Fiber Centerline Research Articles

Abstract 4262: AI-driven analysis of collagen characteristics in the tumor immune microenvironment predicts immune checkpoint inhibitors treatment responsiveness in gastric cancer

Abstract Background: Recent studies indicate a correlation between increased stromal composition in tumors and resistance to immune checkpoint inhibitors (ICIs) across various cancer types. Specifically, the extracellular matrix (ECM) within the stromal compartment significantly influence the tumor microenvironment, affecting cancer cell proliferation and immune cell recruitment. However, identifying and quantifying the ECM at the collagen fiber level in whole slide images (WSIs) is challenging due to its inherent heterogeneity and complexity. In this work, we utilized statistical analysis and a deep learning (DL)-based approach to identify and assess collagen in the ECM, examining its characteristics in relation to the ICIs treatment response in gastric cancer. Methods: We analyzed 116 WSIs of hematoxylin and eosin (H&E) stained slides before ICIs treatment from three institutions. Responders were defined as those with a complete response or partial response (N=18); those with stable disease (SD) (N=34) or progressive disease (PD) (N=64) were considered non-responders (N=98). For non-responders, post-treatment H&E slides were available for 21 patients; SD (N=5) and PD (N=16). We applied a DL-based second harmonic generation (SHG) synthesis algorithm to tumor regions identified by a tumor detection model. We then calculated various collagen properties, including volume, fiber orientation and alignment, waviness, straightness, and fiber thickness, based on collagen fiber centerlines extracted from the generated SHG images. An attention-based multiple instance learning (ABMIL) model was trained to automatically predict the treatment response solely based on collagen features. The dataset of 116 pre-treatment slides was divided into training and validation subsets using stratified 3-fold cross-validation. The area under the curve (AUC) was used to assess the performance of the trained models. Results: Collagen fiber features were significantly associated with ICIs response. Notably, ICIs responders have a decrease in collagen waviness and fiber orientation variation (p-value<0.05, by Mann-Whitney U test). Among non-responders, the PD group showed a statistical decrease in waviness after treatment (p-value<0.05, by Mann-Whitney U test). The ABMIL model achieved a mean AUC score of 0.823, with a standard deviation of 0.023. Conclusion: This study provides a comprehensive analysis of collagen in the ECM, and its properties in relation to gastric cancer ICIs response within Tumor Immune Microenvironment. We also developed a model using collagen features and demonstrated high accuracy in predicting treatment response. Citation Format: Minji Kim, Ekaterina Redekop, Yoonho Choi, Sung Hak Lee, Jae-Ho Cheong, Yu Aoki, Kohei Shitara, Tae Hyun Hwang. AI-driven analysis of collagen characteristics in the tumor immune microenvironment predicts immune checkpoint inhibitors treatment responsiveness in gastric cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4262.

Collagen fiber centerline tracking in fibrotic tissue via deep neural networks with variational autoencoder-based synthetic training data generation.

The role of fibrillar collagen in the tissue microenvironment is critical in disease contexts ranging from cancers to chronic inflammations, as evidenced by many studies. Quantifying fibrillar collagen organization has become a powerful approach for characterizing the topology of collagen fibers and studying the role of collagen fibers in disease progression. We present a deep learning-based pipeline to quantify collagen fibers' topological properties in microscopy-based collagen images from pathological tissue samples. Our method leverages deep neural networks to extract collagen fiber centerlines and deep generative models to create synthetic training data, addressing the current shortage of large-scale annotations. As a part of this effort, we have created and annotated a collagen fiber centerline dataset, with the hope of facilitating further research in this field. Quantitative measurements such as fiber orientation, alignment, density, and length can be derived based on the centerline extraction results. Our pipeline comprises three stages. Initially, a variational autoencoder is trained to generate synthetic centerlines possessing controllable topological properties. Subsequently, a conditional generative adversarial network synthesizes realistic collagen fiber images from the synthetic centerlines, yielding a synthetic training set of image-centerline pairs. Finally, we train a collagen fiber centerline extraction network using both the original and synthetic data. Evaluation using collagen fiber images from pancreas, liver, and breast cancer samples collected via second-harmonic generation microscopy demonstrates our pipeline's superiority over several popular fiber centerline extraction tools. Incorporating synthetic data into training further enhances the network's generalizability. Our code is available at https://github.com/uw-loci/collagen-fiber-metrics.

Integrated computational framework for modeling chopped fiber composites at the mesoscale

A new microstructure reconstruction algorithm, fully integrated with a non-iterative meshing algorithm named CISAMR, is introduced for synthesizing finite element (FE) models of chopped fiber composite (CFC) microstructures. This automated computational framework enables generating densely-packed CFC microstructures with desired statistical descriptors such as the volume fraction, diameter distribution, and spatial arrangement of fibers. The reconstruction process involves two main phases, starting with a virtual packing algorithm relying on Non-Uniform Rational B-Splines (NURBS) representation of fiber centerlines to efficiently detect/eliminate overlapping fibers. An explicit dynamic FE compression simulation is then pursued to increase the fiber volume fraction by modeling them as rigid shells and taking into account contact forces between fibers during the simulation. The shape/location of fibers excessively close to domain boundaries or intersecting them at sharp angles are then slightly modified in the resulting microstructure to ensure the construction of a high-quality conforming mesh using parallel CISAMR. Several example problems are provided to show the ability of this modeling framework for synthesizing and simulating the linear elastic response of polymer matrix CFCs with embedded glass fibers. We also demonstrate the application of high-fidelity FE models generated using CISAMR for simulating the nonlinear failure response of such composites.

A single-layer based numerical method for the slender body boundary value problem

Fluid flows containing dilute or dense suspensions of thin fibers are widespread in biological and industrial processes. To describe the motion of a thin immersed fiber, or to describe the forces acting on it, it is convenient to work with one-dimensional fiber centerlines and force densities rather than two-dimensional surfaces and surface tractions. Slender body theories offer ways to model and simulate the motion of immersed fibers using only one-dimensional data. However, standard formulations can break down when the fiber surface comes close to intersecting itself or other fibers. In this paper we introduce a numerical method for a recently derived three-dimensional slender body boundary value problem that can be stated entirely in terms of a one-dimensional distribution of forces on the centerline. The method is based on a new completed single-layer potential formulation of fluid velocity which removes the nullspace associated with the unmodified single layer potential. We discretize the model and present numerical results demonstrating the good conditioning and improved performance of the method in the presence of near-intersections. To avoid the modeling and numerical choices involved with free ends, we consider closed fibers.

An integral model based on slender body theory, with applications to curved rigid fibers

We propose a novel integral model describing the motion of both flexible and rigid slender fibers in viscous flow and develop a numerical method for simulating dynamics of curved rigid fibers. The model is derived from nonlocal slender body theory (SBT), which approximates flow near the fiber using singular solutions of the Stokes equations integrated along the fiber centerline. In contrast to other models based on (singular) SBT, our model yields a smooth integral kernel which incorporates the (possibly varying) fiber radius naturally. The integral operator is provably negative definite in a nonphysical idealized geometry, as expected from the partial differential equation theory. This is numerically verified in physically relevant geometries. We discuss the convergence and stability of a numerical method for solving the integral equation. The accuracy of the model and method is verified against known models for ellipsoids. Finally, we develop an algorithm for computing dynamics of rigid fibers with complex geometries in the case where the fiber density is much greater than that of the fluid, for example, in turbulent gas-fiber suspensions.

Skeleton optimization of neuronal morphology based on three-dimensional shape restrictions

BackgroundNeurons are the basic structural unit of the brain, and their morphology is a key determinant of their classification. The morphology of a neuronal circuit is a fundamental component in neuron modeling. Recently, single-neuron morphologies of the whole brain have been used in many studies. The correctness and completeness of semimanually traced neuronal morphology are credible. However, there are some inaccuracies in semimanual tracing results. The distance between consecutive nodes marked by humans is very long, spanning multiple voxels. On the other hand, the nodes are marked around the centerline of the neuronal fiber, not on the centerline. Although these inaccuracies do not seriously affect the projection patterns that these studies focus on, they reduce the accuracy of the traced neuronal skeletons. These small inaccuracies will introduce deviations into subsequent studies that are based on neuronal morphology files.ResultsWe propose a neuronal digital skeleton optimization method to evaluate and make fine adjustments to a digital skeleton after neuron tracing. Provided that the neuronal fiber shape is smooth and continuous, we describe its physical properties according to two shape restrictions. One restriction is designed based on the grayscale image, and the other is designed based on geometry. These two restrictions are designed to finely adjust the digital skeleton points to the neuronal fiber centerline. With this method, we design the three-dimensional shape restriction workflow of neuronal skeleton adjustment computation. The performance of the proposed method has been quantitatively evaluated using synthetic and real neuronal image data. The results show that our method can reduce the difference between the traced neuronal skeleton and the centerline of the neuronal fiber. Furthermore, morphology metrics such as the neuronal fiber length and radius become more precise.ConclusionsThis method can improve the accuracy of a neuronal digital skeleton based on traced results. The greater the accuracy of the digital skeletons that are acquired, the more precise the neuronal morphologies that are analyzed will be.

Mechanical properties prediction of injection molded short/long carbon fiber reinforced polymer composites using micro X-ray computed tomography

This paper addresses the challenge of reconstructing nonuniformly orientated fiber-reinforced polymer composites (FRPs) with three-dimensional (3D) geometric complexity, especially for fibers with curvatures, and proposes a framework using micro X-ray computed tomography (μXCT) images to quantify the fiber characteristics in 3D space for elastic modulus prediction. The FRP microstructure is first obtained from the μXCT images. Then, the fiber centerlines are efficiently extracted with the proposed fiber reconstruction algorithm, i.e., iterative template matching, and the 3D coordinates of the fiber centerlines are adopted for quantitative characterization of the fiber morphology. Finally, Young's modulus is predicted using the Halpin-Tsai model and laminate analogy approach, and the fiber configuration averaging method with the consideration of the fiber morphology. The new framework is demonstrated on both injection-molded short and long carbon fiber-reinforced polymer composites, whose fiber morphology and predicted mechanical properties are validated through previous pyrolysis and quasi-static tensile tests, respectively.

Theoretical Justification and Error Analysis for Slender Body Theory

AbstractSlender body theory facilitates computational simulations of thin fibers immersed in a viscous fluid by approximating each fiber using only the geometry of the fiber centerline curve and the line force density along it. However, it has been unclear how well slender body theory actually approximates Stokes flow about a thin but truly three‐dimensional fiber, in part due to the fact that simply prescribing data along a 1D curve does not result in a well‐posed boundary value problem for the Stokes equations in ℝ3. Here, we introduce a PDE problem to which slender body theory (SBT) provides an approximation, thereby placing SBT on firm theoretical footing. The slender body PDE is a new type of boundary value problem for Stokes flow where partial Dirichlet and partial Neumann conditions are specified everywhere along the fiber surface. Given only a 1D force density along a closed fiber, we show that the flow field exterior to the thin fiber is uniquely determined by imposing a fiber integrity condition: the surface velocity field on the fiber must be constant along cross sections orthogonal to the fiber centerline. Furthermore, a careful estimation of the residual, together with stability estimates provided by the PDE well‐posedness framework, allows us to establish error estimates between the slender body approximation and the exact solution to the above problem. The error is bounded by an expression proportional to the fiber radius (up to logarithmic corrections) under mild regularity assumptions on the 1D force density and fiber centerline geometry. © 2019 Wiley Periodicals, Inc.

New algorithms for virtual reconstruction of heterogeneous microstructures

A new set of algorithms is introduced for the virtual reconstruction of heterogeneous material microstructures, in which morphologies of embedded particles/fibers are explicitly represented. As a pre-processing phase, a shape library containing morphologies of heterogeneities, parameterized in terms of Non-Uniform Rational B-Splines (NURBS), is extracted from digital data such as micro-computed tomography images. Two packing algorithms are then introduced to reconstruct an initial (raw) periodic microstructure: In the first approach, a set of hierarchical bounding boxes approximating particle shapes are employed to check for overlap. The second approach, which is specialized for fibrous microstructures, uses the NURBS representation of fiber centerlines during the packing process. An optimization phase is applied to build the final microstructural model, relying either on the Genetic Algorithm to selectively eliminate some of the inclusions or their sequential relocation within the raw microstructure. The objective functions of this optimization phase are designed to replicate the target statistical microstructural descriptors such as the volume fraction, size distribution, and spatial arrangement of inclusions. Several example problems are presented to show the application of these algorithms for synthesizing various heterogeneous microstructures, as well as their finite element modeling and simulation.

A Lightweight Tangible 3D Interface for Interactive Visualization of Thin Fiber Structures

We present a prop-based, tangible interface for 3D interactive visualization of thin fiber structures. These data are commonly found in current bioimaging datasets, for example second-harmonic generation microscopy of collagen fibers in tissue. Our approach uses commodity visualization technologies such as a depth sensing camera and low-cost 3D display. Unlike most current uses of these emerging technologies in the games and graphics communities, we employ the depth sensing camera to create a fish-tank stereoscopic virtual reality system at the scientist's desk that supports tracking of small-scale gestures with objects already found in the work space. We apply the new interface to the problem of interactive exploratory visualization of three-dimensional thin fiber data. A critical task for the visual analysis of these data is understanding patterns in fiber orientation throughout a volume.The interface enables a new, fluid style of data exploration and fiber orientation analysis by using props to provide needed passive-haptic feedback, making 3D interactions with these fiber structures more controlled. We also contribute a low-level algorithm for extracting fiber centerlines from volumetric imaging. The system was designed and evaluated with two biophotonic experts who currently use it in their lab. As compared to typical practice within their field, the new visualization system provides a more effective way to examine and understand the 3D bioimaging datasets they collect.

Onset of whipping in the melt blowing process

Melt blowing is a common process for the production of polymer fibers, but its effectiveness can be greatly influenced by the occurrence of fiber whipping. This paper reports the results of a systematic parametric study aimed at elucidating the factors that determine the onset of whipping in the melt blowing process. A quasi-one-dimensional model developed by Yarin and co-workers [J. Appl. Phys. 108 (2010) 034913] is analyzed in detail and extended to consider the influence of different constitutive models. It is found that whipping requires sufficiently large levels of inertia in the fiber and air, as well as a strong inlet perturbation. Strong longitudinal stresses along the jet axis are required too, although elasticity is not necessary. Also essential are a drag force that varies along the fiber length, and an air velocity that decreases in magnitude away from the fiber centerline. Comparison of results from several different constitutive models suggests that melt inertia rather than melt rheology is the more dominant factor in controlling fiber shapes. The results of the present work are expected to be helpful for the development of more quantitatively accurate models of the melt blowing process.

Estimation of Local Orientations in Fibrous Structures With Applications to the Purkinje System

The extraction of the cardiac Purkinje system (PS) from intensity images is a critical step toward the development of realistic structural models of the heart. Such models are important for uncovering the mechanisms of cardiac disease and improving its treatment and prevention. Unfortunately, the manual extraction of the PS is a challenging and error-prone task due to the presence of image noise and numerous fiber junctions. To deal with these challenges, we propose a framework that estimates local fiber orientations with high accuracy and reconstructs the fibers via tracking. Our key contribution is the development of a descriptor for estimating the orientation distribution function (ODF), a spherical function encoding the local geometry of the fibers at a point of interest. The fiber/branch orientations are identified as the modes of the ODFs via spherical clustering and guide the extraction of the fiber centerlines. Experiments on synthetic data evaluate the sensitivity of our approach to image noise, width of the fiber, and choice of the mode detection strategy, and show its superior performance compared to those of the existing descriptors. Experiments on the free-running PS in an MR image also demonstrate the accuracy of our method in reconstructing such sparse fibrous structures.

NUMERICAL EVIDENCE FOR THE NON-EXISTENCE OF STATIONARY SOLUTIONS OF THE EQUATIONS DESCRIBING ROTATIONAL FIBER SPINNING

The stationary, isothermal rotational spinning process of fibers is considered. The investigations are concerned with the case of large Reynolds (δ = 3/ Re ≪ 1) and small Rossby numbers (ε ≪ 1). Modelling the fibers as a Newtonian fluid and applying slender body approximations, the process is described by a two-point boundary value problem of ODEs. The involved quantities are the coordinates of the fiber's centerline, the fluid velocity and viscous stress. The inviscid case δ = 0 is discussed as a reference case. For the viscous case δ > 0 numerical simulations are carried out. Transfering some properties of the inviscid limit to the viscous case, analytical bounds for the initial viscous stress of the fiber are obtained. A good agreement with the numerical results is found. These bounds give strong evidence, that for δ > 3ε2 no physical relevant stationary solution can exist.

Systematic derivation of an asymptotic model for the dynamics of curved viscous fibers

AbstractThis paper presents a slender body theory for the dynamics of a curved inertial viscous Newtonian fiber. Neglecting surface tension and temperature dependence, the fiber flow is modeled as a three‐dimensional free boundary value problem in terms of instationary incompressible Navier–Stokes equations. From regular asymptotic expansions in powers of the slenderness parameter, leading‐order balance laws for mass (cross‐section) and momentum are derived that combine the unrestricted motion of the fiber centerline with the inner viscous transport. The physically reasonable form of the one‐dimensional fiber model results thereby from the introduction of the intrinsic velocity that characterizes the convective terms. For the numerical investigation of the viscous, gravitational and rotational effects on the fiber dynamics, a finite volume approach on a staggered grid with implicit upwind flux discretization is applied. Copyright © 2007 John Wiley & Sons, Ltd.

An isothermal model for high-speed spinning of liquid crystalline polymer fibers – coupling of flow, orientation, and crystallization

Experiments show high-speed spinning of many synthetic fibers is accompanied by the partial crystallization of the initially amorphous melt. This crystallization is induced by some combination of the extensional flow, molecular orientation and extension, and temperature dependence. The crystallization of the material couples back to affect the fiber rheology, through a process of stress-hardening. Recently Forest, Wang and Bechtel have derived isothermal 1-D equations from macroscopic approximations of Doi-type liquid crystalline polymer (LCP) kinetic equations; their equations model the coupling between the extensional flow and the molecular orientation. Here we extend that model to include a fully three-way coupling of flow, orientation and crystallization mechanics, adopting an Avrami crystallization law with an orientation-dependent rate constant as proposed by Ziabicki, and the Kikutani stress-hardening law. We show that there is a significant difference in the fiber diameter profile, orientation and crystallinity with the inclusion of the stress-hardening to the extensional flow. Of interest here also is the role of crystallization in localizing the drawdown into a so-called `neck' region. We illustrate important qualitative and quantitative effects due to the development of crystallization: a pronounced drop in the fiber radius, an increased rapid fluid acceleration, and enhanced molecular alignment with the fiber centerline axis. These three features all occur over the short orientation-induced crystallization lengthscale, and are isolated in the first 20% of the crystallizing region along the spinline for draw ratios typical of high-speed spinning.