electromechanical coupling in membrane with prestin associated with mechanical perturbations in the cell. We consider pressure or strains applied step-wise or at a constant rate and compute the time course of the resulting electric charge. The results obtained here are important for the analysis of electromechanical properties of membranes, cells, and biological materials as well as for a better understanding of the mechanism of hearing and the role of the protein prestin in this mechanism.

Stereocilia membrane deformation: implications for the gating spring and mechanotransduction channel.

In hair cells, although mechanotransduction channels have been localized to tips of shorter stereocilia of the mechanically sensitive hair bundle, little is known about how force is transmitted to the channel. Here, we use a biophysical model of the membrane-channel complex to analyze the nature of the gating spring compliance and channel arrangement. We use a triangulated surface model and Monte Carlo simulation to compute the deformation of the membrane under the action of tip link force. We show that depending on the gating spring stiffness, the compliant component of the gating spring arises from either the membrane alone or a combination of the membrane and a tether that connects the channel to the actin cytoskeleton. If a bundle is characterized by relatively soft gating springs, such as those of the bullfrog sacculus, the need for membrane reinforcement by channel tethering then depends on membrane parameters. With stiffer gating springs, such as those from rat outer hair cells, the channel must be tethered for all biophysically realistic parameters of the membrane. We compute the membrane forces (resultants), which depend on membrane tension, bending modulus, and curvature, and show that they can determine the fate of the channel.

Cell crawling assisted by contractile stress induced retraction.

Cell locomotion is a result of a series of synchronized chemo-mechanical processes. Crawling-type cell locomotion consists of three steps: protrusion, translocation, and retraction. Previous works have shown that both protrusion and retraction can produce cell movement. For the latter, a cell derives its propulsive force from retraction induced protrusion mechanism, which was experimentally verified by Chen (1979, "Induction of Spreading During Fibroblast Movement," J. Cell Biol., 81, pp. 684-691). In this paper, using finite element method, we take a computational biomimetic approach to study cell crawling assisted by contractile stress induced de-adhesion at the rear of the focal adhesion zone (FAZ). We assume the formation of the FAZ is driven by receptor-ligand bonds and nonspecific interactions. The contractile stress is generated due to the molecular activation of the intracellular actin-myosin machinery. The exerted contractile stress and its time dependency are modeled in a phenomenological manner as a two-spring mechanosensor proposed by Schwarz (2006, "Focal Adhesions as Mechanosensors: The Two-Spring Model," BioSystems, 83(2-3), pp. 225-232). Through coupling the kinetics of receptor-ligand bonds with contractile stress, de-adhesion can be achieved when the stall value of the contractile stress is larger than a critical one. De-adhesion at the rear end of the FAZ causes a redistribution of elastic energy and induces cell locomotion. Parametric studies were conducted to investigate the connection between the cell locomotion speed and stall stress, and receptor-ligand kinetics. Finally, we provide a scaling relationship that can be used to estimate the cell locomotion speed.

A computational biomimetic study of cell crawling.

Cell locomotion is a result of a series of synchronized chemo-mechanical processes. Previous extensive experimental studies have revealed many chemo-mechanical processes that may contribute to cell locomotion. In parallel, theoretical works have been developed to provide deeper insight. To date, however, direct simulations of cell locomotion on a substrate have not been seen. In this paper, a finite element-based computational model is developed to study amoeboid type of cell crawling phenomenon. Here, a cell is modeled as a 2D fluid-filled elastic vesicle, which establishes its interaction with a rigid substrate through a kinetics-based cellular adhesion model. The cell derives its motion through a differential bond breaking at the trailing edge and bond formation at the leading edge. This mechanism of crawling authenticates the hypothesis that cell locomotion can be facilitated by breaking the adhesive bonds at the rear edge, which was initially proposed by Chen (J Cell Biol 90: 187-200, 1981).

Micromechanical model for elasticity of the cell cytoskeleton.

Semiflexible polymer networks, such as cell cytoskeleton, differ significantly from their flexible counterparts in their deformation energy storage mechanism. As a result, the network elasticity is governed by both enthalpic and entropic variations. In addition, the enthalpic effect shows two distinct regimes of energy storage mechanism, the affine and nonaffine regimes. In the past, computation-based modeling on random networks, such as the Mikado model, was used to demonstrate the physical mechanism of mechanical deformation of semiflexible networks. These models are computationally intensive and hence are difficult to apply to studying whole cells. In this paper, we develop a micromechanical model to predict the average macroscopic elastic properties of a random, semiflexible, biopolymer network. The model employs a unit cell consisting of four semiflexible chains and four equivalent axial-bending springs. The proposed unit-cell-based micromechanical model represents a statistically average realization of the actual network and gives the average mechanical properties, such as the shear modulus. Comparisons between the model predictions and Mikado model results confirm that this micromechanical model captures the essential deformation physics revealed from previous studies on the actual network and is capable of predicting the transition between nonaffine and affine deformations. This model can be used to develop efficient continuum constitutive models of the cytoskeleton in the future.