Matrix Texture Research Articles

Electronic structure and spin texture of Mo/N co-doped polar 2D-SiC

Incorporation of Mo and surface symmetry are critical in determining the spin polarisation and texture of a two-dimensional matrix. The spin-polarised band structure and spin texture of a honeycomb monolayer consisting of Mo doped into a 4H-SiC (0 0 1) substrate are investigated using the first-principles plane-wave pseudopotential method based on the density functional theory. The four-fold degenerate energy band that crosses the Fermi level in the interval Γ → M → K → Γ reveals that d electrons with small-polaron behaviour act as itinerate conductive electrons, accounting for the magnetic anisotropy of the material. Interestingly, the spin-texture results demonstrate that both Rashba and Dresselhaus-type spin–orbit couplings coexist in this polar matrix. The incorporation of an N atom within a 2D-SiC-Mo monolayer not only increases the total magnetic moment but also widens the energy band gap of the 2D matrix.

Automatic Process Parameters Tuning and Surface Roughness Estimation for Laser Cleaning

An image analysis-based two-stage process parameters tuning and Surface Roughness (SR) estimation algorithm is proposed for the laser cleaning application. A Cartesian coordinate robot is utilized to collect image and implement cleaning. Before cleaning, in order to tune the proper laser parameters, first, the environment lighting is controlled for the metal image collection. Second, lots of classification features are computed for the images above. The Gray-Level Co-occurrence Matrix (GLCM) texture features, the concavo-convex region features, the histogram symmetry difference feature, and the imaging thermophysical property features are computed. Third, the initial laser parameters are created randomly and an iteration computation is performed: a Support Vector Machine (SVM) is used to forecast the cleaning effect; its inputs include the classification features and the initial laser parameters; its output is the cleaning effect degree. If the SVM output cannot fulfill user’s demand, the laser parameters will be updated randomly. This iteration will be implemented constantly until the SVM output becomes valid. Then the laser cleaning will be performed. When estimating SR for the cleaned metal, multiple image features are calculated for the images after cleaning. The features include the Tamura coarseness, some GLCM features, and the convex region feature. To improve the prediction precision, different feature combinations are used for different cleaning effects. The linear function and the 3-order polynomial function are considered for the SR estimation. After tests, the accuracies of SVM, the SR prediction function, and the integrated SR control and estimation algorithm can be 90.0%, 80.0% and 80.0% approximately.

Effect of MR Field Strength on the Texture Features of Cerebral T2-FLAIR Images: A Pilot Study

Objective To investigate effect of MR field strength on texture features of cerebral T2 fluid attenuated inversion recovery (T2-FLAIR) images. Methods We acquired cerebral 3D T2-FLAIR images of thirty patients who were diagnosed with ischemic white matter lesion (WML) with MR-1.5T and MR-3.0T scanners. Histogram texture features which included mean signal intensity (Mean), Skewness and Kurtosis, and gray level co-occurrence matrix (GLCM) texture features which included angular second moment (ASM), Contrast, Correlation, Inverse difference moment (IDM) and Entropy, of regions of interest located in the area of WML and normal white matter (NWM) were measured by ImageJ software. The texture parameters acquired with MR-1.5T scanning were compared with MR-3.0T scanning. Results The Mean of both WML and NWM obtained with MR-1.5T scanning was significantly lower than that acquired with MR-3.0T (P<0.001), while Skewness and Kurtosis between MR-1.5T and MR-3.0T scanning showed no significant difference (P>0.05). ASM, Correlation and IDM of both WML and NWM acquired with MR-1.5T revealed significantly lower values than those with MR-3.0T (P<0.001), while Contrast and Entropy acquired with MR-1.5T showed significantly higher values than those with MR-3.0T (P<0.001). Conclusion MR field strength showed no significant effect on histogram textures, while had significant effect on GLCM texture features of cerebral T2-FLAIR images, which indicated that it should be cautious to explain the texture results acquired based on the different MR field strength.

Multi-class discrimination of lymphadenopathy by using dual-modal ultrasound radiomics with elastography and B-mode ultrasound

The purpose of our study is to evaluate the diagnostic performance of radiomics in multi-class discrimination of lymphadenopathy based on elastography and B-mode dual-modal ultrasound images. We retrospectively analyzed a total of 251 lymph nodes (89 benign lymph nodes, 70 lymphoma and 92 metastatic lymph nodes) from 248 patients, which were examined by both elastography and B-mode sonography. Firstly, radiomic features were extracted from multimodal ultrasound images, including shape features, intensity statistics features and gray-level co-occurrence matrix texture features. Secondly, three feature selection methods based on information theory were used on the radiomic features to select different subsets of radiomic features, consisting of conditional infomax feature extraction, conditional mutual information maximization, and double input symmetric relevance. Thirdly, the support vector machine classifier was performed for diagnosis of lymphadenopathy on each radiomic subsets. Finally, we fused the results from different modalities and different radiomic feature subsets with Adaboost to improve the performance of lymph node classification. The results showed that the accuracy and overall F1 score with five-fold cross-validation were 76.09%±1.41% and 75.88%±4.32%, respectively. Moreover, when considering on benign lymph nodes, lymphoma or metastatic lymph nodes respectively, the area under the receiver operating characteristic curve of multi-class classification were 0.77, 0.93 and 0.84, respectively. This study indicates that radiomic features derived from multimodal ultrasound images are benefit for diagnosis of lymphadenopathy. It is expected to be useful in clinical differentiation of lymph node diseases.

CP-Violation phase analysis via nontrivial correlation of quarks and leptons in 3 + 1 scenario

The existence and mysterious nature of sterile neutrinos are revolutionizing physics from the particle level to the cosmological scales. The recent results from the MiniBooNE experiment at Fermi-lab observed far more [Formula: see text] appearance than expected, which have provided a hint about the possible existence of sterile neutrinos. The results, if confirmed in future experiments, will have significant implications for cosmology and astroparticle physics. This will require new neutrino mass models to accommodate these additional degrees of freedom. In respect to that, this work is just an extension of our recent work toward the CP phase analysis of Quark-Lepton Complementarity (QLC) model in a [Formula: see text] scenario. The parametrization of CKM4 and PMNS4 using Monte Carlo Simulation is used to estimate the texture of nontrivial correlation matrix [Formula: see text]. As such, we have successfully investigated the constrained values for sterile neutrino parameters, and also predicted the values for Dirac CP-Violation phase and the CP re-phasing invariant [Formula: see text]. The results obtained are consistent with the data available from various experiments, like No[Formula: see text]A, MINOS, SuperK and IceCube-DeepCore. Furthermore, this analysis would be very important in view of growing sterile neutrino experiments.

Quantitative Thermal Imaging Using Grey-level Run Length Matrix Texture Features Correlate to Radiation-Induced Skin Toxicity

Quantitative Thermal Imaging Using Grey-level Run Length Matrix Texture Features Correlate to Radiation-Induced Skin Toxicity

Feasibility of contrast-enhanced MRI derived textural features to predict overall survival in locally advanced breast cancer.

The prognosis for women with locally advanced breast cancer (LABC) is poor and there is a need for better treatment stratification. Gray-level co-occurrence matrix (GLCM) texture analysis of magnetic resonance (MR) images has been shown to predict pathological response and could become useful in stratifying patients to more targeted treatments. To evaluate the ability of GLCM textural features obtained before neoadjuvant chemotherapy to predict overall survival (OS) seven years after diagnosis of patients with LABC. This retrospective study includes data from 55 patients with LABC. GLCM textural features were extracted from segmented tumors in pre-treatment dynamic contrast-enhanced 3-T MR images. Prediction of OS by GLCM textural features was assessed and compared to predictions using traditional clinical variables. Linear mixed-effect models showed significant differences in five GLCM features (f1, f2, f5, f10, f11) between survivors and non-survivors. Using discriminant analysis for prediction of survival, GLCM features from 2 min post-contrast images achieved a classification accuracy of 73% (P < 0.001), whereas traditional prognostic factors resulted in a classification accuracy of 67% (P = 0.005). Using a combination of both yielded the highest classification accuracy (78%, P < 0.001). Median values for features f1, f2, f10, and f11 provided significantly different survival curves in Kaplan-Meier analysis. This study shows a clear association between textural features from post-contrast images obtained before neoadjuvant chemotherapy and OS seven years after diagnosis. Further studies in larger cohorts should be undertaken to investigate how this prognostic information can be used to benefit treatment stratification.

Multimodal feature learning and fusion on B-mode ultrasonography and sonoelastography using point-wise gated deep networks for prostate cancer diagnosis.

B-mode ultrasonography and sonoelastography are used in the clinical diagnosis of prostate cancer (PCa). A combination of the two ultrasound (US) modalities using computer aid may be helpful for improving the diagnostic performance. A technique for computer-aided diagnosis (CAD) of PCa is presented based on multimodal US. Firstly, quantitative features are extracted from both B-mode US images and sonoelastograms, including intensity statistics, regional percentile features, gray-level co-occurrence matrix (GLCM) texture features and binary texture features. Secondly, a deep network named PGBM-RBM2 is proposed to learn and fuse multimodal features, which is composed of the point-wise gated Boltzmann machine (PGBM) and two layers of the restricted Boltzmann machines (RBMs). Finally, the support vector machine (SVM) is used for prostatic disease classification. Experimental evaluation was conducted on 313 multimodal US images of the prostate from 103 patients with prostatic diseases (47 malignant and 56 benign). Under five-fold cross-validation, the classification sensitivity, specificity, accuracy, Youden's index and area under the receiver operating characteristic (ROC) curve with the PGBM-RBM2 were 87.0%, 88.8%, 87.9%, 75.8% and 0.851, respectively. The results demonstrate that multimodal feature learning and fusion using the PGBM-RBM2 can assist in the diagnosis of PCa. This deep network is expected to be useful in the clinical diagnosis of PCa.

Land cover mapping of the Mekong Delta to support natural resource management with multi-temporal Sentinel-1A synthetic aperture radar imagery

Synthetic aperture radar (SAR) has great potential for land cover mapping, especially in tropical regions, where frequent cloud cover obstructs optical remote sensing. The aim of this study is to demonstrate the utility of Sentinel-1A SAR imagery for characterising land use and land cover (LULC) in Bac Lieu, a rapidly developing province of the Mekong Delta, Vietnam, to support natural resource management for land use planning and monitoring. Twenty-one SAR images acquired in 2016 were classified in a four-step process. Firstly, the SAR images were pre-processed to produce Grey Level Co-occurrence Matrix (GLCM) texture images. Then, to reduce the effects of rainfall variation confounding the classification, the images were divided into two categories: dry season (Jan–April) and wet season (May–December) and three input image sets were produced: 1) a single-date composite image (two VV-VH bands), 2) a multi-temporal composite image (eight bands for dry season and sixteen bands for wet season) and 3) a multi-temporal and textural composite image (sixteen bands for dry season and thirty two bands for wet season). We then applied Support Vector Machines (SVM) and Random Forest (RF) classifiers to characterise urban, forest, aquaculture, and rice paddy field for the three input image sets. We tested a combination of input images and classification algorithms and found that no matter the classification algorithms used, multi-temporal images had a higher overall classification accuracy than single-date images and that differences between classification algorithms were minimal. The most accurate classification resulted from a combination of the SVM classifier and multi-temporal and textural input image sets from the dry and wet seasons; with an overall accuracy of 94.81% and Kappa coefficient of 0.92. Our land cover mapping showed that Bac Lieu province is dominated by shrimp farming and rice paddy fields, with natural coastal forested ecosystems only made up a small proportion of the province. The results demonstrated the potential use of SAR as an up-to-date complementary data source of land cover information for local authorities, to support their land use master plan and to monitor illegal land use changes. Long-term monitoring of land use in Bac Lieu province, as well as in the Mekong Delta, will be crucial for decision makers, to ensure sustainability and food security in this region.

Effects of Al2O3 coating induced structural and mechanical changes of C/C composites

Different C/C composites, identified as Cf/C, CAl2O3/C, C1600/C and C1600-Al2O3/C, were fabricated by condensation of different treated 3D needle-punched carbon fiber preforms through a thermal gradient chemical vapor infiltration (TG-CVI) method. Effects of high temperature treatment and Al2O3 coating of the carbon fiber preforms on matrix texture and mechanical properties of C/C composites were investigated. SL PyC is obtained for Cf/C and C1600/C, which preforms are without Al2O3 coating. Whereas the texture of carbon matrix for CAl2O3/C, and C1600-Al2O3/C changes to RL under the same deposition conditions, indicating that Al2O3 coating layer significantly affects the texture of carbon matrix. Heat treatment of the carbon fiber preforms at 1600 °C for 4 h obviously increase the toughness of the C/C composites. The fracture toughness increased by 226% from 0.938 MPa/m2 for Cf/C to 3.061 MPa/m2 for C1600-Al2O3/C. The significant toughness increasing is attributed to the unique fracture mode for C1600-Al2O3/C, where a hollow “cylindrical” PyC matrix is pulled out and a corresponding big hole is left with a central fiber in it. The flexural strength, however, is mainly affected by the final density of the C/C composites.

Static and Dynamic Algorithms for Terrain Classification in UAV Aerial Imagery

The ability to precisely classify different types of terrain is extremely important for Unmanned Aerial Vehicles (UAVs). There are multiple situations in which terrain classification is fundamental for achieving a UAV’s mission success, such as emergency landing, aerial mapping, decision making, and cooperation between UAVs in autonomous navigation. Previous research works describe different terrain classification approaches mainly using static features from RGB images taken onboard UAVs. In these works, the terrain is classified from each image taken as a whole, not divided into blocks; this approach has an obvious drawback when applied to images with multiple terrain types. This paper proposes a robust computer vision system to classify terrain types using three main algorithms, which extract features from UAV’s downwash effect: Static textures- Gray-Level Co-Occurrence Matrix (GLCM), Gray-Level Run Length Matrix (GLRLM) and Dynamic textures- Optical Flow method. This system has been fully implemented using the OpenCV library, and the GLCM algorithm has also been partially specified in a Hardware Description Language (VHDL) and implemented in a Field Programmable Gate Array (FPGA)-based platform. In addition to these feature extraction algorithms, a neural network was designed with the aim of classifying the terrain into one of four classes. Lastly, in order to store and access all the classified terrain information, a dynamic map, with this information was generated. The system was validated using videos acquired onboard a UAV with an RGB camera.

Supervised fusion approach of local features extracted from SAR images for detecting deforestation changes

Deforestation has become a major problem consisting of a continuous regression of forested areas in the world, and for this purpose, an efficient detection of these changes has become more than necessary. In this work, a new method for deforestation change detection is proposed. This approach is based on a supervised fusion of local texture features extracted from SAR images. ALOS PALSAR (Advanced Land Observation Satellite Phased Array type L-band Synthetic Aperture Radar) multi-temporal data have been used in this work. Normalised radar cross-section (NRCS) and polarimetric features extracted from HH and HV polarised data allowed recognising different categories of land covers termed as NRCS classification. Grey-level co-occurrence matrix (GLCM) texture features were extracted by using a different moving window sizes applied on local regions previously obtained by binarisation of the NRCS results. A total of 300 samples of regions and five GLCM characteristics have been used here. The detection of deforestation appears clearly in the resulted images with a very satisfactory precision of the reached regions, and the obtained results of the proposed supervised approach have indeed led to very good detection results of the deforestation change.

Quantifying Subresolution 3D Morphology of Bone with Clinical Computed Tomography

The aim of this study was to quantify sub-resolution trabecular bone morphometrics, which are also related to osteoarthritis (OA), from clinical resolution cone beam computed tomography (CBCT). Samples (n = 53) were harvested from human tibiae (N = 4) and femora (N = 7). Grey-level co-occurrence matrix (GLCM) texture and histogram-based parameters were calculated from CBCT imaged trabecular bone data, and compared with the morphometric parameters quantified from micro-computed tomography. As a reference for OA severity, histological sections were subjected to OARSI histopathological grading. GLCM and histogram parameters were correlated to bone morphometrics and OARSI individually. Furthermore, a statistical model of combined GLCM/histogram parameters was generated to estimate the bone morphometrics. Several individual histogram and GLCM parameters had strong associations with various bone morphometrics (|r| > 0.7). The most prominent correlation was observed between the histogram mean and bone volume fraction (r = 0.907). The statistical model combining GLCM and histogram-parameters resulted in even better association with bone volume fraction determined from CBCT data (adjusted R2 change = 0.047). Histopathology showed mainly moderate associations with bone morphometrics (|r| > 0.4). In conclusion, we demonstrated that GLCM- and histogram-based parameters from CBCT imaged trabecular bone (ex vivo) are associated with sub-resolution morphometrics. Our results suggest that sub-resolution morphometrics can be estimated from clinical CBCT images, associations becoming even stronger when combining histogram and GLCM-based parameters.

WorldView-2 Data for Hierarchical Object-Based Urban Land Cover Classification in Kigali: Integrating Rule-Based Approach with Urban Density and Greenness Indices

The emergence of high-resolution satellite data, such as WorldView-2, has opened the opportunity for urban land cover mapping at fine resolution. However, it is not straightforward to map detailed urban land cover and to detect urban deprived areas, such as informal settlements, in complex urban environments based merely on high-resolution spectral features. Thus, approaches integrating hierarchical segmentation and rule-based classification strategies can play a crucial role in producing high quality urban land cover maps. This study aims to evaluate the potential of WorldView-2 high-resolution multispectral and panchromatic imagery for detailed urban land cover classification in Kigali, Rwanda, a complex urban area characterized by a subtropical highland climate. A multi-stage object-based classification was performed using support vector machines (SVM) and a rule-based approach to derive 12 land cover classes with the input of WorldView-2 spectral bands, spectral indices, gray level co-occurrence matrix (GLCM) texture measures and a digital terrain model (DTM). In the initial classification, confusion existed among the informal settlements, the high- and low-density built-up areas, as well as between the upland and lowland agriculture. To improve the classification accuracy, a framework based on a geometric ruleset and two newly defined indices (urban density and greenness density indices) were developed. The novel framework resulted in an overall classification accuracy at 85.36% with a kappa coefficient at 0.82. The confusion between high- and low-density built-up areas significantly decreased, while informal settlements were successfully extracted with the producer and user’s accuracies at 77% and 90% respectively. It was revealed that the integration of an object-based SVM classification of WorldView-2 feature sets and DTM with the geometric ruleset and urban density and greenness indices resulted in better class separability, thus higher classification accuracies in complex urban environments.

Microstructural evolution and mechanical behavior of Mg/Al laminated composite sheet by novel corrugated rolling and flat rolling

AZ31B Mg/5052 Al laminated composite sheet with superior mechanical property and good sheet shape was fabricated by novel corrugated roll/flat roll corrugating rolling and flat roll/flat roll flat rolling (CFR) at elevated temperatures. The microstructural evolution and mechanical behavior variation of the composite sheet under CFR hot rolling condition were investigated by experimental and numerical simulation methods. The results show that first-pass corrugated rolling induces relatively high and inhomogeneous shear strain in the Mg side and various corrugated interface positions at a low reduction of 35%. It leads to significant grain refinement, outstanding basal texture modification as well as tensile twin activation Mg alloy, which perfectly coordinates the plastic deformation between hard-to-deform Mg and ductile Al and achieve the strong interface metallurgy bonding in Mg/Al sheet and also significantly improves the mechanical properties of Mg/Al composite sheet. The annealing before second-pass flat rolling causes the pronounced static recrystallization and grain growth of Mg alloy in corrugated-rolling Mg/Al sheet, and the formation of ~25 μm thick interfacial compound layer consisted of Mg17Al12 and Al3Mg2 two sublayers. Second-pass flat rolling with a reduction of 30% further imposes severe inhomogeneous shear deformation on the composite sheet, which results in the formation of a wave-shape three-dimensional bonding interface, and the interface compound layers cracking at peak and waist interface positions or fragmenting into well-aligned discrete particles to the bonding interface, in addition to substantial matrix microstructure refinement, texture modification and twin activation in Mg alloy. Such interface structure greatly enhances the interface bonding force of the Mg/Al composite sheet by the larger interface contact area, mechanical gearing effect as well as particle pinning effect. Hence the significant microstructural refinement, strong interface bonding and low residual stress induced by CFR process account for the excellent performance of the Mg/Al laminated composite sheet.

Using neighborhood gray tone difference matrix texture features on dual time point PET/CT images to differentiate malignant from benign FDG-avid solitary pulmonary nodules

ObjectiveLung cancer usually presents as a solitary pulmonary nodule (SPN) on diagnostic imaging during the early stages of the disease. Since the early diagnosis of lung cancer is very important for treatment, the accurate diagnosis of SPNs has much importance. The aim of this study was to evaluate the discriminant power of dual time point imaging (DTPI) PET/CT in the differentiation of malignant and benign FDG-avid solitary pulmonary nodules by using neighborhood gray-tone difference matrix (NGTDM) texture features.MethodsRetrospective analysis was carried out on 116 patients with SPNs (35 benign and 81 malignant) who had DTPI 18F-FDG PET/CT between January 2005 and May 2015. Both PET and CT images were acquired at 1 h and 3 h after injection. The SUVmax and NGTDM texture features (coarseness, contrast, and busyness) of each nodule were calculated on dual time point images. Patients were randomly divided into training and validation datasets. Receiver operating characteristic (ROC) curve analysis was performed on all texture features in the training dataset to calculate the optimal threshold for differentiating malignant SPNs from benign SPNs. For all the lesions in the testing dataset, two visual interpretation scores were determined by two nuclear medicine physicians based on the PET/CT images with and without reference to the texture features.ResultsIn the training dataset, the AUCs of delayed busyness, delayed coarseness, early busyness, and early SUVmax were 0.87, 0.85, 0.75 and 0.75, respectively. In the validation dataset, the AUCs of visual interpretations with and without texture features were 0.89 and 0.80, respectively.ConclusionCompared to SUVmax or visual interpretation, NGTDM texture features derived from DTPI PET/CT images can be used as good predictors of SPN malignancy. Improvement in discriminating benign from malignant nodules using SUVmax and visual interpretation can be achieved by adding busyness extracted from delayed PET/CT images.

Evaluation of TP53/PIK3CA mutations using texture and morphology analysis on breast MRI

PurposeSomatic mutations in TP53 and PIK3CA genes, the two most frequent genetic alternations in breast cancer, are associated with prognosis and therapeutic response. This study predicted the presence of TP53 and PIK3CA mutations in breast cancer by using texture and morphology analyses on breast MRI. Materials and methodsA total of 107 breast cancers (dataset A) from The Cancer Imaging Archive (TCIA) consisting of 40 TP53 mutation cancer and 67 cancers without TP53 mutation; 35 PIK3CA mutations cancer and 72 without PIK3CA mutation. 122 breast cancer (dataset B) from Seoul National University Hospital containing 54 TP53 mutation cancer and 68 without mutations were used in this study. At first, the tumor area was segmented by a region growing method. Subsequently, gray level co-occurrence matrix (GLCM) texture features were extracted after ranklet transform, and a series of features including compactness, margin, and ellipsoid fitting model were used to describe the morphological characteristics of tumors. Lastly, a logistic regression was used to identify the presence of TP53 and PIK3CA mutations. The classification performances were evaluated by accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Taking into account the trade-offs of sensitivity and specificity, the overall performances were evaluated by using receiver operating characteristic (ROC) curve analysis. ResultsThe GLCM texture feature based on ranklet transform is more capable of recognizing TP53 and PIK3CA mutations than morphological feature, especially for the TP53 mutation that achieves statistically significant. The area under the ROC curve (AUC) for TP53 mutation dataset A and dataset B achieved 0.78 and 0.81 respectively. For PIK3CA mutation, the AUC of ranklet texture feature was 0.70. ConclusionTexture analysis of segmented tumor on breast MRI based on ranklet transform is potential in recognizing the presence of TP53 mutation and PIK3CA mutation.

Experimental analysis of sand grain size mapping using UAV remote sensing

ABSTRACTTo map sand grain size distribution using Unmanned Aerial Vehicle (UAV) digital cameras, we conducted experiments for six sand classes with the UAV images acquired at six different heights. We used the maximum likelihood classification and the Grey Level Co-occurrence Matrix texture indices on the images and evaluated classification accuracy under each flight height/image resolution. The image at the spatial resolution of 1.23 mm can effectively separate all six classes. The image of lower resolutions were only effective for coarser classes with an overall accuracy of 83.92% or better: the 2.48 mm pixel size was good for the four classes coarser than medium sand; 3.78 mm was good for the three classes coarser than coarse sand; 4.89 and 5.42 mm were good for the two coarsest classes of sand and gravel. This research is the first comprehensive study to test different UAV survey quality control parameters for sand classification, which could serve as a guidance for UAV survey parameter selection such as flight height and quality of digital cameras.

A Comparative Study of Texture and Convolutional Neural Network Features for Detecting Collapsed Buildings After Earthquakes Using Pre- and Post-Event Satellite Imagery

The accurate and quick derivation of the distribution of damaged building must be considered essential for the emergency response. With the success of deep learning, there is an increasing interest to apply it for earthquake-induced building damage mapping, and its performance has not been compared with conventional methods in detecting building damage after the earthquake. In the present study, the performance of grey-level co-occurrence matrix texture and convolutional neural network (CNN) features were comparatively evaluated with the random forest classifier. Pre- and post-event very high-resolution (VHR) remote sensing imagery were considered to identify collapsed buildings after the 2010 Haiti earthquake. Overall accuracy (OA), allocation disagreement (AD), quantity disagreement (QD), Kappa, user accuracy (UA), and producer accuracy (PA) were used as the evaluation metrics. The results showed that the CNN feature with random forest method had the best performance, achieving an OA of 87.6% and a total disagreement of 12.4%. CNNs have the potential to extract deep features for identifying collapsed buildings compared to the texture feature with random forest method by increasing Kappa from 61.7% to 69.5% and reducing the total disagreement from 16.6% to 14.1%. The accuracy for identifying buildings was improved by combining CNN features with random forest compared with the CNN approach. OA increased from 85.9% to 87.6%, and the total disagreement reduced from 14.1% to 12.4%. The results indicate that the learnt CNN features can outperform texture features for identifying collapsed buildings using VHR remotely sensed space imagery.

Automatic classification of focal liver lesions based on MRI and risk factors.

ObjectivesAccurate classification of focal liver lesions is an important part of liver disease diagnostics. In clinical practice, the lesion type is often determined from the abdominal MR examination, which includes T2-weighted and dynamic contrast enhanced (DCE) MR images. To date, only T2-weighted images are exploited for automatic classification of focal liver lesions. In this study additional MR sequences and risk factors are used for automatic classification to improve the results and to make a step forward to a clinically useful aid for radiologists.Materials and methodsClinical MRI data sets of 95 patients with in total 125 benign lesions (40 adenomas, 29 cysts and 56 hemangiomas) and 88 malignant lesions (30 hepatocellular carcinomas (HCC) and 58 metastases) were included in this study. Contrast curve, gray level histogram, and gray level co-occurrence matrix texture features were extracted from the DCE-MR and T2-weighted images. In addition, risk factors including the presence of steatosis, cirrhosis, and a known primary tumor were used as features. Fifty features with the highest ANOVA F-score were selected and fed to an extremely randomized trees classifier. The classifier evaluation was performed using the leave-one-out principle and receiver operating characteristic (ROC) curve analysis.ResultsThe overall accuracy for the classification of the five major focal liver lesion types is 0.77. The sensitivity/specificity is 0.80/0.78, 0.93/0.93, 0.84/0.82, 0.73/0.56, and 0.62/0.77 for adenoma, cyst, hemangioma, HCC, and metastasis, respectively.ConclusionThe proposed classification system using features derived from clinical DCE-MR and T2-weighted images, with additional risk factors is able to differentiate five common types of lesions and is a step forward to a clinically useful aid for focal liver lesion diagnosis.