-
Development of an Unpaired Deep Neural Network for Synthesizing X-ray Fluoroscopic Images from Digitally Reconstructed Tomography in Image Guided Radiotherapy
Authors:
Chisako Hayashi,
Shinichiro Mori,
Yasukuni Mori,
Lim Taehyeung,
Hiroki Suyari,
Hitoshi Ishikawa
Abstract:
Purpose The purpose of this study was to develop and evaluate a deep neural network (DNN) capable of generating flat-panel detector (FPD) images from digitally reconstructed radiography (DRR) images in lung cancer treatment, with the aim of improving clinical workflows in image-guided radiotherapy.
Methods A modified CycleGAN architecture was trained on paired DRR-FPD image data obtained from pa…
▽ More
Purpose The purpose of this study was to develop and evaluate a deep neural network (DNN) capable of generating flat-panel detector (FPD) images from digitally reconstructed radiography (DRR) images in lung cancer treatment, with the aim of improving clinical workflows in image-guided radiotherapy.
Methods A modified CycleGAN architecture was trained on paired DRR-FPD image data obtained from patients with lung tumors. The training dataset consisted of over 400 DRR-FPD image pairs, and the final model was evaluated on an independent set of 100 FPD images. Mean absolute error (MAE), peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and Kernel Inception Distance (KID) were used to quantify the similarity between synthetic and ground-truth FPD images. Computation time for generating synthetic images was also measured.
Results Despite some positional mismatches in the DRR-FPD pairs, the synthetic FPD images closely resembled the ground-truth FPD images. The proposed DNN achieved notable improvements over both input DRR images and a U-Net-based method in terms of MAE, PSNR, SSIM, and KID. The average image generation time was on the order of milliseconds per image, indicating its potential for real-time application. Qualitative evaluations showed that the DNN successfully reproduced image noise patterns akin to real FPD images, reducing the need for manual noise adjustments.
Conclusions The proposed DNN effectively converted DRR images into realistic FPD images for thoracic cases, offering a fast and practical method that could streamline patient setup verification and enhance overall clinical workflow. Future work should validate the model across different imaging systems and address remaining challenges in marker visualization, thereby fostering broader clinical adoption.
△ Less
Submitted 1 March, 2025;
originally announced March 2025.
-
Spatiotemporal forecasting of vertical track alignment with exogenous factors
Authors:
Katsuya Kosukegawa,
Yasukuni Mori,
Hiroki Suyari,
Kazuhiko Kawamoto
Abstract:
To ensure the safety of railroad operations, it is important to monitor and forecast track geometry irregularities. A higher safety requires forecasting with higher spatiotemporal frequencies, which in turn requires capturing spatial correlations. Additionally, track geometry irregularities are influenced by multiple exogenous factors. In this study, a method is proposed to forecast one type of tr…
▽ More
To ensure the safety of railroad operations, it is important to monitor and forecast track geometry irregularities. A higher safety requires forecasting with higher spatiotemporal frequencies, which in turn requires capturing spatial correlations. Additionally, track geometry irregularities are influenced by multiple exogenous factors. In this study, a method is proposed to forecast one type of track geometry irregularity, vertical alignment, by incorporating spatial and exogenous factor calculations. The proposed method embeds exogenous factors and captures spatiotemporal correlations using a convolutional long short-term memory. The proposed method is also experimentally compared with other methods in terms of the forecasting performance. Additionally, an ablation study on exogenous factors is conducted to examine their individual contributions to the forecasting performance. The results reveal that spatial calculations and maintenance record data improve the forecasting of vertical alignment.
△ Less
Submitted 23 February, 2023; v1 submitted 7 November, 2022;
originally announced November 2022.
-
Advantages of $q$-logarithm representation over $q$-exponential representation from the sense of scale and shift on nonlinear systems
Authors:
Hiroki Suyari,
Hiroshi Matsuzoe,
Antonio M. Scarfone
Abstract:
Addition and subtraction of observed values can be computed under the obvious and implicit assumption that the scale unit of measurement should be the same for all arguments, which is valid even for any nonlinear systems. This paper starts with the distinction between exponential and non-exponential family in the sense of the scale unit of measurement. In the simplest nonlinear model…
▽ More
Addition and subtraction of observed values can be computed under the obvious and implicit assumption that the scale unit of measurement should be the same for all arguments, which is valid even for any nonlinear systems. This paper starts with the distinction between exponential and non-exponential family in the sense of the scale unit of measurement. In the simplest nonlinear model ${dy}/{dx}=y^{q}$, it is shown how typical effects such as rescaling and shift emerge in the nonlinear systems and affect observed data. Based on the present results, the two representations, namely the $q$-exponential and the $q$-logarithm ones, are proposed. The former is for rescaling, the latter for unified understanding with a fixed scale unit. As applications of these representations, the corresponding entropy and the general probability expression for unified understanding with a fixed scale unit are presented. For the theoretical study of nonlinear systems, $q$-logarithm representation is shown to have significant advantages over $q$-exponential representation.
△ Less
Submitted 13 December, 2019; v1 submitted 16 September, 2019;
originally announced September 2019.
-
$α$-divergence derived as the generalized rate function in a power-law system
Authors:
Hiroki Suyari,
Antonio Maria Scarfone
Abstract:
The generalized binomial distribution in Tsallis statistics (power-law system) is explicitly formulated from the precise $q$-Stirling's formula. The $α$-divergence (or $q$-divergence) is uniquely derived from the generalized binomial distribution in the sense that when $α\rightarrow-1$ (i.e., $q\rightarrow1$) it recovers KL divergence obtained from the standard binomial distribution. Based on thes…
▽ More
The generalized binomial distribution in Tsallis statistics (power-law system) is explicitly formulated from the precise $q$-Stirling's formula. The $α$-divergence (or $q$-divergence) is uniquely derived from the generalized binomial distribution in the sense that when $α\rightarrow-1$ (i.e., $q\rightarrow1$) it recovers KL divergence obtained from the standard binomial distribution. Based on these combinatorial considerations, it is shown that $α$-divergence (or $q$-divergence) is appeared as the generalized rate function in the large deviation estimate in Tsallis statistics.
△ Less
Submitted 11 May, 2014;
originally announced May 2014.
-
Large deviation estimates involving deformed exponential functions
Authors:
Jan Naudts,
Hiroki Suyari
Abstract:
We study large deviation properties of probability distributions with either a compact support or a fat tail by comparing them with q-deformed exponential distributions. Our main result is a large deviation property for probability distributions with a fat tail.
We study large deviation properties of probability distributions with either a compact support or a fat tail by comparing them with q-deformed exponential distributions. Our main result is a large deviation property for probability distributions with a fat tail.
△ Less
Submitted 30 May, 2015; v1 submitted 13 April, 2014;
originally announced April 2014.
-
Multiplicative duality, q-triplet and (mu,nu,q)-relation derived from the one-to-one correspondence between the (mu,nu)-multinomial coefficient and Tsallis entropy Sq
Authors:
Hiroki Suyari,
Tatsuaki Wada
Abstract:
We derive the multiplicative duality "q<->1/q" and other typical mathematical structures as the special cases of the (mu,nu,q)-relation behind Tsallis statistics by means of the (mu,nu)-multinomial coefficient. Recently the additive duality "q<->2-q" in Tsallis statistics is derived in the form of the one-to-one correspondence between the q-multinomial coefficient and Tsallis entropy. A slight g…
▽ More
We derive the multiplicative duality "q<->1/q" and other typical mathematical structures as the special cases of the (mu,nu,q)-relation behind Tsallis statistics by means of the (mu,nu)-multinomial coefficient. Recently the additive duality "q<->2-q" in Tsallis statistics is derived in the form of the one-to-one correspondence between the q-multinomial coefficient and Tsallis entropy. A slight generalization of this correspondence for the multiplicative duality requires the (mu,nu)-multinomial coefficient as a generalization of the q-multinomial coefficient. This combinatorial formalism provides us with the one-to-one correspondence between the (mu,nu)-multinomial coefficient and Tsallis entropy Sq, which determines a concrete relation among three parameters mu, nu and q, i.e., nu(1-mu)+1=q which is called "(mu,nu,q)-relation" in this paper. As special cases of the (mu,nu,q)-relation, the additive duality and the multiplicative duality are recovered when nu=1 and nu=q, respectively. As other special cases, when nu=2-q, a set of three parameters (mu,nu,q) is identified with the q-triplet (q_{sen},q_{rel},q_{stat}) recently conjectured by Tsallis. Moreover, when nu=1/q, the relation 1/(1-q_{sen})=1/alpha_{min}-1/alpha_{max} in the multifractal singularity spectrum f(alpha) is recovered by means of the (mu,nu,q)-relation.
△ Less
Submitted 5 February, 2007;
originally announced February 2007.
-
A two-parameter generalization of Shannon-Khinchin Axioms and the uniqueness theorem
Authors:
Tatsuaki Wada,
Hiroki Suyari
Abstract:
Based on the one-parameter generalization of Shannon-Khinchin (SK) axioms presented by one of the authors, and utilizing a tree-graphical representation, we have further developed the SK Axioms in accordance with the two-parameter entropy introduced by Sharma-Taneja, Mittal, Borges-Roditi, and Kaniadakis-Lissia-Scarfone. The corresponding unique theorem is proved. It is shown that the obtained t…
▽ More
Based on the one-parameter generalization of Shannon-Khinchin (SK) axioms presented by one of the authors, and utilizing a tree-graphical representation, we have further developed the SK Axioms in accordance with the two-parameter entropy introduced by Sharma-Taneja, Mittal, Borges-Roditi, and Kaniadakis-Lissia-Scarfone. The corresponding unique theorem is proved. It is shown that the obtained two-parameter Shannon additivity is a natural consequence from the Leibniz rule of the two-parameter Chakrabarti-Jagannathan difference operator.
△ Less
Submitted 5 August, 2006;
originally announced August 2006.
-
Scaling property and the generalized entropy uniquely determined by a fundamental nonlinear differential equation
Authors:
Hiroki Suyari,
Tatsuaki Wada
Abstract:
We derive a scaling property from a fundamental nonlinear differential equation whose solution is the so-called q-exponential function. A scaling property has been believed to be given by a power function only, but actually more general expression for the scaling property is found to be a solution of the above fundamental nonlinear differential equation. In fact, any power function is obtained b…
▽ More
We derive a scaling property from a fundamental nonlinear differential equation whose solution is the so-called q-exponential function. A scaling property has been believed to be given by a power function only, but actually more general expression for the scaling property is found to be a solution of the above fundamental nonlinear differential equation. In fact, any power function is obtained by restricting the domain of the q-exponential function appropriately. As similarly as the correspondence between the exponential function and Shannon entropy, an appropriate generalization of Shannon entropy is expected for the scaling property. Although the q-exponential function is often appeared in the optimal distributions of some one-parameter generalized entropies such as Renyi entropy, only Tsallis entropy is uniquely derived from the algebra of the q-exponential function, whose uniqueness is shown in the two ways in this paper.
△ Less
Submitted 7 August, 2006; v1 submitted 31 July, 2006;
originally announced August 2006.
-
$κ$-generalization of Stirling approximation and multinominal coefficients
Authors:
T. Wada,
H. Suyari
Abstract:
Stirling approximation of the factorials and multinominal coefficients are generalized based on the one-parameter ($κ$) deformed functions introduced by Kaniadakis [Phys. Rev. E \textbf{66} (2002) 056125]. We have obtained the relation between the $κ$-generalized multinominal coefficients and the $κ$-entropy by introducing a new $κ$-product operation.
Stirling approximation of the factorials and multinominal coefficients are generalized based on the one-parameter ($κ$) deformed functions introduced by Kaniadakis [Phys. Rev. E \textbf{66} (2002) 056125]. We have obtained the relation between the $κ$-generalized multinominal coefficients and the $κ$-entropy by introducing a new $κ$-product operation.
△ Less
Submitted 1 October, 2005;
originally announced October 2005.
-
$κ$-generalization of Gauss' law of error
Authors:
T. Wada,
H. Suyari
Abstract:
Based on the $κ$-deformed functions ($κ$-exponential and $κ$-logarithm) and associated multiplication operation ($κ$-product) introduced by Kaniadakis (Phys. Rev. E \textbf{66} (2002) 056125), we present another one-parameter generalization of Gauss' law of error. The likelihood function in Gauss' law of error is generalized by means of the $κ$-product. This $κ$-generalized maximum likelihood pr…
▽ More
Based on the $κ$-deformed functions ($κ$-exponential and $κ$-logarithm) and associated multiplication operation ($κ$-product) introduced by Kaniadakis (Phys. Rev. E \textbf{66} (2002) 056125), we present another one-parameter generalization of Gauss' law of error. The likelihood function in Gauss' law of error is generalized by means of the $κ$-product. This $κ$-generalized maximum likelihood principle leads to the {\it so-called} $κ$-Gaussian distributions.
△ Less
Submitted 12 May, 2005;
originally announced May 2005.
-
The unique non self-referential q-canonical distribution and the physical temperature derived from the maximum entropy principle in Tsallis statistics
Authors:
Hiroki Suyari
Abstract:
The maximum entropy principle in Tsallis statistics is reformulated in the mathematical framework of the q-product, which results in the unique non self-referential q-canonical distribution. As one of the applications of the present formalism, we theoretically derive the physical temperature which coincides with that already obtained in accordance with the generalized zeroth law of thermodynamic…
▽ More
The maximum entropy principle in Tsallis statistics is reformulated in the mathematical framework of the q-product, which results in the unique non self-referential q-canonical distribution. As one of the applications of the present formalism, we theoretically derive the physical temperature which coincides with that already obtained in accordance with the generalized zeroth law of thermodynamics.
△ Less
Submitted 22 March, 2005; v1 submitted 13 February, 2005;
originally announced February 2005.
-
Mathematical structure derived from the q-multinomial coefficient in Tsallis statistics
Authors:
Hiroki Suyari
Abstract:
We present the conclusive mathematical structure behind Tsallis statistics. We obtain mainly the following five theoretical results: (i) the one-to-one correspondence between the q-multinomial coefficient and Tsallis entropy, (ii) symmetry behind Tsallis statistics, (iii) the numerical computations revealing the existence of the central limit theorem in Tsallis statistics, (iv) Pascal's triangle…
▽ More
We present the conclusive mathematical structure behind Tsallis statistics. We obtain mainly the following five theoretical results: (i) the one-to-one correspondence between the q-multinomial coefficient and Tsallis entropy, (ii) symmetry behind Tsallis statistics, (iii) the numerical computations revealing the existence of the central limit theorem in Tsallis statistics, (iv) Pascal's triangle in Tsallis statistics and its properties, (v) the self-similarity of the q-product $\otimes_{q}$ leading to successful applications in Tsallis statistics. In particular, the third result (iii) provides us with a mathematical representation of a convincible answer to the physical problem: "Why so many power-law behaviors exist in nature universally ?"
△ Less
Submitted 27 January, 2004;
originally announced January 2004.
-
q-Stirling's formula in Tsallis statistics
Authors:
Hiroki Suyari
Abstract:
We present the q-Stirling's formula using the q-product determined by Tsallis entropy as nonextensive generalization of the usual Stirling's formula. The numerical computations and the proof are also shown. Moreover, we derive and prove some inequalities and equalities concerning the q-Stirling's formula. Applying the q-Stirling's formula, the mathematical foundations of Tsallis statistics can b…
▽ More
We present the q-Stirling's formula using the q-product determined by Tsallis entropy as nonextensive generalization of the usual Stirling's formula. The numerical computations and the proof are also shown. Moreover, we derive and prove some inequalities and equalities concerning the q-Stirling's formula. Applying the q-Stirling's formula, the mathematical foundations of Tsallis statistics can be obtained as similarly as those of Boltzmann-Gibbs statistics.
△ Less
Submitted 27 January, 2004;
originally announced January 2004.
-
Law of Error in Tsallis Statistics
Authors:
Hiroki Suyari,
Makoto Tsukada
Abstract:
Gauss' law of error is generalized in Tsallis statistics such as multifractal systems, in which Tsallis entropy plays an essential role instead of Shannon entropy. For the generalization, we apply the new multiplication operation determined by the q-logarithm and the q-exponential functions to the definition of the likelihood function in Gauss' law of error. The maximum likelihood principle lead…
▽ More
Gauss' law of error is generalized in Tsallis statistics such as multifractal systems, in which Tsallis entropy plays an essential role instead of Shannon entropy. For the generalization, we apply the new multiplication operation determined by the q-logarithm and the q-exponential functions to the definition of the likelihood function in Gauss' law of error. The maximum likelihood principle leads us to finding Tsallis distribution as nonextensively generalization of Gaussian distribution.
△ Less
Submitted 4 February, 2004; v1 submitted 27 January, 2004;
originally announced January 2004.
-
Generalization of Shannon-Khinchin Axioms to Nonextensive Systems and the Uniqueness Theorem
Authors:
Hiroki Suyari
Abstract:
The Shannon-Khinchin axioms are generalized to nonextensive systems and the uniqueness theorem for the nonextensive entropy is proved rigorously. In the present axioms, Shannon additivity is used as additivity in contrast to pseudoadditivity in Abe's axioms. The results reveal that Tsallis entropy is the simplest among all nonextensive entropies which can be obtained from the generalized Shannon…
▽ More
The Shannon-Khinchin axioms are generalized to nonextensive systems and the uniqueness theorem for the nonextensive entropy is proved rigorously. In the present axioms, Shannon additivity is used as additivity in contrast to pseudoadditivity in Abe's axioms. The results reveal that Tsallis entropy is the simplest among all nonextensive entropies which can be obtained from the generalized Shannon-Khinchin axioms.
△ Less
Submitted 28 May, 2002; v1 submitted 2 May, 2002;
originally announced May 2002.
-
Three classes of nonextensive entropies characterized by Shannon additivity and pseudoadditivity
Authors:
Hiroki Suyari
Abstract:
Nonextensive entropies are divided into three classes, each of which is characterized by Shannon additivity and pseudoadditivity. One of the three classes has properties of both additivities. The remaining classes have only one property of the two additivities, respectively. An example of nonextensive entropy is shown concretely for each class. In particular, one class is found to consist of onl…
▽ More
Nonextensive entropies are divided into three classes, each of which is characterized by Shannon additivity and pseudoadditivity. One of the three classes has properties of both additivities. The remaining classes have only one property of the two additivities, respectively. An example of nonextensive entropy is shown concretely for each class. In particular, one class is found to consist of only the Tsallis entropy. More precisely, the Tsallis entropy is proved to be uniquely determined by only these two additivities. The present classification using these two distinct additivities reveals unique characteristics of the Tsallis entropy.
△ Less
Submitted 28 May, 2002; v1 submitted 1 May, 2002;
originally announced May 2002.
-
Nonextensive Entropies derived from Form Invariance of Pseudoadditivity
Authors:
Hiroki Suyari
Abstract:
The form invariance of pseudoadditivity is shown to determine the structure of nonextensive entropies. Nonextensive entropy is defined as the appropriate expectation value of nonextensive information content, similar to the definition of Shannon entropy. Information content in a nonextensive system is obtained uniquely from generalized axioms by replacing the usual additivity with pseudoadditivi…
▽ More
The form invariance of pseudoadditivity is shown to determine the structure of nonextensive entropies. Nonextensive entropy is defined as the appropriate expectation value of nonextensive information content, similar to the definition of Shannon entropy. Information content in a nonextensive system is obtained uniquely from generalized axioms by replacing the usual additivity with pseudoadditivity. The satisfaction of the form invariance of the pseudoadditivity of nonextensive entropy and its information content is found to require the normalization of nonextensive entropies. The proposed principle requires the same normalization as that derived in [A.K. Rajagopal and S. Abe, Phys. Rev. Lett. {\bf 83}, 1711 (1999)], but is simpler and establishes a basis for the systematic definition of various entropies in nonextensive systems.
△ Less
Submitted 1 May, 2002; v1 submitted 17 August, 2001;
originally announced August 2001.
-
Quantum mutual entropy for Jaynes-Cummings model
Authors:
Shigeru Furuichi,
Masanori Ohya,
Hiroki Suyari
Abstract:
The dynamics of an atom on the Jaynes-Cummings model has been studied by an atomic inversion, von Neumann entropy and so on. In this letter, we will treat the Jaynes-Cummings model as a problem in non-equilibrium statistical mechanics and apply quantum mutual entropy to study the irreversible dynamics of a state for the atom on this model.
The dynamics of an atom on the Jaynes-Cummings model has been studied by an atomic inversion, von Neumann entropy and so on. In this letter, we will treat the Jaynes-Cummings model as a problem in non-equilibrium statistical mechanics and apply quantum mutual entropy to study the irreversible dynamics of a state for the atom on this model.
△ Less
Submitted 14 April, 1998;
originally announced April 1998.
