Machine vision 3D skin texture analysis for detection of melanoma

To distinguish malignant pulmonary nodules from benign ones is of much importance in computer-aided diagnosis of lung diseases. Compared to many previous methods which are based on shape or growth assessing of nodules, this proposed three-dimensional (3D) texture feature based approach extracted fifty kinds of 3D textural features from gray level, gradient and curvature co-occurrence matrix, and more derivatives of the volume data of the nodules. To evaluate the presented approach, the Lung Image Database Consortium public database was downloaded. Each case of the database contains an annotation file, which indicates the diagnosis results from up to four radiologists. In order to relieve partial-volume effect, interpolation process was carried out to those volume data with image slice thickness more than 1mm, and thus we had categorized the downloaded datasets to five groups to validate the proposed approach, one group of thickness less than 1mm, two types of thickness range from 1mm to 1.25mm and greater than 1.25mm (each type contains two groups, one with interpolation and the other without). Since support vector machine is based on statistical learning theory and aims to learn for predicting future data, so it was chosen as the classifier to perform the differentiation task. The measure on the performance was based on the area under the curve (AUC) of Receiver Operating Characteristics. From 284 nodules (122 malignant and 162 benign ones), the validation experiments reported a mean of 0.9051 and standard deviation of 0.0397 for the AUC value on average over 100 randomizations.

The mainstream CNN-based remote sensing (RS) image semantic segmentation approaches typically rely on massively labeled training data. Such a paradigm struggles with the problem of RS multi-view scene segmentation with limited labeled views due to the lack of consideration of 3D information within the scene. In this paper, we propose “Implicit Ray-Transformer (IRT)” based on Implicit Neural Representation (INR) for RS scene semantic segmentation with sparse labels (5% of the images being labeled). We explore a new way of introducing the multi-view 3D structure priors to the task for accurate and view-consistent semantic segmentation. The proposed method includes a two-stage learning process. In the first stage, we optimize a neural field to encode the color and 3D structure of the remote sensing scene based on multi-view images. In the second stage, we design a Ray Transformer to leverage the relations between the neural field 3D features and 2D texture features for learning better semantic representations. Different from previous methods that only consider 3D priors or 2D features, we incorporate additional 2D texture information and 3D priors by broadcasting CNN features to different point features along the sampled ray. To verify the effectiveness of the proposed method, we construct a challenging dataset containing six synthetic sub-datasets collected from the Carla platform and three real sub-datasets from Google Maps. Experiments show that the proposed method outperforms the CNN-based methods and the state-of-the-art INR-based segmentation methods in quantitative and qualitative metrics. The ablation study shows that under a limited number of fully annotated images, the combination of the 3D structure priors and 2D texture can significantly improve the performance and effectively complete missing semantic information in novel views. Experiments also demonstrate the proposed method could yield geometry-consistent segmentation results against illumination changes and viewpoint changes. Our data and code will be public.

The purpose of the study was to determine if CT texture features of untreated hepatic metastatic colorectal cancer (CRC) relate to pathologic features and clinical outcomes. Tumor texture analysis was performed on single hepatic metastatic lesions on pre-treatment contrast-enhanced CT scans in 77 pts (mean age 58, 34F/43M) using a novel tool. Measures of heterogeneity, including entropy, kurtosis, skewness, mean, mean positive pixels (MPP), and standard deviation (SD) of pixel distribution histogram were derived with filter values corresponding to fine (spatial scaling factor (ssf) 2), medium (ssf 3, 4), and coarse textures (ssf 5, 6). Texture parameters were correlated with tumor grade, baseline serum CEA, and KRAS mutation status. Overall survival was also correlated using Cox proportional hazards models. Single-slice 2D vs. whole-tumor volumetric 3D texture analysis was compared in a subcohort of 20 patients. Entropy, MPP, and SD at medium filtration levels were significantly associated with tumor grade (MPP ssf 3 P=0.002, SD ssf 3 P=0.004, entropy ssf 4 P=0.007). Skewness was negatively associated KRAS mutation (P=0.02). Entropy at coarse filtration levels was associated with survival (Hazard ratio (HR) for death 0.65, 95% CI 0.44-0.95, P=0.03). Texture results for 2D and 3D analysis were similar. CT texture features, particularly entropy, MPP, and SD, are significantly associated with tumor grade in untreated CRC liver metastases. Tumor entropy at coarse filters correlates with overall survival. Single-slice 2D texture analysis appears to be adequate.

3D solid textures are important data for computer graphics applications because they are useful for synthesizing photo realistic patterns such as stone, smoke, water, clouds, fire, and wood. Unlike typical 2D textures, 3D solid textures are represented in voxel forms, which use three-dimensional coordinates. Various kinds of research have been conducted for pattern analysis of 2D textures to help development of computer applications. However, pattern analysis of 3D solid textures has not been investigated sufficiently even though 3D data related software is emerging. In this research, we have extended higher order local autocorrelation (HLAC) 2D mask patterns to 3D mask patterns so that pattern analysis software can handle not only 2D texture images but also 3D solid textures. Our experimental system can classify and look for 3D solid textures with similar patterns from a 3D solid texture database by using 3D HLAC masks. We have examined (1) similarity retrieval of 3D solid textures and (2) pattern classification of 3D solid textures by using the system.

Lately, 3D imaging techniques have achieved a lot of progress due to recent developments in 3D sensor technologies. This leads to a great interest regarding 3D image feature extraction and classification techniques. As pointed out in literature, one of the most important and discriminative features in images is the textural content. Within this context, we propose a texture feature extraction technique for volumetric images with improved discrimination power. The method could be used in textured volumetric data classification tasks. To achieve this, we fuse two complementary pieces of information, feature vectors derived from Local Binary Patterns (LBP) and the Gray-Level Co-occurrence Matrix-based methods. They provide information regarding the image pattern and the contrast, homogeneity and local anisotropy in the volumetric data, respectively. The performance of the proposed technique was evaluated on a public dataset consisting of volumetric textured images affected by several transformations. The classifiers used are the Support Vector Machine, k-Nearest Neighbours and Random Forest. Our method outperforms other handcrafted 3D or 2D texture feature extraction methods and typical deep-learning networks. The proposed technique improves the discrimination power and achieves promising results even if the number of images per class is relatively small.

In this study emotion-based face expression recognition framework has been proposed using a machine vision (MV) approach. The face emotion dataset has been collected local survey in Bahawalpur city dataset divide into 3 classes happy, sad, and angry. A total of 600 images of size (256 x 256) were transformed into a gray level format and employed a median filter for noise removal. Three non-overlapping regions of interest (ROIs) of size (50 x 50) have been taken and analyze 1800 (600 x 3) ROIs on the overall dataset. Total 45 Statistical features named as texture, histogram, and binary features were extracted. Select optimize features using the correlation-based feature section technique. The optimized dataset employed of MV classifiers namely random forest (RF), logistic (Lg), and J48 are obtained very promising accuracy 96.33%, 95.67%, and 95.33% respectively.

This paper suggests 3D co-occurrence texture features by extending the concept of co-occurrence feature to the 3D world. The suggested 3D features are described as a 3D co-occurrence matrix by using a co-occurrence histogram of digital elevations at two contiguous positions. With the addition of 3D co-occurrence features, we encounter the high dimensionality problem in the classification process. Since these ANN (Artificial Neural Networks) clustering algorithms are known as robust in this situation, FCM (Fuzzy C-mean) and GBFCM (Gradient Based Fuzzy C-mean) clustering algorithms are employed to implement the terrain classifier. Experimental results show that the classification accuracy with the addition of 3D co-occurrence features is significantly improved over the conventional classification method only with 2D features.

PD08-03 3D TEXTURE ANALYSIS IN ABDOMINAL CT AIDED BY MACHINE LEARNING CLASSIFIERS PREDICTS SHOCK WAVE LITHOTRIPSY SUCCESS

This paper presents a new feature extraction technique for 3D texture classification based on Local Binary Patterns-derived operators and state-of-the-art denoising techniques, with good invariance properties to different transformations and robust to noise. Our approach consists in the computation of textural features obtained using the Block Matching and 3D Filtering Extended Local Binary Patterns for each of the three orthogonal planes and their concatenation. The proposed method achieves a promising classification performance on a publicly available dataset of 3D textured images. The obtained results prove that the presented feature extraction technique is above traditional 2D texture descriptors and typical CNN deep learning-based methods.

Collecting pavement texture information is crucial to understand the characteristics of a road surface and to have essential data to support road maintenance. Traditional texture assessment techniques often require expensive equipment and complex operations. To ensure cost sustainability and reduce traffic closure times, this study proposes a rapid, cost-effective, and non-invasive surface texture assessment technique. This technology consists of capturing a set of images of a road surface with a mobile phone; then, the images are used to reconstruct the 3D surface with photogrammetric processing and derive the roughness parameters to assess the pavement texture. The results indicate that pavement images taken by a mobile phone can reconstruct the 3D surface and extract texture features with accuracy, meeting the requirements of a time-effective documentation. To validate the effectiveness of this technique, the surface structure of the pavement was analyzed in situ using a 3D structured light projection scanner and rigorous photogrammetry with a high-end reflex camera. The results demonstrated that increasing the point cloud density can enhance the detail level of the real surface 3D representation, but it leads to variations in road surface roughness parameters. Therefore, appropriate density should be chosen when performing three-dimensional reconstruction using mobile phone images. Mobile phone photogrammetry technology performs well in detecting shallow road surface textures but has certain limitations in capturing deeper textures. The texture parameters and the Abbott curve obtained using all three methods are comparable and fall within the same range of acceptability. This finding demonstrates the feasibility of using a mobile phone for pavement texture assessments with appropriate settings.

Bone architectural information can be derived from structural indices calculated from high resolution magnetic resonance (MR) images. However, high resolution scans are time consuming and prone to motion artifacts and hence are not routinely feasible. The purpose of this study is to determine if a correlation exists between 3D structural indices calculated from high-resolution MR images and 3D co-occurrence texture features calculated from lower resolution MR images. Regression analysis indicates a strong correlation between the structural indices and the texture features. This study highlights the potential of using surrogate texture markers extracted from readily acquired clinical MR images to quantify bone architecture, circumventing the need for high resolution MR imaging.

Iris presentation attack detection (PAD) is still an unsolved problem mainly due to the various spoof attack strategies and poor generalisation on unseen attackers. In this paper, the merits of both light field (LF) imaging and deep learning (DL) are leveraged to combine 2D texture and 3D geometry features for iris liveness detection. By exploring off‐the‐shelf deep features of planar‐oriented and sequence‐oriented deep neural networks (DNNs) on the rendered focal stack, the proposed framework excavates the differences in 3D geometric structure and 2D spatial texture between bona fide and spoofing irises captured by LF cameras. A group of pre‐trained DL models are adopted as feature extractor and the parameters of SVM classifiers are optimised on a limited number of samples. Moreover, two‐branch feature fusion further strengthens the framework's robustness and reliability against severe motion blur, noise, and other degradation factors. The results of comparative experiments indicate that variants of the proposed framework significantly surpass the PAD methods that take 2D planar images or LF focal stack as input, even recent state‐of‐the‐art (SOTA) methods fined‐tuned on the adopted database. Presentation attacks, including printed papers, printed photos, and electronic displays, can be accurately detected without fine‐tuning a bulky CNN. In addition, ablation studies validate the effectiveness of fusing geometric structure and spatial texture features. The results of multi‐class attack detection experiments also verify the good generalisation ability of the proposed framework on unseen presentation attacks.

Objective: Susceptibility-weighted imaging (SWI) has been used to identify neurodegeneration in amyotrophic lateral sclerosis (ALS) through qualitative gross visual comparison of signal intensity. The aim of this study was to quantitatively identify cerebral degeneration in ALS on SWI using texture analysis. Methods: SW images were acquired from 17 ALS patients (58.4 ± 10.3 years, 13M/4F, ALSFRS-R 41.2 ± 4.1) and 18 healthy controls (56.3 ± 17.6 years, 9M/9F) at 4.7 tesla. Textures were computed within the precentral gyrus and basal ganglia and compared between patients and controls using ANCOVA with age and gender as covariates. Texture features were correlated with clinical measures in patients. Texture features found to be significantly different between patients and controls in the precentral gyrus were then used in a whole-brain 3D texture analysis. Results: The texture feature autocorrelation was significantly higher in ALS patients compared to healthy controls in the precentral gyrus and basal ganglia (p < 0.05). Autocorrelation correlated significantly with clinical measures such as disease progression rate and finger tapping speed (p < 0.05). Whole brain 3D texture analysis using autocorrelation revealed differences between ALS patients and controls within the precentral gyrus on SWI images (p < 0.001). Conclusion: Texture analysis on SWI can quantitatively identify cerebral differences between ALS patients and controls.

Local fractal and multifractal features for volumic texture characterization

409P - 3D based texture analysis serving as potential diagnostic factor in discriminating primary central nervous system lymphoma from metastatic brain tumours: A preliminary study