Computed Texture Features Research Articles

Modeling visual aesthetic perception: bridges between computed texture features and perceived beauty qualities in semantic experiments.

The exploration of the potential relationship between computable low-level texture, such as features extracted from color and texture, and the perceived high-level aesthetic properties, such as warm or cold, soft or hard, has been a hot research topic of neuroaesthetics. First, the selection and clustering of aesthetic antonyms used to represent the aesthetic properties of visual texture are completed through two semantic differential experiments. Subsequently, 151 visual textures are rated according to the selected aesthetic antonyms by participants in a third semantic differential experiment. Third, 106 textural features are extracted using four different image analysis algorithms to describe the low-level characteristics of visual textures. Finally, the construction and evaluation of the visual aesthetic perception model based on multiple linear and nonlinear regression algorithms are discussed. We analyzed the frequency of each aesthetic antonym selected from 20 pairs of semantic antonyms, and the most frequently mentioned 8 pairs of semantic antonyms were selected as the core set for model building. The extracted low-level features are highly correlative. Of the correlation coefficients based on absolute values, 3383 are less than 0.75, accounting for 14.84% of the total. The correlation coefficients were larger than 0.5 accounts for 27.29% of the total. Through neighborhood component analysis, the top 10 low-level features are selected with lower correlation. The gap between low-level calculated features and high-level perceived aesthetic emotions can be bridged by a brain-inspired model of visual aesthetic perception.

Quantifying invasiveness of clinical stage IA lung adenocarcinoma with computed tomography texture features

ObjectivesThe study objectives were to establish and validate a nomogram for pathological invasiveness prediction in clinical stage IA lung adenocarcinoma and to help identify those potentially unsuitable for sublobar resection–based computed tomography texture features. MethodPatients with clinical stage IA lung adenocarcinoma who underwent surgery at Guangdong Provincial People's Hospital between January 2015 and October 2018 were retrospectively reviewed. All surgically resected nodules were pathologically classified into less-invasive and invasive cohorts. Each nodule was manually segmented, and its computerized texture features were extracted. Clinicopathological and computed tomographic texture features were compared between 2 cohorts. A nomogram for distinguishing the pathological invasiveness was established and validated. ResultsAmong 428 enrolled patients, 249 were diagnosed with invasive pathological subtypes. Smoking status (odds ratio, 2.906; 95% confidence interval, 1.285-6.579; P = .011), mean computed tomography attenuation value (odds ratio, 1.005, 95% confidence interval, 1.002-1.007; P < .001), and entropy (odds ratio, 8.536, 95% confidence interval, 3.478-20.951; P < .001) were identified as independent predictors for pathological invasiveness by multivariate logistics regression analysis. The nomogram showed good calibration (P = .182) with an area under the curve of 0.849 when validated with testing set data. Decision curve analysis indicated the potentially clinical usefulness of the model with respect to treat-all or treat-none scenario. Compared with intraoperative frozen-section, the nomogram performed better in pathological invasiveness diagnosis (area under the curve, 0.815 vs 0.670; P = .00095). ConclusionsWe established and validated a nomogram to compute the probability of invasiveness of clinical stage IA lung adenocarcinoma with great calibration, which may contribute to decisions related to resection extent.

Texture Analysis in the Diagnosis of Primary Breast Cancer: Comparison of High-Resolution Dedicated Breast Positron Emission Tomography (dbPET) and Whole-Body PET/CT.

Objective: This retrospective study aimed to compare the ability to classify tumor characteristics of breast cancer (BC) of positron emission tomography (PET)-derived texture features between dedicated breast PET (dbPET) and whole-body PET/computed tomography (CT).Methods: Forty-four BCs scanned by both high-resolution ring-shaped dbPET and whole-body PET/CT were analyzed. The primary BC was extracted with a standardized uptake value (SUV) threshold segmentation method. On both dbPET and PET/CT images, 38 texture features were computed; their ability to classify tumor characteristics such as tumor (T)-category, lymph node (N)-category, molecular subtype, and Ki67 levels was compared. The texture features were evaluated using univariate and multivariate analyses following principal component analysis (PCA). AUC values were used to evaluate the diagnostic power of the computed texture features to classify BC characteristics.Results: Some texture features of dbPET and PET/CT were different between Tis-1 and T2-4 and between Luminal A and other groups, respectively. No association with texture features was found in the N-category or Ki67 level. In contrast, receiver-operating characteristic analysis using texture features' principal components showed that the AUC for classification of any BC characteristics were equally good for both dbPET and whole-body PET/CT.Conclusions: PET-based texture analysis of dbPET and whole-body PET/CT may have equally good classification power for BC.

Radial textures: a new algorithm to analyze meat quality on MRI

Magnetic Resonance Imaging (MRI) was introduced as an alternative to destructive methods for determining the quality parameters of meat products, since MRI is a non-destructive, non-ionizing, and innocuous technique. The use of MRI is linked to the use of computer vision algorithms and data mining techniques. Thus, the success of MRI could not be understood without referring to the implementation of powerful algorithms to handle all data generated. In this study, a new texture algorithm was developed in order to obtain new features based on the radial distribution of the textures from MRI. This new algorithm is called Radial Texture Algorithm (RTA) and it was compared with four classical texture features approaches. The results obtained allow describing the semantic context of the texture features for this new algorithm and the relationship among the features from this algorithm and the classical approaches. Besides, the results obtained by means of these computer vision algorithms were correlated with the results obtained by means of physico-chemical and sensory parameters of the Iberian loins. All of computer vision algorithms achieved correlation coefficients higher than 0.500, noting that RTA reached the highest correlation coefficients in almost all cases. These high correlation coefficients confirm the proposed algorithm as an alternative to the other computational texture features approaches in order to compute the physico-chemical and sensory parameters of meat products in an accurate, fast, non-destructive and efficient way.

Radiogenomics of rectal adenocarcinoma in the era of precision medicine: A pilot study of associations between qualitative and quantitative MRI imaging features and genetic mutations

ObjectiveTo investigate associations between genetic mutations and qualitative as well as quantitative features on MRI in rectal adenocarcinoma at primary staging. MethodsIn this retrospective study, patients with rectal adenocarcinoma, genome sequencing, and pretreatment rectal MRI were included. Statistical analysis was performed to evaluate associations between qualitative features obtained from subjective evaluation of rectal MRI and gene mutations as well as between quantitative textural features and gene mutations. For the qualitative evaluation, Fisher’s Exact test was used to analyze categorical associations and Wilcoxon Rank Sum test was used for continuous clinical variables. For the quantitative evaluation, we performed manual segmentation of T2-weighted images for radiomics-based quantitative image analysis. Thirty-four texture features consisting of first order intensity histogram-based features (n = 4), second order Haralick textures (n = 5), and Gabor-edge based Haralick textures were computed at two different orientations. Consensus clustering was performed with 34 computed texture features using the K-means algorithm with Euclidean distance between the texture features. The clusters resulting from the algorithm were then used to enumerate the prevalence of gene mutations in those clusters. ResultsIn 65 patients, 45 genes were mutated in more than 3/65 patients (5%) and were included in the statistical analysis. Regarding qualitative imaging features, on univariate analysis, tumor location was significantly associated with APC (p = 0.032) and RASA1 mutation (p = 0.032); CRM status was significantly associated with ATM mutation (p = 0.021); and lymph node metastasis was significantly associated with BRCA2 (p = 0.046) mutation. However, these associations were not significant after adjusting for multiple comparisons. Regarding quantitative imaging features, Cluster C1 had tumors with higher mean Gabor edge intensity compared with cluster C2 (θ = 0°, p = 0.018; θ = 45°, p = 0.047; θ = 90°, p = 0.037; cluster C3 (θ = 0°, p = 0.18; θ = 45°, p = 0.1; θ = 90°, p = 0.052), and cluster C4 (θ = 0°, p = 0.016; θ = 45°, p = 0.033; θ = 90°, p = 0.014) suggesting that the cluster C1 had tumors with more distinct edges or heterogeneous appearance compared with other clusters. ConclusionsAlthough this preliminary study showed promising associations between quantitative features and genetic mutations, it did not show any correlation between qualitative features and genetic mutations. Further studies with larger sample size are warranted to validate our preliminary data.

A computerised framework for characterisation of breast tissue using mammographic images

In this study various experiments have been conducted for four-class and two-class breast tissue pattern characterisation using various texture feature models in transform domain. These experiments have been conducted on 480 mammographic images taken from the DDSM dataset. From each image, ROI of predefined size 128 × 128 pixels has been extracted from the core location of the breast tissue where remarkable amount of glandular tissues are found. Various statistical parameters have been computed from each ROI using transform domain texture feature models. The computed texture feature vectors are passed to the classification phase. The classification phase is consisted of SVM classifier. From the exhaustive experiments conducted in this study, it can be concluded that the maximum accuracy of 86.2% has been obtained for four-class breast tissue characterisation and the maximum accuracy of 91.2% has been obtained for two-class breast tissue characterisation.

Segmentation of Apples in Aerial Images under Sixteen Different Lighting Conditions Using Color and Texture for Optimal Irrigation

Due to the changes in the lighting intensity and conditions throughout the day, machine vision systems used in precision agriculture for irrigation management should be prepared for all possible conditions. For this purpose, a complete segmentation algorithm has been developed for a case study on apple fruit segmentation in outdoor conditions using aerial images. This algorithm has been trained and tested using videos with 16 different light intensities from apple orchards during the day. The proposed segmentation algorithm consists of five main steps: (1) transforming frames in RGB to CIE L*u*v* color space and applying thresholds on image pixels; (2) computing texture features of local standard deviation; (3) using intensity transformation to remove background pixels; (4) color segmentation applying different thresholds in RGB space; and (5) applying morphological operators to refine the results. During the training process of this algorithm, it was observed that frames in different light conditions had more than 58% color sharing. Results showed that the accuracy of the proposed segmentation algorithm is higher than 99.12%, outperforming other methods in the state of the art that were compared. The processed images are aerial photographs like those obtained from a camera installed in unmanned aerial vehicles (UAVs). This accurate result will enable more efficient support in the decision making for irrigation and harvesting strategies.

A Texture Analysis–Based Prediction Model for Lymph Node Metastasis in Stage IA Lung Adenocarcinoma

Some clinical N0 lung adenocarcinomas have been pathologically diagnosed as N1 or N2. To improve the preoperative diagnostic accuracy of lymph node disease, we developed a prediction model for lymph node metastasis in cT1 N0 M0 lung adenocarcinoma based on computed tomography texture analysis and clinical characteristics to estimate the probability of lymph node metastasis. The records of 501 consecutive patients with cT1 N0 M0 lung adenocarcinoma who underwent computed tomography scan and pulmonary resection with systematic lymph nodes dissection or lymph nodes sampling were reviewed. Each nodule was manually segmented, and its computerized texture features were extracted. Multivariate logistic regression with fivefold validation was used to estimate independent predictors and build the prediction model. The prediction model was then externally validated. A nomogram was developed based on logistic regression results. Among 501 patients, 41 were diagnosed withpositive lymph nodes (8.18%). Four independent predictors were identified: the skewness and 90th percentile of computed tomography number, nodule compactness, and carcinoembryonic antigen level. This model showed good calibration (Hosmer-Lemeshow test, p= 0.337), with an area under the curve of 0.883(95% confidence interval, 0.842 to 0.924; p < 0.001). The area under the curve was 0.808 (95% confidence interval, 0.735 to 0.880) when validated with independent data. A model based on computerized textures and carcinoembryonic antigen level can assess the lymph node status of patients with cT1 N0 M0 lung adenocarcinoma preoperatively, which could assist surgeons in making subsequent clinical decisions.

Effect of Cooking Conditions on Quality Characteristics of Confit Cod: Prediction by MRI

Abstract This study evaluates the effect of low temperature (60, 64 and 68 °C) and different times (15 and 20 min) of cooking on physicochemical and sensory characteristics of confit cod and analyzes confit cod in a non-destructive way by means of magnetic resonance imaging (MRI) and computer vision techniques. Higher scores for acceptability and flavor in 60 °C–20 min and 64 °C–20 min samples and some physicochemical differences were found. These results allow distinguishing four groups of confit cods as a function of cooking conditions – 60 °C–15 min/68 °C–15 min/68 °C–20 min/60 °C–20 min, 64 °C–15 min, 64 °C–20 min – and proposing to cook confit cod at 60–64 °C during 20 min. Prediction by means of computational texture features from MRI gave moderate-to-good correlation (0.6–0.75) for six quality attributes and very good-to-excellent relationship (0.75–1) for other six. Thus, computational texture features seem to be appropriate to determine the quality attributes of confit cod in a non-destructive way.

Optimization of MRI Acquisition and Texture Analysis to Predict Physico-chemical Parameters of Loins by Data Mining

The main objective of this study was to configure the acquisition and analysis of low-field magnetic resonance imaging (MRI) to predict physico-chemical characteristics of Iberian loin, evaluating the use of different MRI sequences (spin echo, SE; gradient echo, GE; turbo 3D, T3D), computational texture feature methods (GLCM, NGLDM, GLRLM, GLCM + NGLDM + GLRLM), and data mining techniques (multiple linear regression, MLR; isotonic regression, IR). Moderate to very good correlation coefficients and low mean absolute error were found when applying MLR or IR on any method of computational texture features from MRI acquired with SE or GE. For T3D sequence, accurate results are only obtained by applying IR on GLCM or GLCM + NGLDM + GLRLM methods. Considering not only the accuracy of the methodology but also consumed time and required resources, the use of SE sequences for MRI acquisition, GLCM method for MRI texture analysis, and MLR could be indicated for prediction physico-chemical characteristics of loin.

P1.03-024 Accuracy of Combined Semantic and Computational CT Features in Predicting Non-Small Cell Lung Cancer Subtype

With improvements in molecular treatment, it is increasingly important to differentiate non-small cell lung cancer (NSCLC) subtypes, i.e., adenocarcinoma(ADCA) from squamous cell cancer(SCCA). Many patients cannot undergo invasive biopsy procedures and so non-invasive classification methods would be helpful in their management. Most studies using CT scans for this purpose have used either semantic (visual assessment of CT images by a radiologist) or computational texture features, yielding modest accuracy. We hypothesized that combined semantic and computational assessment of CT scans would improve the accuracy of CT in NSCLC classification. 67 patients (38 ADC, 29 SCCA) underwent contrast-enhanced chest CT for lung cancer staging. Tumor volumes of interests (VOI) were drawn semi-automatically. 10 qualitative semantic and 361 computational texture features were derived from the VOIs. Univariate and multivariate logistic regression models(MLRM) were developed for combinations of semantic and texture features. Sensitivity, specificity, and area under the receiver operating characteristic (AUROC) curve were computed. 10-fold cross-validation was used to prevent overfitting. Univariate models found two semantic (air-bronchogram, shape) and five texture parameters (wavelet-transform based: GLCM_Correlation, GLRL_LGRE, GLRL_LGRE, GLRL_LGRE, and original VOI-based GLSZM_ZSN[1]) to be most predictive of tumor class (p-value <0.01). Sensitivity, specificity, and AUROC for MLRM utilizing semantic features alone was 64.2%, 73.3%, and 0.76, and that of MLRM for texture features alone was 74.6%, 72.3%, and 0.79, respectively. For combined model involving semantic and texture features (i.e., air-bronchogram and GLCM_Correlation), respective values were 81.2%, 90%, and 0.9. [1] GLCM: gray-level cooccurence matrix, GLRL_LGRE: gray-level run-length matrix-derived low gray run emphasis, GLSZM_ZSN: Gray-level size-zone matrix-derived zone-size nonuniformity. Combined semantic and computational texture assessment of lung cancer CT images is highly accurate in differentiation of SCCA and ADCA.

Aesthetics by Numbers: Links between Perceived Texture Qualities and Computed Visual Texture Properties.

Our world is filled with texture. For the human visual system, this is an important source of information for assessing environmental and material properties. Indeed—and presumably for this reason—the human visual system has regions dedicated to processing textures. Despite their abundance and apparent relevance, only recently the relationships between texture features and high-level judgments have captured the interest of mainstream science, despite long-standing indications for such relationships. In this study, we explore such relationships, as these might be used to predict perceived texture qualities. This is relevant, not only from a psychological/neuroscience perspective, but also for more applied fields such as design, architecture, and the visual arts. In two separate experiments, observers judged various qualities of visual textures such as beauty, roughness, naturalness, elegance, and complexity. Based on factor analysis, we find that in both experiments, ~75% of the variability in the judgments could be explained by a two-dimensional space, with axes that are closely aligned to the beauty and roughness judgments. That a two-dimensional judgment space suffices to capture most of the variability in the perceived texture qualities suggests that observers use a relatively limited set of internal scales on which to base various judgments, including aesthetic ones. Finally, for both of these judgments, we determined the relationship with a large number of texture features computed for each of the texture stimuli. We find that the presence of lower spatial frequencies, oblique orientations, higher intensity variation, higher saturation, and redness correlates with higher beauty ratings. Features that captured image intensity and uniformity correlated with roughness ratings. Therefore, a number of computational texture features are predictive of these judgments. This suggests that perceived texture qualities—including the aesthetic appreciation—are sufficiently universal to be predicted—with reasonable accuracy—based on the computed feature content of the textures.

Data Mining on MRI-Computational Texture Features to Predict Sensory Characteristics in Ham

In this study, data mining technique was applied on computational texture features obtained from the analysis of magnetic resonance imaging (MRI) of hams, with the main objective of determining sensory attributes of dry-cured ham non-destructively. For that, fresh and dry-cured hams were scanned and then the MRI images were analyzed by three methods of computational texture features. Data mining was applied on the computational texture features from fresh and dry-cured hams for obtaining prediction equations of the sensory attributes of dry-cured hams. The correlation coefficient (R) was used to analyze the results. Accurate prediction was found for 13 sensory attributes as a function of computational texture features of fresh ham, and three from dry-cured ham. In addition, a sensory analysis of dry-cured hams was also carried out to validate the predicted results. Similar values were found between the predicted attributes and those determined by sensory analysis. Thus, it is possible to predict sensory attributes of dry-cured hams by applying data mining on computational texture features of MRI from fresh and dry-cured hams. This supposes the chance of determining non-destructively sensory attributes of dry-cured hams, even before the curing process starts.

Aesthetic perception of visual textures: a holistic exploration using texture analysis, psychological experiment, and perception modeling

Modeling human aesthetic perception of visual textures is important and valuable in numerous industrial domains, such as product design, architectural design, and decoration. Based on results from a semantic differential rating experiment, we modeled the relationship between low-level basic texture features and aesthetic properties involved in human aesthetic texture perception. First, we compute basic texture features from textural images using four classical methods. These features are neutral, objective, and independent of the socio-cultural context of the visual textures. Then, we conduct a semantic differential rating experiment to collect from evaluators their aesthetic perceptions of selected textural stimuli. In semantic differential rating experiment, eights pairs of aesthetic properties are chosen, which are strongly related to the socio-cultural context of the selected textures and to human emotions. They are easily understood and connected to everyday life. We propose a hierarchical feed-forward layer model of aesthetic texture perception and assign 8 pairs of aesthetic properties to different layers. Finally, we describe the generation of multiple linear and non-linear regression models for aesthetic prediction by taking dimensionality-reduced texture features and aesthetic properties of visual textures as dependent and independent variables, respectively. Our experimental results indicate that the relationships between each layer and its neighbors in the hierarchical feed-forward layer model of aesthetic texture perception can be fitted well by linear functions, and the models thus generated can successfully bridge the gap between computational texture features and aesthetic texture properties.

NIMG-67COMPUTER EXTRACTED TEXTURE DESCRIPTORS FROM DIFFERENT TISSUE COMPARTMENTS WITHIN THE TUMOR HABITAT ON TREATMENT-NAÏVE MRI PREDICT CLINICAL SURVIVAL IN GLIOBLASTOMA PATIENTS

PURPOSE: Glioblastoma Multiforme (GBM) is a highly aggressive and heterogeneous brain tumor. Currently all GBM patients are given “one-dose-fit-all” treatment. Identification of prognostic markers can allow for personalized therapy options for GBM. The underlying hypothesis of this study is that the heterogeneity in GBM (due to subtle variations in tumor enhancement, cellular density, necrosis that are not visually appreciable on volumetric analysis of MRI) is prognostic and can be captured using computerized texture descriptors extracted from within different tumor compartments (enhancing tumor, necrotic core, edema). These compartments together define tumor “habitat”, and computerized texture features from within the habitat can be predictive of short-term (STS) ( 24-months). METHODS: A total of 58 3Tesla MRI studies (27 STS, 35 LTS) with GdT1C, FLAIR, and T2w protocols were obtained from the TCIA repository. Enhancing tumor, and necrotic regions on Gd-T1 and edematous region jointly on T2w and FLAIR, were segmented by an expert using Slicer 3D. 135 2-D texture descriptors, and volumetric measurements were extracted on a per-voxel basis from each of the 3 sub-regions (tumor, necrosis, edema) on every MRI protocol. Feature selection was used to identify most discriminative features with a random forest classifier trained via 3-fold cross validation. Kaplan-Meier (KM) survival curves were used for analysis, with correction for multiple hypothesis testing to identify features that were significantly (p < 0.05) correlated with survival. RESULTS: Contributions from top 15 texture features within the tumor habitat, (edema, necrotic core, enhanced tumor), were together most significantly associated with survival, across Gd-T1-C (p = 0.003), FLAIR (p = 0.006), and T2w (p = 0.04) as compared to individual features, and volumetric measurements from the tumor habitat. CONCLUSION: Computerized texture features when jointly interrogated across compartments within the tumor habitat appear more prognostic of clinical survival in GBMs than features from enhancing tumor and tumor volume alone.

Association of computerized texture features on MRI with early treatment response following laser ablation for neuropathic cancer pain: preliminary findings.

Laser interstitial thermal therapy (LITT) has recently emerged as a new treatment modality for cancer pain management that targets the cingulum (pain center in the brain) and has shown promise over radio frequency (RF)-based ablation, due to magnetic resonance image (MRI) guidance that allows for precise ablation. Since laser ablation for pain management is currently exploratory and is only performed at a few centers worldwide, its short- and long-term effects on the cingulum are currently unknown. Traditionally, treatment effects for neurological conditions are evaluated by monitoring changes in intensities and/or volume of the ablation zone on post-treatment Gadolinium-contrast T1-w (Gd-T1) MRI. However, LITT introduces subtle localized changes corresponding to tissues response to treatment, which may not be appreciable on visual inspection of volumetric or intensity changes. Additionally, different MRI protocols [Gd-T1, T2w, gradient echo sequence (GRE), fluid-attenuated inversion recovery (FLAIR)] are known to capture complementary diagnostic information regarding the patient's response to treatment; the utility of these MRI protocols has so far not been investigated to evaluate early and localized response to LITT treatment in the context of neuropathic cancer pain. In this work, we present the first attempt at (a)examining early treatment-related changes on a per-voxel basis via quantitative comparison of computer-extracted texture descriptors across pre- and post-LITT multiparametric (MP-MRI) (Gd-T1, T2w, GRE, FLAIR), subtle microarchitectural texture changes that may not be appreciable on original MR intensities or volumetric differences, and (b)investigating the efficacy of different MRI protocols in accurately capturing immediate post-treatment changes reflected (1)within and (2)outside the ablation zone. A retrospective cohort of four patient studies comprising pre- and immediate (24h) post-LITT 3 Tesla Gd-T1, T2w, GRE, and FLAIR acquisitions was considered. Our quantitative approach first involved intensity standardization to allow for grayscale MR intensities acquired pre- and post-LITT to have a fixed tissue-specific meaning within the same imaging protocol, the same body region, and within the same patient. An affine registration was then performed on individual post-LITT MRI protocols to a reference MRI protocol pre-LITT. A total of 78 computerized texture features (co-occurrence matrix homogeneity, neighboring gray-level dependence matrix, Gabor) are then extracted from pre- and post-LITT MP-MRI on a per-voxel basis. Quantitative, voxelwise comparison of the changes in MRI texture features between pre- and post-LITT MRI indicate that (a)Gabor texture features at specific orientations were highly sensitive as well as specific in predicting subtle microarchitectural changes within and around the ablation zone pre- and post-LITT, (b)FLAIR was identified as the most sensitive MRI protocol in identifying early treatment changes yielding a normalized percentage change of 360% within the ablation zone relative to its pre-LITT value, and (c)GRE was identified as the most sensitive MRI protocol in quantifying changes outside the ablation zone post-LITT. Our preliminary results thus indicate potential for noninvasive computerized MP-MRI features over volumetric features in determining localized microarchitectural early focal treatment changes post-LITT for neuropathic cancer pain treatment.

TU‐G‐204‐05: The Effects of CT Acquisition and Reconstruction Conditions On Computed Texture Feature Values of Lung Lesions

Purpose:Texture features have been investigated as a biomarker of response and malignancy. Because these features reflect local differences in density, they may be influenced by acquisition and reconstruction parameters. The purpose of this study was to investigate the effects of radiation dose level and reconstruction method on features derived from lung lesions.Methods:With IRB approval, 33 lung tumor cases were identified from clinically indicated thoracic CT scans in which the raw projection (sinogram) data were available. Based on a previously‐published technique, noise was added to the raw data to simulate reduced‐dose versions of each case at 25%, 10% and 3% of the original dose. Original and simulated reduced dose projection data were reconstructed with conventional and two iterative‐reconstruction settings, yielding 12 combinations of dose/recon conditions. One lesion from each case was contoured. At the reference condition (full dose, conventional recon), 17 lesions were randomly selected for repeat contouring (repeatability). For each lesion at each dose/recon condition, 151 texture measures were calculated. A paired differences approach was employed to compare feature variation from repeat contours at the reference condition to the variation observed in other dose/recon conditions (reproducibility). The ratio of standard deviation of the reproducibility to repeatability was used as the variation measure for each feature.Results:The mean variation (standard deviation) across dose levels and kernel was significantly different with a ratio of 2.24 (±5.85) across texture features (p=0.01). The mean variation (standard deviation) across dose levels with conventional recon was also significantly different with 2.30 (7.11) (p=0.025). The mean variation across reconstruction settings of original dose has a trend in showing difference with 1.35 (2.60) among all features (p=0.09).Conclusion:Texture features varied considerably with variations in dose and reconstruction condition. Care should be taken to standardize these conditions when using texture as a quantitative feature.This effort supported in part by a grant from the National Cancer Institute's Quantitative Imaging Network (QIN): U01 CA181156; The UCLA Department of Radiology has a Master Research Agreement with Siemens Healthcare; Dr. McNitt‐Gray has previously received research support from Siemens Healthcare.

Could linear model bridge the gap between low-level statistical features and aesthetic emotions of visual textures?

Aesthetic perceptions are common in everyday life, even though we are not aware of these natural behavior and ability. The exploration of the processing pathways in human aesthetic perception of visual textures by modeling is vital and valuable in important domains such as product design and architecture. After conducting semantic differential rating experiments, we model the relationship between low-level texture features and aesthetic properties involved in human aesthetic texture perception. In this paper, we use multiple linear regression models to bridge the gap between computation texture features and aesthetic emotions that are more interpretable and clearer in structure than artificial neural network models. In feature extraction stage, four different categories of low level features are used to objectively represent the characteristics of visual textures. To reduce the information redundancy of the feature set and the complexity of the prediction model, the dimensionality reduction of the extracted feature set using stochastic neighbor embedding is implanted. In psychological experiment stage, the semantic differential rating experiment is conducted to collect the aesthetic perceptions of selected texture stimuli from participants, and then the aesthetic properties are assigned to three different model layers based on the neural mechanisms of aesthetic perception. Before model building, an evaluation of the presentation ability of feature subsets is firstly carried out to investigate whether there is a potential relationship between 8 high level aesthetic properties and the feature set including 106 texture features. Next, we describe the generation of a multiple linear regression model for aesthetic prediction by taking a reduced feature set as dependent variables. Experimental results indicate that the hierarchical feed-forward layer model of aesthetic texture perception proposed in our research can link computational texture features with aesthetic texture properties quite well.

Computerized texture analysis of persistent part-solid ground-glass nodules: differentiation of preinvasive lesions from invasive pulmonary adenocarcinomas.

To retrospectively investigate the value of computerized three-dimensional texture analysis for differentiation of preinvasive lesions from invasive pulmonary adenocarcinomas (IPAs) that manifest as part-solid ground-glass nodules (GGNs). The institutional review board approved this retrospective study with a waiver of patients' informed consent. The study consisted of 86 patients with 86 pathologic analysis-confirmed part-solid GGNs (mean size, 16 mm ± 5.4 [standard deviation]) who had undergone computed tomographic (CT) imaging between January 2005 and October 2011. Each part-solid GGN was manually segmented and its computerized texture features were quantitatively extracted by using an in-house software program. Multivariate logistic regression analysis was performed to investigate the differentiating factors of preinvasive lesions from IPAs. Three-layered artificial neural networks (ANNs) with a back-propagation algorithm and receiver operating characteristic curve analysis were used to build a discriminating model with texture features and to evaluate its discriminating performance. Pathologic analysis confirmed 58 IPAs (seven minimally invasive adenocarcinomas and 51 invasive adenocarcinomas) and 28 preinvasive lesions (four atypical adenomatous hyperplasias and 24 adenocarcinomas in situ). IPAs and preinvasive lesions exhibited significant differences in various histograms and volumetric parameters (P < .05). Multivariate analysis revealed that smaller mass (adjusted odds ratio, 0.092) and higher kurtosis (adjusted odds ratio, 3.319) are significant differentiators of preinvasive lesions from IPAs (P < .05). With mean attenuation, standard deviation of attenuation, mass, kurtosis, and entropy, the ANNs model showed excellent accuracy in differentiation of preinvasive lesions from IPAs (area under the curve, 0.981). In part-solid GGNs, higher kurtosis and smaller mass are significant differentiators of preinvasive lesions from IPAs, and preinvasive lesions can be accurately differentiated from IPAs by using computerized texture analysis. Online supplemental material is available for this article.

Abstract P2-02-12: Computer derived image features on DCE-MRI appear to distinguish estrogen receptor-positive breast cancers with low and high oncotype DX recurrence scores

Abstract The Oncotype DX (ODX) is a 21 panel gene-expression based assay for identifying which Estrogen Receptor-positive (ER+) breast cancer (BCa) patients are candidates for adjuvant chemotherapy. The objective of this research was to identify whether computerized texture features on a staging DCE-MRI can distinguish ER+ BCa with low and high ODX recurrence scores (RS) (i.e. to distinguish which ER+ BCa patients are more likely to benefit from adjuvant hormonal therapy from those who require chemotherapy). This would provide a non-invasive, imaging based, pre-therapeutic assessment tool for predicting the appropriate treatment regimen. This work, to the best of our knowledge, is the first attempt to quantitatively correlate low versus high risk stratification via computer derived MRI measurements to corresponding risk stratification via the ODX assay. 52 ER+ BCa patient studies with high (&amp;gt;30, N = 28) and low (&amp;lt;18, N = 24) ODX RS were available for this study from two sites; 16 breast MRIs from the Boston Medical Center using a Phillips 1.5T magnet with a 7-channel breast coil, and 36 MRIs from the Case Medical Center using a Siemens 1.5T magnet with a 8-channel breast coil. All datasets included T1w images obtained prior to, during, and after administration of 0.1 mmol/kg of Gd-DTPA and corresponding ODX RS. For each study a radiologist picked a representative slice showing the tumor and then manually segmented the region of interest (ROI) containing the lesion. Computerized image analysis tools developed in-house via the MATLAB© programming platform were applied to the manually segmented lesion ROI for each of the 52 MRI studies to quantitatively characterize the lesion via a set of (a) 6 shape, (b) 3 pharmacokinetic (Ktrans, ve, kep) based on Tofts model (PK), (c) 12 enhancement kinetic (EK), (d) 12 intensity kinetic (IK), (e) 312 textural kinetic (TK), (f) 6 dynamic local binary pattern (DLBP), and (g) 5 dynamic histogram of oriented gradient (DHoG) features. The computer extracted features were evaluated via a linear discriminant analysis (LDA) classifier in terms of their ability to distinguish ER+ BCa as having a low or high ODX RS via a 2-fold randomized cross validation scheme. At each iteration, half of the studies were randomly selected from the 52 cases and used for training the LDA classifier and the remaining 26 studies were used for independent testing. This process was repeated 200 times. Classification performance was evaluated by area under the ROC curve (AUC). Higher AUC values suggest a stronger relationship between risk stratification via MRI attributes and ODX. Table 1Feature classAccuracy (μ±Δ)AUC (μ±Δ)DHoG87.07%±5.66%0.89±0.04DLBP85.86%±7.82%0.83±0.07EK82.36%±8.46%0.80±0.06PK81.14%±7.55%0.78±0.07TK75.93%±6.65%0.76±0.08IK76.43%±7.23%0.75±0.12Shape71.04%±6.81%0.70±0.06 Table 1 illustrates the mean and standard deviation in accuracy and AUC values over 200 runs of randomized cross validation. DHoG, DBLP and EK features yielded the highest classification accuracy and AUC. Although lesion shape has been shown to be important for discriminating benign and malignant lesions on MRI, shape appears to be less useful in distinguishing between ER+ BCa lesions with low and high ODX RS. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-02-12.