TBS-BAO: fully automated beam angle optimization for IMRT guided by a total-beam-space reference plan

Properly selected beam angles contribute to the quality of radiotherapy treatment plans. However, the Beam Angle Optimization (BAO) problem is difficult to solve to optimality due to its non-convex discrete nature with many local minima. In this study, we propose TBS-BAO, a novel approach for solving the BAO problem, and test it for non-coplanar robotic CyberKnife radiotherapy for prostate cancer. First, an ideal Pareto-optimal reference dose distribution is automatically generated using a priori multi-criterial fluence map optimization (FMO) to generate a plan that includes all candidate beams (total-beam-space, TBS). Then, this ideal dose distribution is reproduced as closely as possible in a subsequent segmentation/beam angle optimization step (SEG/BAO), while limiting the number of allowed beams to a user-selectable preset value. SEG/BAO aims at a close reproduction of the ideal dose distribution. For each of 33 prostate SBRT patients, 18 treatment plans with different pre-set numbers of allowed beams were automatically generated with the proposed TBS-BAO. For each patient, the TBS-BAO plans were then compared to a plan that was automatically generated with an alternative BAO method (Erasmus-iCycle) and to a high-quality manually generated plan. TBS-BAO was able to automatically generate plans with clinically feasible numbers of beams (∽25), with a quality highly similar to corresponding 91-beam ideal reference plans. Compared to the alternative Erasmus-iCycle BAO approach, similar plan quality was obtained for 25-beam segmented plans, while computation times were reduced from 10.7 hours to 4.8/1.5 hours, depending on the applied pencil-beam resolution in TBS-BAO. 25-beam TBS-BAO plans had similar quality as manually generated plans with on average 48 beams, while delivery times reduced from 22.3 to 18.4/18.1 min. TBS reference plans could effectively steer the discrete non-convex BAO.

A New Discrete Implicit Monte Carlo Scheme for Simulating Radiative Transfer Problems

We present a new algorithm for radiative transfer—based on a statistical Monte Carlo approach—that does not suffer from teleportation effects, on the one hand, and yields smooth results, on the other hand. Implicit Monte Carlo (IMC) techniques for modeling radiative transfer have existed from the 1970s. When they are used for optically thick problems, however, the basic algorithm suffers from “teleportation” errors, where the photons propagate faster than the exact physical behavior, due to the absorption-blackbody emission processes. One possible solution is to use semianalog Monte Carlo, in its new implicit form (ISMC), which uses two kinds of particles, photons and discrete material particles. This algorithm yields excellent teleportation-free results, but it also produces noisier solutions (relative to classic IMC), due to its discrete nature. Here, we derive a new Monte Carlo algorithm, Discrete Implicit Monte Carlo (DIMC), which also uses the idea of two kinds of discrete particles, and thus does not suffer from teleportation errors. DIMC implements the IMC discretization and creates new radiation photons for each time step, unlike ISMC. Using the continuous absorption technique, DIMC yields smooth results like classic IMC. One of the main elements of the algorithm is the avoidance of the explosion of the particle population, by using particle merging. We test the new algorithm on 1D and 2D cylindrical problems, and show that it yields smooth, teleportation-free results. We finish by demonstrating the power of the new algorithm on a classic radiative hydrodynamic problem—an opaque radiative shock wave. This demonstrates the power of the new algorithm for astrophysical scenarios.

A Novel Occupancy Mapping Framework for Risk-Aware Path Planning in Unstructured Environments

In the context of autonomous robots, one of the most important tasks is to prevent potential damage to the robot during navigation. For this purpose, it is often assumed that one must deal with known probabilistic obstacles, then compute the probability of collision with each obstacle. However, in complex scenarios or unstructured environments, it might be difficult to detect such obstacles. In these cases, a metric map is used, where each position stores the information of occupancy. The most common type of metric map is the Bayesian occupancy map. However, this type of map is not well suited for computing risk assessments for continuous paths due to its discrete nature. Hence, we introduce a novel type of map called the Lambda Field, which is specially designed for risk assessment. We first propose a way to compute such a map and the expectation of a generic risk over a path. Then, we demonstrate the benefits of our generic formulation with a use case defining the risk as the expected collision force over a path. Using this risk definition and the Lambda Field, we show that our framework is capable of doing classical path planning while having a physical-based metric. Furthermore, the Lambda Field gives a natural way to deal with unstructured environments, such as tall grass. Where standard environment representations would always generate trajectories going around such obstacles, our framework allows the robot to go through the grass while being aware of the risk taken.

Nonconvex Sparse Regularization for Deep Neural Networks and Its Optimality

Recent theoretical studies proved that deep neural network (DNN) estimators obtained by minimizing empirical risk with a certain sparsity constraint can attain optimal convergence rates for regression and classification problems. However, the sparsity constraint requires knowing certain properties of the true model, which are not available in practice. Moreover, computation is difficult due to the discrete nature of the sparsity constraint. In this letter, we propose a novel penalized estimation method for sparse DNNs that resolves the problems existing in the sparsity constraint. We establish an oracle inequality for the excess risk of the proposed sparse-penalized DNN estimator and derive convergence rates for several learning tasks. In particular, we prove that the sparse-penalized estimator can adaptively attain minimax convergence rates for various nonparametric regression problems. For computation, we develop an efficient gradient-based optimization algorithm that guarantees the monotonic reduction of the objective function.

Generalised S-System-Type Equation: Sensitivity of the Deterministic and Stochastic Models for Bone Mechanotransduction

The formalism of a bone cell population model is generalised to be of the form of an S-System. This is a system of nonlinear coupled ordinary differential equations (ODEs), each with the same structure: the change in a variable is equal to a difference in the product of a power-law functions with a specific variable. The variables are the densities of a variety of biological populations involved in bone remodelling. They will be specified concretely in the cases of a specific periodically forced system to describe the osteocyte mechanotransduction activities. Previously, such models have only been deterministically simulated causing the populations to form a continuum. Thus, very little is known about how sensitive the model of mechanotransduction is to perturbations in parameters and noise. Here, we revisit this assumption using a Stochastic Simulation Algorithm (SSA), which allows us to directly simulate the discrete nature of the problem and encapsulate the noisy features of individual cell division and death. Critically, these stochastic features are able to cause unforeseen dynamics in the system, as well as completely change the viable parameter region, which produces biologically realistic results.

Reconstructing and counting genomic fragments through tagmentation-based haploid phasing

Single-cell sequencing provides a new level of granularity in studying the heterogeneous nature of cancer cells. For some cancers, this heterogeneity is the result of copy number changes of genes within the cellular genomes. The ability to accurately determine such copy number changes is critical in tracing and understanding tumorigenesis. Current single-cell genome sequencing methodologies infer copy numbers based on statistical approaches followed by rounding decimal numbers to integer values. Such methodologies are sample dependent, have varying calling sensitivities which heavily depend on the sample’s ploidy and are sensitive to noise in sequencing data. In this paper we have demonstrated the concept of integer-counting by using a novel bioinformatic algorithm built on our library construction chemistry in order to detect the discrete nature of the genome.

Cartesian Difference Categories

Cartesian differential categories are categories equipped with a differential combinator which axiomatizes the directional derivative. Important models of Cartesian differential categories include classical differential calculus of smooth functions and categorical models of the differential $\lambda$-calculus. However, Cartesian differential categories cannot account for other interesting notions of differentiation of a more discrete nature such as the calculus of finite differences. On the other hand, change action models have been shown to capture these examples as well as more "exotic" examples of differentiation. But change action models are very general and do not share the nice properties of Cartesian differential categories. In this paper, we introduce Cartesian difference categories as a bridge between Cartesian differential categories and change action models. We show that every Cartesian differential category is a Cartesian difference category, and how certain well-behaved change action models are Cartesian difference categories. In particular, Cartesian difference categories model both the differential calculus of smooth functions and the calculus of finite differences. Furthermore, every Cartesian difference category comes equipped with a tangent bundle monad whose Kleisli category is again a Cartesian difference category.

Fastness of Protective and Decorative Coatings Construction Products and Structures to the Action of Wind

The article provides information about the stress state protective and decorative coatings during aging. The influence of the discrete nature of the substrate on the change in the stress state of the coatings from the action of the wind load is considered. It was revealed that for each type of coating there is its own critical value of the pore size (unfilled paint), the excess of which leads to cracking of the coatings. A method is proposed for selecting the optimal coating thickness.

Towards Measuring Terahertz Photon Statistics by a Superconducting Bolometer

Statistical distributions of the analog readings of an antenna-coupled THz superconducting bolometer were measured and analyzed under a special type of irradiation by low-energy fluxes of THz photons with Poisson photon statistics and controllable mean photon numbers. The photons were generated via low-gain parametric down-conversion in pulse-pumped Mg:LiNbO3 crystal placed to a cooled cryostat together with the bolometer NbN film. Results of theoretical approximation of experimental histograms reveal the discrete nature of THz detection by superconducting bolometers and open a way for studying their quantum characteristics. It is shown that bolometer readings per pulse consist of discrete counts (“single charges”), with the mean number linearly dependent on the number of input photons. Contributions of single counts to a total analog reading are statistically distributed according to the normal law, with average values slightly depending on the number of counts in each reading. A general formula is proposed to describe the relationship between continuous statistical distribution of the bolometer readings and discrete quantum statistics of the incident photons.

Origins of 1/f-like tissue oxygenation fluctuations in the murine cortex

The concentration of oxygen in the brain spontaneously fluctuates, and the distribution of power in these fluctuations has a 1/f-like spectra, where the power present at low frequencies of the power spectrum is orders of magnitude higher than at higher frequencies. Though these oscillations have been interpreted as being driven by neural activity, the origin of these 1/f-like oscillations is not well understood. Here, to gain insight of the origin of the 1/f-like oxygen fluctuations, we investigated the dynamics of tissue oxygenation and neural activity in awake behaving mice. We found that oxygen signal recorded from the cortex of mice had 1/f-like spectra. However, band-limited power in the local field potential did not show corresponding 1/f-like fluctuations. When local neural activity was suppressed, the 1/f-like fluctuations in oxygen concentration persisted. Two-photon measurements of erythrocyte spacing fluctuations and mathematical modeling show that stochastic fluctuations in erythrocyte flow could underlie 1/f-like dynamics in oxygenation. These results suggest that the discrete nature of erythrocytes and their irregular flow, rather than fluctuations in neural activity, could drive 1/f-like fluctuations in tissue oxygenation.