Skin Lesion Classification Based on Surface Fractal Dimensions and Statistical Color Cluster Features Using an Ensemble of Machine Learning Techniques

(1) Background: An approach for skin cancer recognition and classification by implementation of a novel combination of features and two classifiers, as an auxiliary diagnostic method, is proposed. (2) Methods: The predictions are made by k-nearest neighbor with a 5-fold cross validation algorithm and a neural network model to assist dermatologists in the diagnosis of cancerous skin lesions. As a main contribution, this work proposes a descriptor that combines skin surface fractal dimension and relevant color area features for skin lesion classification purposes. The surface fractal dimension is computed using a 2D generalization of Higuchi’s method. A clustering method allows for the selection of the relevant color distribution in skin lesion images by determining the average percentage of color areas within the nevi and melanoma lesion areas. In a classification stage, the Higuchi fractal dimensions (HFDs) and the color features are classified, separately, using a kNN-CV algorithm. In addition, these features are prototypes for a Radial basis function neural network (RBFNN) classifier. The efficiency of our algorithms was verified by utilizing images belonging to the 7-Point, Med-Node, and PH2 databases; (3) Results: Experimental results show that the accuracy of the proposed RBFNN model in skin cancer classification is 95.42% for 7-Point, 94.71% for Med-Node, and 94.88% for PH2, which are all significantly better than that of the kNN algorithm. (4) Conclusions: 2D Higuchi’s surface fractal features have not been previously used for skin lesion classification purpose. We used fractal features further correlated to color features to create a RBFNN classifier that provides high accuracies of classification.

Particulate Formation During High Explosive Detonations: How Oxygen Balance Drives Parameters Pertinent to Atmospheric Lofting

High explosive (HE) detonations reach pressures and temperatures that extend beyond normal environmental conditions, thereby permitting access to various carbon and metal allotropes of different morphologies, sizes and surface structures. The products of HE detonations are dependent on multiple parameters, including the chemical and physical properties of the starting material and atmospheric conditions (i.e. oxygen). One important factor is the HE oxygen balance, which is the extent to which the material can be oxidized. Insensitive HEs are designed to resist external stimuli that would cause detonation in conventional HEs. The insensitive HEs are negatively oxygen balanced and therefore produce not only gaseous species but solid carbon products during detonation. Insensitive HEs were studied, Composition B-3 and PBX 9501, with steady and overdriven geometries in an oxygen-free atmosphere that reached different pressure and temperature regimes. Small angle x-ray scattering provided the size and surface structure of the resulting particulates. Composition B-3 primary particles were 157.0 ± 0.3 Å and 199.5 ± 0.3 Å for steady and overdriven detonations; where PBX 9501 primary particles were larger than Composition B-3 at 300 ± 6 Å and 334.5 ± 0.3 Å for steady and overdriven detonations. The two compounds formed contrasting primary particles with different cluster structures, in the Composition B-3 steady detonation the particles were agglomerated into a surface fractal with rough surfaces where as the PBX 9501 was a mass fractal cluster with smooth surface primary particles. In the overdriven detonation the primary particles were reversed, Composition B-3 was agglomerated into a mass fractal structure with smooth surfaces and PBX 9501 had a surface fractal structure with a rough surface primary particles. Scanning electron microscopy provided a snapshot of the morphology of the materials on the micron length scale, supporting the observation of x-ray scattering that the Composition B-3 particulates/agglomerates are smaller than the PBX 9501. Raman spectroscopy provided information as to the carbon bonding of the detonation soot, showing significantly more product variation in Composition B-3 than PBX 9501, likely due the poor oxygen balance of Composition B-3 leading to more complex carbon bonding formations. Finally, x-ray photoelectron spectroscopy showed how the difference in the oxygen balance of the HE fuel directly relates to the amount of carbon-oxygen bonding that is present in the final products, where PBX 9501 had significantly more oxygen on the surface of the particulates. We used two HEs to understand the detonation pathways for both synthesis and atmospheric processes; where the chemical constituents of the particulates can promote processes such as self-lofting and aerosol-cloud interactions after the particles are launched into the troposphere or stratosphere during detonation.

Study on the Influence of Three Factors on Mass Loss and Surface Fractal Dimension of Concrete in Sulfuric Acid Environments

When exposed to sulfuric acid environments, the service life of concrete structures would be reduced due to the high alkalinity of concrete. The influence of three factors including water/cement ratio, the pH value of the solution, and the chemical composition of the aggregate on the resistance of concrete subjected to sulfuric acid has been widely investigated by previous researchers. This paper aims to investigate the influence of these three factors on the durability evaluation indicators including mass loss and surface fractal dimension through orthogonal experiments, which has been reported rarely in previous research. Four combinations of coarse and fine aggregate including gravel and river sand, gravel and crushed marble sand, crushed marble stone and river sand, and crushed marble stone and marble sand were adopted, and three water/cement ratios including 0.35, 0.45, and 0.55 were selected, and the sulfuric acid solution pH values 0.95, 2, and 4 were chosen in this paper. The results showed that the larger the water/cement ratio, the smaller the mass loss and the surface fractal dimension of the specimens, and with the decrease of the pH value of the sulfuric acid solution, the mass loss and the surface fractal dimension of the specimens would be increased. The concrete specimen containing gravel and river sand had the greatest surface fractal dimension and greatest mass loss, while the concrete specimen containing crushed marble sand had a smaller surface fractal dimension and a smaller mass loss. The dominant and secondary order of three factors on mass loss and surface fractal dimension of concrete subjected to sulfuric acid was the pH value of the solution > the chemical composition of the aggregate > the water/cement ratio.

Structure and Strength of Artificial Soils Containing Monomineral Clay Fractions

Structure and strength are responsible for soil physical properties. This paper determines in a uniaxial compression test the strength of artificial soils containing different proportions of various clay-size minerals (cementing agents) and silt-size feldspar/quartz (skeletal particles). A novel empirical model relating the maximum stress and the Young’s modulus to the mineral content basing on the Langmuir-type curve was proposed. By using mercury intrusion porosimetry (MIP), bulk density (BD), and scanning electron microscopy (SEM), structural parameters influencing the strength of the soils were estimated and related to mechanical parameters. Size and shape of particles are considered as primary factors responsible for soil strength. In our experiments, the soil strength depended primarily on the location of fine particles in respect to silt grains and then, on a mineral particle size. The surface fractal dimension of mineral particles played a role of a shape parameter governing soil strength. Soils containing minerals of higher surface fractal dimensions (rougher surfaces) were more mechanically resistant. The two latter findings appear to be recognized herein for the first time.

Spectral and Texture Properties of Hydrophobic Aerogel Powders Obtained from Room Temperature Drying

Attenuated Total Reflectance Infrared (ATR-IR) spectroscopy and texture measurements based on nitrogen adsorption-desorption isotherms are combined to characterize silica aerogel granules with different degrees of hydrophobicity. The aerogels were prepared from tetraethoxysilane via a room temperature hydrolysis-gelation process, solvent exchange, hydrophobization, and drying at subcritical conditions. The dependencies between the texture properties, pore architectures, surface fractal dimensions, and degree of hydrophobicity of the samples are extracted from the ATR-IR spectra and the adsorption-desorption isotherms. The IR absorption in the region of the Si-O-Si and Si-OH vibrations is used for a description of the structural and chemical changes in aerogel powders connected with their surface hydrophobization. The Frenkel–Halsey–Hill (FHH) theory is applied to determine the surface fractal dimension of the powder species.

Experimental Investigation of the Matrix Pore Size Distribution and Inner Surface Fractal Dimension of Different-Structure High Rank Coals

To investigate the nanopore characteristics of different-structure high-rank coal matrices, four samples were collected from the Zhaozhuang Mine, a typical mine highly prone to coal and gas outbursts in Shanxi Province, China, and the nanopore size distribution was measured using a lowtemperature nitrogen adsorption method for these four samples. Based on the nitrogen adsorption isotherms, the inner surface fractal dimensions of micropores, meso- and macropores, and full-scale matrix pores were estimated using a Frenkel–Halsey–Hill (FFH) fractal model, and the relationships between the fractal dimensions and pore parameters were discussed. The inner surface of the high-rank coal matrix is heterogeneous. The inner surface fractal dimensions of micropores (D1) and full-scale matrix pores (DT) can be arranged in a descending sequence for the differentstructure coals: mylonitized coal > granulated coal > cataclastic coal > intact coal, while the inner surface fractal dimensions of meso- and macropores (D2) are in an ascending sequence. With the increasing deformation degree of coals, some macropores and mesopores are transformed into smaller pores, such as micropores; the total pore volume (PV) and total specific surface area (SSA) increase; the SSA percentage and PV percentage of meso- and micropores increase; and those of macropores decrease. The average pore width (APW) is positively correlated with D2 and negatively correlated with D1 and DT. The tectonic deformation enhances the irregularity of micropores and the heterogeneity of the whole coal matrix pores and decreases the inner surface roughness of meso- and macropores.

Experimental Study on Fragmentation Behavior of Specimen with Rockbolts

The effect of rockbolt density on fractured specimens and the distribution law of fragments were studied using the fractal method. The results show that with the increase of rockbolt density, the strength of specimens is improved. Its degree of fragmentation is increased, but the average size of fragments is decreased. The fragmentation and surface fractal dimensions are increased, their difference value is 1 in theory, but the value is slightly less than 1 in practical application due to the uncertain factors such as anisotropy of specimen. There is a positive correlation between the deformation of surrounding rock and its fractal dimension. The surface fractal dimension of surrounding rock can give a quantitative description of its degree of fragmentation and can give a comprehensive reflection to the roof stability. The support mechanism of rockblots can be understood as improving the probability of refragmentation of fragments, making its distribution gradation close to the optimal one with minimum porosity, thus reducing the deformation of surrounding rock.