A Robust Learning Algorithm Based on Particle Swarm Optimization for Pi-Sigma Artificial Neural Networks.

Artificial neural networks (ANNs) have been frequently used in forecasting problems in recent years. One of the most popular types of ANNs in these days is Pi-Sigma artificial neural networks (PS-ANNs). PS-ANNs have a high order ANN structure and they use both multiplicative and additive neuron models in their architecture. PS-ANNs produce superior forecasting performance because of their high order structure. PS-ANNs are affected negatively by an outlier or outliers in a data set because of having a multiplicative neuron model in their architecture. In this study, a new robust learning algorithm based on particle swarm optimization and Huber's loss function for PS-ANNs is proposed. To evaluate the performance of the proposed method, Dow Jones stock exchange and Australian beer consumption data sets are analyzed and the obtained results are compared with many ANNs types proposed in the literature. Besides, the performance of the proposed method in outlier cases is also investigated by injecting outliers into these data sets. It is seen that the proposed learning algorithm has a satisfying performance both the data have an outlier or outliers' case and original case.

A new automatic forecasting method based on a new input significancy test of a single multiplicative neuron model artificial neural network.

The model adequacy and input significance tests have not been proposed as features for the specification of a single multiplicative neuron model artificial neural networks in the literature. Moreover, there is no systematic approach based on hypothesis tests for using single multiplicative neuron model artificial neural networks for forecasting purposes like classical time series forecasting methods. In this study, new methods are proposed to solve these problems. The performance of the proposed test procedures is investigated in a simulation study. According to simulation results, the proposed tests have very good performance. Moreover, the test procedures are illustrated by using two real-world examples. The second contribution of the paper is that an automatic forecasting method is proposed based on input significance and model adequacy tests and the particle swarm optimization-based learning algorithm. The proposed automatic forecasting method is applied to M4 competition hourly data sets, and it is the best pure machine learning method among others in the competition. The proposed automatic forecasting method is more accurate than all benchmarks, such as MLP, RNN, and ETS, which were proposed by the competition organizers.

Threshold single multiplicative neuron artificial neural networks for non-linear time series forecasting.

Single multiplicative neuron artificial neural networks have different importance than many other artificial neural networks because they do not have complex architecture problem, too many parameters and they need more computation time to use. In single multiplicative neuron artificial neural network, it is assumed that there is a one data generation process for time series. Many time series need an assumption that they have two data generation process or more. Based on this idea, the threshold model structure can be employed in a single multiplicative neuron model artificial neural network for taking into considering data generation processes problem. In this study, a new artificial neural network type is proposed and it is called a threshold single multiplicative neuron artificial neural network. It is assumed that time series have two data generation processes according to the architecture of single multiplicative neuron artificial neural network. Training algorithms are proposed based on harmony search algorithm and particle swarm optimization for threshold single multiplicative neuron artificial neural network. The proposed method is tested by various time series data sets and compared with well-known forecasting methods by considering different error measures. Finally, the performance of the proposed method is evaluated by a simulation study.

Laterality of the thyroid nodules, anatomic and sonographic, as an estimator of thyroid malignancy and its neoplastic nature by comparing the Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) and histopathology.

PURPOSE: To assess the association between the topographic and sonographic laterality of the thyroid nodules and the malignancy for those who had undergone ultrasonography (US)-guided fine-needle aspiration (FNA) (US-FNA) and following relevant indicated thyroidectomy. METHODS: A retrospective analysis from April 2011 to October 2015 was conducted by enrolling the documents of 501 consecutive eligible patients with 601 thyroid nodules who had undergone neck US, Doppler US, and US-FNA. The prediction of malignancy by means of laterality of 95 thyroid nodules with undetermined cytology on the basis of the Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) was evaluated histopathologically with comparison of three locations, separately. RESULTS: Six hundred and one nodules in 501 cases were studied and 249 nodules (49.8%) were topographically located at the right lobe (Location 1/Loc1), while 255 (42.4%) at the left lobe (Location 2/Loc2), 46 (7.7%) at the isthmus (Location 3/Loc3), and 1 (0.2%) was an accessory thyroid gland (Location 4/Loc4). Three different comparisons were performed regarding the locations, which revealed that the specificity did not change regarding the locations while the sensitivity of Loc3 was higher than that of Loc1 and Loc2. CONCLUSIONS: The preliminary data of 4.5-year single-center study proved that the isthmus location may be more beneficial to estimate the malignancy on the basis of toposonographic laterality of the nodules with undetermined cytology. This notewothy outcome may be considered particularly for the challenging cases with undetermined cytology in Endocrine Surgery and Thyroidology.

Can cut-off points of 10 and 15 mm of thyroid nodule predict malignancy on the basis of three diagnostic tools: i) strain elastography, ii) the Bethesda System for Reporting Thyroid Cytology with 27-gauge fine-needle, and iii) histopathology?

PURPOSE: This study evaluated whether the cut-offs 10 and 15 mm can help distinguish malignant from benign nodules regarding three diagnostic tools: i) strain elastography (SE), ii) the Bethesda System for Reporting Thyroid Cytopathology (TBSRTC), iii) histopathology. METHODS: From 2012 to April 2015, a retrospective analysis was conducted by enrolling the data of 425 consecutive eligible patients with 500 thyroid nodules. The efficacy of the nodule size, as of the cut-offs, on the estimation for malignancy had been analysed on the basis of the three diagnostic tools. RESULTS: Of the 500 thyroid nodules examined, 80 (16.0%) were under 10 mm and 420 (84.0%) were over 10 mm in diameter. No significant difference was found between over 10 mm with i) TES (Tsukuba Elasticity Score) 4 and 5, area under the curve (AUC) 0.531, ii) TBSRTC (The Bethesda System for Reporting Thyroid Cytopathology) III, IV, V, VI, undetermined and malignant cytology, AUC 0.517, iii) malignant histopathology, AUC 0.509. Similarly, no significance difference was recognized between over 15 mm with i) TES 4 and 5, AUC 0.623, ii) undetermined and malignant cytology, AUC 0.455, iii) malignant histopathology, AUC 0.515 by McNemar test. However, size over 15 mm may strengthen the prediction among TES 4 and 5 and malignant histopathology, as weakens in undetermined and malignant cytology. CONCLUSIONS: These preliminary data of 3-year single-center study suggest that assignment of 10 and 15 mm as the cut-off points of the thyroid nodules may not be predictive of malignancy on the basis of three diagnostic tools. Nevertheless, higher cut-off may corrobarate the correlation with TES 4 and 5 and malignant histopathology while attenuation with TBSRTC III, IV, V, and VI, confront with the lower one, 10 mm.

Comparing the Effect of Different Voxel Resolutions for Assessment of Vertical Root Fracture of Permanent Teeth.

BACKGROUND: The teeth with undiagnosed vertical root fractures (VRFs) are likely to receive endodontic treatment or retreatment, leading to frustration and inappropriate endodontic therapies. Moreover, many cases of VRFs cannot be diagnosed definitively until the extraction of tooth. OBJECTIVES: This study aimed to assess the use of different voxel resolutions of two different cone beam computerized tomography (CBCT) units in the detection VRFs in vitro. MATERIALS AND METHODS: The study material comprised 74 extracted human mandibular single rooted premolar teeth without root fractures that had not undergone any root-canal treatment. Images were obtained by two different CBCT units. Four image sets were obtained as follows: 1) 3D Accuitomo 170, 4 × 4 cm field of view (FOV) (0.080 mm(3)); 2) 3D Accuitomo 170. 6 × 6 cm FOV (0.125 mm(3)); 3) NewTom 3G, 6˝ (0.16 mm(3)) and 4) NewTom 3G, 9˝ FOV (0.25 mm(3)). Kappa coefficients were calculated to assess both intra- and inter-observer agreements for each image set. RESULTS: No significant differences were found among observers or voxel sizes, with high average Z (Az) results being reported for all groups. Both intra- and inter-observer agreement values were relatively better for 3D Accuitomo 170 images than the images from NewTom 3G. The highest Az and kappa values were obtained with 3D Accuitomo 170, 4 × 4 cm FOV (0.080 mm(3)) images. CONCLUSION: No significant differences were found among observers or voxel sizes, with high Az results reported for all groups.

Evaluation of the prevalence of bifid mandibular condyle detected on cone beam computed tomography images in a Turkish population / Evaluación de la prevalencia de cóndilo mandibular bífido detectado en tomografía computadorizada cone-beam en una población turca

The aim of this study was to assess the frequency of the BMC phenomenon in a Turkish patient population. Cone beam computed tomography (CBCT) images of 2634 consecutive patients were retrospectively reviewed. The Chi-squared test was used to determine potential differences in the distribution of BMCs when stratified by sex and side. Among the 2634 patients, 42 (1.7%) patients were found to have BMC. Of these 42 patients, 22 were female (0.8%) and 20 were male (0.7%) with age ranging from 29 to 68 years (mean age 47.47). Among the 42 patients, 39 (92.8%) of the BMCs were unilateral and three (7.1%) were bilateral. Approximately 24 cases (53.3%) were on the right side, and 21 cases (46.6%) were on the left side. All of the BMCs showed a mediolateral orientation. The mean depth of the BMC was 2.55 mm in males and 2.68 mm in females. 2 patients have symptoms whereas the other patients were atraumatic and asymptomatic. BMC is a rare condition that might be more prevalent in the Turkish population. Greater detailed information regarding BMC could be obtained by the widespread use of CBCT in epidemiological studies.

El objetivo de este estudio fue evaluar la frecuencia del fenómeno CMB en una población de pacientes de Turquía. Se revisaron imágenes consecutivas de tomografía computarizada (CBCT) de 2.634 pacientes retrospectivamente. Se utilizó la prueba de Chi-cuadrado para determinar las posibles diferencias en la distribución de CMB estratificado por sexo y lado. Entre los 2.634 pacientes, se encontró que 42 (1,7%) pacientes tenían CMB. De estos 42 pacientes, 22 eran mujeres (0,8%) y 20 eran varones (0,7%), con edades entre 29 a 68 años (promedio edad 47,47). Entre los 42 pacientes, 39 (92,8%) del CMB fueron unilaterales y tres (7,1%) fueron bilaterales. Aproximadamente 24 casos (53,3%) estaban en el lado derecho, y 21 casos (46,6%) estaban en el lado izquierdo. Todas las CMB mostraron una orientación mediolateral. La profundidad media de la CMB era 2,55 mm en los hombres y 2,68 mm en las mujeres. Dos de los pacientes presentaron síntomas, mientras que en el resto de los pacientes no presentó trauma ni síntomas. CMB es una afección poco común que podría ser más frecuente en la población turca. Mayor información y detalle sobre CMB se podría obtener en estudios epidemiológicos con el uso generalizado de CBCT.

Comparison of intraoral radiography and cone-beam computed tomography for the detection of horizontal root fractures: an in vitro study.

OBJECTIVES: This study aimed to compare the diagnostic accuracy of two different cone-beam computed tomography (CBCT) units with several intraoral radiography techniques for detecting horizontal root fractures. METHODS: The study material comprised 82 extracted human maxillary incisors without root fractures that had not undergone any root canal treatment. Root fractures were created in the horizontal plane in 31 teeth by a mechanical force using a hammer with the tooth placed on a soft foundation as described in a previous study. The teeth were divided into two groups: a control group with no fractures and a test group with fractures. These were randomized to the empty maxillary anterior sockets of a dry human maxilla. Each tooth was imaged at various vertical angles using each of the following modalities: a 3D Accuitomo 170 CBCT, a NewTom 3G CBCT, a VistaScan PSP, a CCD sensor, and conventional film. Specificity and sensitivity for assessing horizontal root fracture by each radiographic technique were calculated. Chi-square statistics were used to evaluate differences between modalities. Kappa statistics assessed the agreement between observers. Results were considered significant at P < 0.05. RESULTS: The kappa values for inter-observer agreement between observers ranged between 0.88 and 0.98 for the 3D Accuitomo 170, 0.82 and 0.91 for the NewTom 3G, and 0.61 and 0.72 for the different types of intraoral images. The diagnostic accuracy for detecting fracture lines in 3D Accuitomo 170 (0.93) was significantly higher than NewTom 3G (0.87), VistaScan (0.71), CCD (0.70), and CF (0.68). CONCLUSIONS: 3D Accuitomo 170 has the highest sensitivity and diagnostic accuracy for detecting horizontal root fracture among the 5 radiographic modalities examined. CBCT should be considered as the most reliable imaging modality of choice for the diagnosis of horizontal root fracture. CLINICAL RELEVANCE: CBCT imaging offers the clear advantage over conventional imaging that traumatized teeth can be visualized in all three dimensions--especially the oro-facial dimension.

Retrospective analysis of nasal soft tissue profile changes with maxillary surgery.

PURPOSE: The aim of this study was to analyze the changes in the position of the nasal and labial soft tissue profile of patients undergoing bimaxillary orthognathic surgery, with special emphasis on the effect on the nasal tip projection. MATERIALS AND METHODS: The lateral cephalometric radiographs of 27 consecutive patients (16 female and 11 male patients; mean age, 22 years) who had undergone maxillary advancement and mandibular setback were studied. The pretreatment and end-of-treatment lateral cephalometric radiographs were selected. The pretreatment and end-of-treatment radiographs were superimposed on the sella-nasion plane, and the case was only included if there had been no change in sella-nasion length (ie, no growth). Analyses of Pearson correlation coefficient and stepwise linear regression tests were used to compare the cephalometric measurements at the beginning and at the end of treatment. Paired-sample t tests were also performed to analyze changes in nasolabial angle (NLA) and columella-lobular angle (CLA). RESULTS: The correlations between vertical movement of nasal tip, A-point, and maxillary incisal tip were important. Although there was an important correlation between nasal and incisal tip, interestingly, there was no correlation between nasal tip and A-point in horizontal movement. According to stepwise linear regression analysis, the best model for horizontal movement of nasal tip was as follows: Nasal anteroposterior movement = 0.241 + 0.188 × Incisal tip anteroposterior movement + 0.153 × Incisal tip superoinferior movement. For vertical movement of nasal tip, the best model was as follows: Nasal superoinferior movement= -1.117 + 0.399 × Incisal tip superoinferior movement + 0.323 × A-point anteroposterior movement. There was no significant relation in angular measurements of NLA and CLA before and after treatment. CONCLUSION: The results of our study suggest that both horizontal and vertical movements of nasal tip were related to incisal tip and A-point movements; however, angular changes in CLA and NLA did not affect the nasal tip.

Radiologic investigation of prevalence, associated pathologies and dental anomalies of non-third molar impacted teeth in Turkish oral patients.

OBJECTIVE: To investigate the prevalence and associated pathologies of impacted teeth in Turkish oral patients. METHODS: A retrospective survey was carried out in 12,129 patients who visited the Department of Oral Diagnosis and Radiology, Ondokuz Mayis University, Faculty of Dentistry, Turkey, from January 2003 to December 2007. The minimum age for inclusion was 14 years and third molar impactions were excluded from the study. To be enrolled in the study, the patient's chart had to contain a panoramic radiograph with supplemental periapical radiographs. One radiologist examined all radiographs to determine the number, orientation and types of impacted teeth and the presence of associated pathologies and developmental dental anomalies associated with this phenomenon. RESULTS: Of the 12,129 patients, 1117 (9.2%) patients aged 14 to 80 years had one or more dental impactions (in total 1356 impacted teeth). The male to female ratio was 1:1.4 (457:660). The maxillary canine teeth were the most commonly encountered (71.5%), followed by the mandibular premolars (8.6%). The analysis of the orientation of the impacted teeth showed that 480 impacted teeth were in a mesioangular position (35.4%), followed by vertical (28.9%), distoangular (18.9%), horizontal (16.5%) and buccolingual (0.3%) orientations. CONCLUSION: The prevalence of non-third molar impacted teeth was 9.2% among Turkish oral patients. The maxillary canines were the most frequent impacted teeth. The most common orientations of impacted teeth were the mesioangular position and vertical orientation. The most frequent associated pathologic change was cystic change.