arXiv subjects

Ari Hietanen

Publications and source records attributed to Ari Hietanen.

At least 19 recordsLinked to original sources

arXiv2025

Ultrafast Deep Learning-Based Scatter Estimation in Cone-Beam Computed Tomography

Purpose: Scatter artifacts drastically degrade the image quality of cone-beam computed tomography (CBCT) scans. Although deep learning-based methods show promise in estimating scatter from CBCT measurements, their deployment in mobile CBCT systems or edge devices is still limited due to the large memory footprint of the networks. This study addresses the issue by applying networks at varying resolutions and suggesting an optimal one, based on speed and accuracy. Methods: First, the reconstruction error in down-up sampling of CBCT scatter signal was examined at six resolutions by comparing four interpolation methods. Next, a recent state-of-the-art method was trained across five image resolutions and evaluated for the reductions in floating-point operations (FLOPs), inference times, and GPU memory requirements. Results: Reducing the input size and network parameters achieved a 78-fold reduction in FLOPs compared to the baseline method, while maintaining comarable performance in terms of mean-absolute-percentage-error (MAPE) and mean-square-error (MSE). Specifically, the MAPE decreased to 3.85% compared to 4.42%, and the MSE decreased to 1.34 \times 10^{-2} compared to 2.01 \times 10^{-2}. Inference time and GPU memory usage were reduced by factors of 16 and 12, respectively. Further experiments comparing scatter-corrected reconstructions on a large, simulated dataset and real CBCT scans from water and Sedentex CT phantoms clearly demonstrated the robustness of our method. Conclusion: This study highlights the underappreciated role of downsampling in deep learning-based scatter estimation. The substantial reduction in FLOPs and GPU memory requirements achieved by our method enables scatter correction in resource-constrained environments, such as mobile CBCT and edge devices.

eess.SP↗

arXiv2024

Utilizing U-Net Architectures with Auxiliary Information for Scatter Correction in CBCT Across Different Field-of-View Settings

Cone-beam computed tomography (CBCT) has become a vital imaging technique in various medical fields but scatter artifacts are a major limitation in CBCT scanning. This challenge is exacerbated by the use of large flat panel 2D detectors. The scatter-to-primary ratio increases significantly with the increase in the size of FOV being scanned. Several deep learning methods, particularly U-Net architectures, have shown promising capabilities in estimating the scatter directly from the CBCT projections. However, the influence of varying FOV sizes on these deep learning models remains unexplored. Having a single neural network for the scatter estimation of varying FOV projections can be of significant importance towards real clinical applications. This study aims to train and evaluate the performance of a U-Net network on a simulated dataset with varying FOV sizes. We further propose a new method (Aux-Net) by providing auxiliary information, such as FOV size, to the U-Net encoder. We validate our method on 30 different FOV sizes and compare it with the U-Net. Our study demonstrates that providing auxiliary information to the network enhances the generalization capability of the U-Net. Our findings suggest that this novel approach outperforms the baseline U-Net, offering a significant step towards practical application in real clinical settings where CBCT systems are employed to scan a wide range of FOVs.

eess.SP↗

arXiv2023

Deep learning based projection domain metal segmentation for metal artifact reduction in cone beam computed tomography

Metal artifact correction is a challenging problem in cone beam computed tomography (CBCT) scanning. Metal implants inserted into the anatomy cause severe artifacts in reconstructed images. Widely used inpainting-based metal artifact reduction (MAR) methods require segmentation of metal traces in the projections as a first step, which is a challenging task. One approach is to use a deep learning method to segment metals in the projections. However, the success of deep learning methods is limited by the availability of realistic training data. It is laborious and time consuming to get reliable ground truth annotations due to unclear implant boundaries and large numbers of projections. We propose to use X-ray simulations to generate synthetic metal segmentation training dataset from clinical CBCT scans. We compare the effect of simulations with different numbers of photons and also compare several training strategies to augment the available data. We compare our model's performance on real clinical scans with conventional region growing threshold-based MAR, moving metal artifact reduction method, and a recent deep learning method. We show that simulations with relatively small number of photons are suitable for the metal segmentation task and that training the deep learning model with full size and cropped projections together improves the robustness of the model. We show substantial improvement in the image quality affected by severe motion, voxel size under-sampling, and out-of-FOV metals. Our method can be easily integrated into the existing projection-based MAR pipeline to get improved image quality. This method can provide a novel paradigm to accurately segment metals in CBCT projections.

eess.IV↗

arXiv2023

Reproducibility analysis of automated deep learning based localisation of mandibular canals on a temporal CBCT dataset

Preoperative radiological identification of mandibular canals is essential for maxillofacial surgery. This study demonstrates the reproducibility of a deep learning system (DLS) by evaluating its localisation performance on 165 heterogeneous cone beam computed tomography (CBCT) scans from 72 patients in comparison to an experienced radiologist's annotations. We evaluated the performance of the DLS using the symmetric mean curve distance (SMCD), the average symmetric surface distance (ASSD), and the Dice similarity coefficient (DSC). The reproducibility of the SMCD was assessed using the within-subject coefficient of repeatability (RC). Three other experts rated the diagnostic validity twice using a 0-4 Likert scale. The reproducibility of the Likert scoring was assessed using the repeatability measure (RM). The RC of SMCD was 0.969 mm, the median (interquartile range) SMCD and ASSD were 0.643 (0.186) mm and 0.351 (0.135) mm, respectively, and the mean (standard deviation) DSC was 0.548 (0.138). The DLS performance was most affected by postoperative changes. The RM of the Likert scoring was 0.923 for the radiologist and 0.877 for the DLS. The mean (standard deviation) Likert score was 3.94 (0.27) for the radiologist and 3.84 (0.65) for the DLS. The DLS demonstrated proficient qualitative and quantitative reproducibility, temporal generalisability, and clinical validity.

physics.med-ph↗

arXiv2022

Comparison of Deep Learning Segmentation and Multigrader-annotated Mandibular Canals of Multicenter CBCT scans

Deep learning approach has been demonstrated to automatically segment the bilateral mandibular canals from CBCT scans, yet systematic studies of its clinical and technical validation are scarce. To validate the mandibular canal localization accuracy of a deep learning system (DLS) we trained it with 982 CBCT scans and evaluated using 150 scans of five scanners from clinical workflow patients of European and Southeast Asian Institutes, annotated by four radiologists. The interobserver variability was compared to the variability between the DLS and the radiologists. In addition, the generalization of DLS to CBCT scans from scanners not used in the training data was examined to evaluate the out-of-distribution generalization capability. The DLS had lower variability to the radiologists than the interobserver variability between them and it was able to generalize to three new devices. For the radiologists' consensus segmentation, used as gold standard, the DLS had a symmetric mean curve distance of 0.39 mm compared to those of the individual radiologists with 0.62 mm, 0.55 mm, 0.47 mm, and 0.42 mm. The DLS showed comparable or slightly better performance in the segmentation of the mandibular canal with the radiologists and generalization capability to new scanners.

cs.LG↗

arXiv2016

A simple method to optimize HMC performance

We present a practical strategy to optimize a set of Hybrid Monte Carlo parameters in simulations of QCD and QCD-like theories. We specialize to the case of mass-preconditioning, with multiple time-step Omelyan integrators. Starting from properties of the shadow Hamiltonian we show how the optimal setup for the integrator can be chosen once the forces and their variances are measured, assuming that those only depend on the mass-preconditioning parameter.

hep-lat↗

arXiv2016

$SU(2)$ Gauge Theory with Two Fundamental Flavours: Scalar and Pseudoscalar Spectrum

We investigate the scalar and pseudoscalar spectrum of the $SU(2)$ gauge theory with $N_f=2$ flavours of fermions in the fundamental representation using non perturbative lattice simulations. We provide first benchmark estimates of the mass of the lightest $0(0^{+})$ ($σ$), $0(0^{-})$ ($η'$) and $1(0^+)$ ($a_0$) states, including estimates of the relevant disconnected contributions. We find $m_{a_0}/F_{\rm{PS}}= 16.7(4.9)$, $m_σ/F_{\rm{PS}}=19.2(10.8)$ and $m_{η'}/F_{\rm{PS}} = 12.8(4.7)$. These values for the masses of light scalar states provide crucial information for composite extensions of the Standard Model from the unified Fundamental Composi te Higgs-Technicolor theory \cite{Cacciapaglia:2014uja} to models of composite dark matter.

hep-lat↗

arXiv2016

SU(2) Gauge Theory with Two Fundamental Flavours: a Minimal Template for Model Building

We investigate the continuum spectrum of the SU(2) gauge theory with $N_f=2$ flavours of fermions in the fundamental representation. This model provides a minimal template which is ideal for a wide class of Standard Model extensions featuring novel strong dynamics that range from composite (Goldstone) Higgs theories to several intriguing types of dark matter candidates, such as the SIMPs. We improve our previous lattice analysis [1] by adding more data at light quark masses, at two additional lattice spacings, by determining the lattice cutoff via a Wilson flow measure of the $w_0$ parameter, and by measuring the relevant renormalisation constants non-perturbatively in the RI'-MOM scheme. Our results for the lightest isovector states in the vector and axial channels, in units of the pseudoscalar decay constant, are $m_V/F_{\rm{PS}}\sim 13.1(2.2)$ and $m_A/F_{\rm{PS}}\sim 14.5(3.6)$ (combining statistical and systematic errors). In the context of the composite (Goldstone) Higgs models, our result for the spin-one resonances are $m_V > 3.2(5)$ TeV and $m_A > 3.6(9)$ TeV, which are above the current LHC constraints. In the context of dark matter models, for the SIMP case our results indicate the occurrence of a compressed spectrum at the required large dark pion mass, which implies the need to include the effects of spin-one resonances in phenomenological estimates.

hep-lat↗

arXiv2015

Template Composite Dark Matter : SU(2) gauge theory with 2 fundamental flavours

We present a non perturbative study of SU(2) gauge theory with two fundamental Dirac flavours. We discuss how the model can be used as a template for composite Dark Matter (DM). We estimate one particular interaction of the DM candidate with the Standard Model : the interaction through photon exchange computing the electric polarizability of the DM candidate. Finally, we briefly discuss the viability of the model given the present experimental constraints.

hep-lat↗

arXiv2015

Wilson Fermions with Four Fermion Interactions

We present a lattice study of a four fermion theory, known as Nambu Jona-Lasinio (NJL) theory, via Wilson fermions. Four fermion interactions naturally occur in several extensions of the Standard Model as a low energy parameterisation of a more fundamental theory. In models of dynamical electroweak symmetry breaking these operators, at an effective level, are used to endow the Standard Model fermions with masses. Furthermore these operators, when sufficiently strong, can drastically modify the fundamental composite dynamics by, for example, turning a strongly coupled infrared conformal theory into a (near) conformal one with desirable features for model building. As first step, we study spontaneous chiral symmetry breaking for the lattice version of the NJL model.

hep-lat↗

arXiv2015

Conformal symmetry vs. chiral symmetry breaking in the SU(3) sextet model

We present new results for the SU(3) "sextet model" with two flavors transforming according to the two-index symmetric representation of the gauge group. The simulations are performed using unimproved Wilson fermions. We measure the meson and baryon spectrum of the theory for multiple bare quark masses at two different lattice spacings. To address the pressing issue of whether the model is inside or below the conformal window, we compare the spectrum to the expectations for a theory with spontaneous chiral symmetry breaking and to those of an IR conformal theory. Regardless of the answer (conformal or chirally broken), the theory is a cornerstone in our understanding of near-conformal and composite dynamics, ranging from Technicolor models to unparticle physics. It is also interesting for the composite dynamics of vector-like singlets with respect to the Standard Model interactions.

hep-lat↗

arXiv2015

Approximating the Ising model on fractal lattices of dimension below two

We construct periodic approximations to the free energies of Ising models on fractal lattices of dimension smaller than two, in the case of zero external magnetic field, using a generalization of the combinatorial method of Feynman and Vodvickenko. Our procedure is applicable to any fractal obtained by the removal of sites of a periodic two dimensional lattice. As a first application, we compute estimates for the critical temperatures of many different Sierpinski carpets and we compare them to known Monte Carlo estimates. The results show that our method is capable of determining the critical temperature with, possibly, arbitrary accuracy and paves the way to determine $T_c$ for any fractal of dimension below two. Critical exponents are more difficult to determine since the free energy of any periodic approximation still has a logarithmic singularity at the critical point implying $α= 0$. We also compute the correlation length as a function of the temperature and extract the relative critical exponent. We find $ν=1$ for all periodic approximation, as expected from universality.

cond-mat.stat-mech↗

arXiv2014

Composite (Goldstone) Higgs Dynamics on the Lattice: Spectrum of SU(2) Gauge Theory with two Fundamental Fermions

We study the meson spectrum of the SU(2) gauge theory with two Wilson fermions in the fundamental representation. The theory unifies both Technicolor and composite Goldstone Boson Higgs models of electroweak symmetry breaking. We have calculated the masses of the lightest spin one vector and axial vector mesons. In addition, we have also obtained preliminary results for the mass of the lightest scalar (singlet) meson state. The simulations have been done with multiple masses and two different lattice spacings for chiral and continuum extrapolations. The spin one meson masses set lower limits for accelerator experiments, whereas the scalar meson will mix with a pGB of the theory and produce two scalar states. The lighter of the states is the 125 GeV Higgs boson, and the heavier would be a new yet unobserved scalar state.

hep-lat↗

arXiv2014

Composite Goldstone Dark Matter: Experimental Predictions from the Lattice

We study, via first principles lattice simulations, the nonperturbative dynamics of $SU(2)$ gauge theory with two fundamental Dirac flavors. The model can be used simultaneously as a template for composite Goldstone boson dark matter and for breaking the electroweak symmetry dynamically. We compute the form factor, allowing us to estimate the associated electromagnetic charge radius. Interestingly we observe that the form factor obeys vector meson dominance even for the two color theory. We finally compare the model predictions with dark matter direct detection experiments. We find that the composite Goldstone boson dark matter cross sections is constrained by the most stringent direct-detection experiments. Our results are a foundation for quantitative new composite dynamics relevant for model building, and are of interest to current experiments.

hep-ph↗

arXiv2014

Schrödinger functional boundary conditions and improvement for N>3

The standard method to calculate non-perturbatively the evolution of the running coupling of a SU(N) gauge theory is based on the Schrödinger functional (SF). In this paper we construct a family of boundary fields for general values of N which enter the standard definition of the SF coupling. We provide spatial boundary conditions for fermions in several representations which reduce the condition number of the squared Dirac operator. In addition, we calculate the O(a) improvement coefficients for N>3 needed to remove boundary cutoff effects from the gauge action. After this, residual cutoff effects on the step scaling function are shown to be very small even when considering non-fundamental representations. We also calculate the ratio of Lambda parameters between the MS-bar and SF schemes.

hep-lat↗

arXiv2014

Fundamental Composite Higgs Dynamics on the Lattice: SU(2) with Two Flavors

In reference [1] a unified description, both at the effective and fundamental Lagrangian level, of models of composite Higgs dynamics was proposed. In the unified framework the Higgs itself can emerge, depending on the way the electroweak symmetry is embedded, either as a pseudo-Goldstone boson or as a massive excitation of the condensate. The most minimal fundamental description consists of an SU(2) gauge theory with two Dirac fermions transforming according to the defining representation of the gauge group. We therefore provide first principle lattice results for the massive spectrum of this theory. We confirm the chiral symmetry breaking phenomenon and determine the lightest spin-one axial and vector masses. The knowledge of the energy scale at which new states will appear at the Large Hadron Collider is of the utmost relevance to guide experimental searches of new physics.

hep-lat↗

arXiv2013

Schrödinger functional boundary conditions and improvement of the SU($N$) pure gauge action for $N>3$

The leading method to study the running coupling constant of non-abelian gauge theories is based on the Schrödinger functional scheme. However, the boundary conditions and $\mathcal{O}(a)$ improvement have not been systematically generalized for theories with more than three colors. These theories have applications in BSM model building as well as in the large $N$ limit. We have studied the boundary conditions and improvement for the pure Yang-Mills theory within the Schrödinger functional scheme. We have determined for all values of $N$ the boundary fields which provide high signal/noise ratio. Additionally, we have calculated the improvement coefficient $c_t$ for the pure gauge to one loop order for SU($N$) gauge theories with $N=2,\ldots,8$ from which $N\geq 4$ are previously unknown.

hep-lat↗

arXiv2013

Predictions for LHC from SO(4) MWT

We investigate the vector, axial and pseudo scalar mass spectrum of an SO(4) - MWT gauge theory with fermions in the vector representation of SO(4). Here we present the preliminary lattice results for the masses of vector and axial vector meson using Wilson fermions. These spectra are crucial for the discovery and to guide the searches of composite dynamics at the LHC.

hep-lat↗