Hidden Markov random field model and Broyden–Fletcher–Goldfarb–Shanno algorithm for brain image segmentation

Guerrout, El-Hachemi; Ait-Aoudia, Samy; Michelucci, Dominique; Mahiou, Ramdane

doi:10.1080/0952813x.2017.1409280

Cited by 12 publications

(11 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This tool allows for the optimization of the ANN architecture by defining the best number of neurons in the hidden layer and the best activation functions of the hidden and output layers (Vale et al 2017). We used the quasi-Newton algorithm developed by Broyden-Fletcher-Goldfarb-Shanno for processing the neural networks in the IPS (BFGS), which has great resolution power for optimization problems and predictions and is the most popular quasi-Newton method (Guerrout et al 2018).…”

Section: Modeling: Training Of Neural Networkmentioning

confidence: 99%

Artificial intelligence associated with satellite data in predicting energy potential in the Brazilian savanna woodland area

Carrijo¹,

Miguel²,

Vale³

et al. 2020

iForest

View full text Add to dashboard Cite

The use of artificial intelligence to generate information of the savanna's energy capacity may support sustainable management of those areas. We assessed the efficacy of artificial neural networks (ANNs) combined with satellite data to estimate the energy potential (Pe) for cerradão, a dense savannah-like vegetation type in Brazil. We conducted a forest inventory for measuring dendrometric variables and sampling woody materials and barks in a cerradão area in the state of Tocantins, Brazil. The Pe of cerradão biomass was estimated based on the observed higher calorific power and drier biomass values. Six vegetation indices were retrieved from a RapidEye image and tested for correlation to choose the optimum vegetation index for biomass modeling. The basal area and the Normalized Difference Vegetation Index were used as predictors in the Pe modeling. We estimated an average of 19.234 ± 0.411 GJ ton-1 and 19.878 ± 1.090 GJ ton-1 for higher heating values of the wood species and barks, respectively, and an average Pe of 1022.660 GJ ha-1. The best ANN showed an error of 11.3% by using a structure of two, eight, and one neurons in the input layer, in the hidden layer, and in the output layer, respectively, as well as activation functions of the tangential and sigmoidal types. The validation tests showed no significant difference between the observed and ANNpredicted values. Based on our results, we concluded that Pe can be efficiently predicted by combining ANNs and remotely sensed data, which ultimately is a promising tool for forest sustainable management of the cerrado ecosystems.

show abstract

Section: Modeling: Training Of Neural Networkmentioning

confidence: 99%

Artificial intelligence associated with satellite data in predicting energy potential in the Brazilian savanna woodland area

Carrijo¹,

Miguel²,

Vale³

et al. 2020

iForest

View full text Add to dashboard Cite

show abstract

“…It constructs an approximate Hesse matrix without using the second-order partial derivative, and increases the onedimentional search along the Newton direction and has the characteristics of fast convergence. The steps of the quasi-Newton search are as follows [25].…”

Section: Differential Perturbation Strategymentioning

confidence: 99%

Lévy Flight Shuffle Frog Leaping Algorithm Based on Differential Perturbation and Quasi-Newton Search

Zhang

Chen

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Lévy flight Shuffle Frog Leaping Algorithm (LSFLA) is a SFLA variant and enhances the performance of SFLA largely, however, it still has some defects, such as poor convergence and low efficiency. So an improved LSFLA, namely, LSFLA based on Differential perturbation and Quasi-Newton search (DQLSFLA), is proposed in this paper. Firstly, the way of updating only one solution which is the worst one at every sub-iteration in LSFLA is replaced with an all-solution updating way in the subgroup to improve the probability of obtaining the best solution, to omit one sorting step at every sub-iteration and the subiteration number parameter setting, to reduce the computational complexity and to enhance the optimization efficiency. Secondly, a random differential perturbation approach and an improved Lévy flight updating one are created and used in the subgroup updating to form DLSFLA, that is, the first half frogs in the subgroup use the random differential perturbation updating approach to improve the global search ability, and the last half adopt the improved Lévy flight updating method to keep the advantage of the Lévy flight updating method and overcome its defects. Finally, quasi-Newton local search is integrated into DLSFLA near the end of the algorithm to improve the convergence speed. Experimental results on the complex functions from CEC-2014 show that DQLSFLA's convergence quality and optimization efficiency are much better than LSFLA's and that compared with quite a few other state-of-the-art algorithms, DQLSFLA has better performance. The results on medical image enhancement and Quadratic Assignment Problem (QAP) also show that DQLSFLA can solve the real word optimization problems better than LSFLA can. INDEX TERMS Intelligent optimization algorithm, shuffled frog leaping algorithm (SFLA), Lévy flight, quasi-Newton search, image enhancement, quadratic assignment problem.

show abstract

“…Some recent publications further improve previous works. For instance, Guerrout et al [7] described the hidden Markov random field theory and Broyden-Fletcher-Goldfarb-Shanno for tissue labeling. Azimbagirad et al [8] proposed a method based on intensity histogram dedicated to Alzheimer's diseased brains.…”

Section: Introductionmentioning

confidence: 99%

MR brain tissue classification based on the spatial information enhanced Gaussian mixture model

Bian¹

2022

THC

View full text Add to dashboard Cite

BACKGROUND: Classifying T1-weighted Magnetic Resonance brain scans into cerebrospinal fluid, gray matter and white matter is one of the most critical tasks in neurodegenerative disease analysis. Since manual delineation is a labor-intensive and time-consuming process, automated methods have been widely adopted for this purpose. One group of commonly used method by biomedical researchers are based on Gaussian mixture model. The main drawbacks of this model include complex computational cost and parameter selection with the presence of imaging defects such as intensity inhomogeneity and noise. OBJECTIVE: To alleviate these aspects, an improved Gaussian mixture model-based method is proposed in this work. METHODS: Standard mixture model was used to formulate individual voxel intensity. A set of spatial weightings were created to represent local tissue characteristics. The emphasis of this method is its “lite” and robust implementation mode highlighted by a dedicated entropy term. The Expectation-Maximization algorithm was then iteratively executed to estimate model parameters. The Maximum a Posteriori criterion was employed to determine for each voxel if it belongs to a certain tissue. RESULTS: The proposed method was validated on both simulated and real MR scans. The averaged Dice coefficient of segmented brain tissues on each dataset ranged between [66.41, 87.42] for cerebrospinal fluid, [80.57, 85.35] for gray matter, and [83.17, 85.63] for white matter. CONCLUSIONS: Experiments illustrated the effectiveness and reliability in tissue classification against imaging defects compared with manually constructed reference standard.

show abstract

Hidden Markov random field model and Broyden–Fletcher–Goldfarb–Shanno algorithm for brain image segmentation

Cited by 12 publications

References 28 publications

Artificial intelligence associated with satellite data in predicting energy potential in the Brazilian savanna woodland area

Artificial intelligence associated with satellite data in predicting energy potential in the Brazilian savanna woodland area

Lévy Flight Shuffle Frog Leaping Algorithm Based on Differential Perturbation and Quasi-Newton Search

MR brain tissue classification based on the spatial information enhanced Gaussian mixture model

Contact Info

Product

Resources

About