RESUMO
Size Functions and Support Vector Machines are used to implement a new automatic classifier of melanocytic lesions. This is mainly based on a qualitative assessment of asymmetry, performed by halving images by several lines through the center of mass, and comparing the two halves in terms of color, mass distribution, and boundary. The program is used, at clinical level, with two thresholds, so that comparison of the two outputs produces a report of low-middle-high risk. Experimental results on 977 images, with cross-validation, are reported.
RESUMO
Mathematical morphology is a tool for extracting image components that are useful for representation and description. The technique consists of a set-theoretic method of image analysis providing a quantitative description of geometrical structures. A simple application of mathematical morphology to a bi-dimensional processing of TEM images of airborne particles allows us to distinguish between particles grown and/or transported in atmosphere under dry conditions or in rainy days by a simple comparison of the corresponding image form factors. The form factors range in the 0.385-0.031 interval in the case of particles sampled in rainy days, and in the 0.103-0.006 interval in the case of non-rainy conditions. The same classification criterion was applied to filters collected under dry conditions and plunged in water. The results demonstrate that a morphological change may be artificially induced to the particle structure. The artificially wet particles, indeed, display an apparent contraction of their structures evidenced by a two-fold increase of the average values of their form factors. The last experiment roughly simulates the impact of particles on membranes of the respiratory tract.
Assuntos
Material Particulado/análise , Fuligem/análise , Água , Computação Matemática , Microscopia Eletrônica de Transmissão , Modelos Teóricos , Tamanho da Partícula , Fuligem/química , Propriedades de SuperfícieRESUMO
BACKGROUND: The clinical diagnosis of melanoma could be difficult for a general practitioner and, in some cases, for dermatologists. To enhance and support the clinical evaluation of pigmented skin lesions a computer-aided diagnosis has been introduced. MATERIALS AND METHODS: Images of melanocytic lesions (477 total, 42 melanomas and 435 melanocytic nevi) evaluated in epiluminescence microscopy and recorded with x16 magnification were selected. A training set of 22 melanomas and 218 nevi was randomized from the dataset. The test set was formed by the complement (the remaining 20 melanomas and 217 nevi). Furthermore, a set of images consisting of 31 melanomas and 103 nevi was selected to compare the discrimination capacity of three general practitioners and three dermatologists with experience in dermoscopy (2 years), and with the automatic data analysis for the melanoma early detection system (ADAM). Sensitivity and specificity were estimated for observer assessments and computer diagnosis. RESULTS: The entire dataset used to test the implementation of the diagnostic algorithms ADAM showed a good sensitivity and specificity performance. Compared with the physicians, the ADAM system showed a slightly higher diagnostic performance in terms of sensitivity and a lower one in terms of specificity. Dermatologists showed higher levels of specificity, but lower levels in terms of sensitivity, when compared with the general practitioners. CONCLUSION: Image analysis has the potential to distinguish nevi and melanomas and to support the clinical diagnosis of melanocytic lesions by the general practitioner.