China 
Africa 
Explore chapters and articles related to this topic
Stochastic Model to Explain the Biology and Epidemiology of the Ultraviolet Induction of Skin Cancer
Published in Ovide Arino, David E. Axelrod, Marek Kimmel, Mathematical population dynamics, 2020
Melanomas are classified into three groups essentially based on growth patterns. Lentigo maligna melanomas (LMMs) tend to grow slowly, perhaps for decades along the skin surface. Superficial spreading melanomas (SSMs) spread radially, perhaps for several years, and then begin to grow vertically and become more virulent. Nodular melanomas (NMs) tend to grow vertically from their inception, typically creating a nodule rising above the skin; they are the most life–threatening form. The boundary between LMMs and SSMs or between SSMs and NMs is blurred and pathologists will sometimes disagree on classification (Larsen, 1980). Most reports suggest that between 50 and 85% of melanoma tumors develop in a preexisting nevus (Ariel, 1980). A nevus, or “mole,” is a benign cluster of pigment-filled melanocytes with a larger, less dendridic shape than their ordinary epidermal counterparts (Hu, 1981). The relationship between melanoma and nevi is based on histological evidence of nevi in conjunction with a tumor, similar characteristics of tumor and nevus cells, and patient recollection of moles at the sites of their melanomas. The relative risk for melanoma appears to increase with the number of nevi, even after adjustments are made for the frequency of sunburns (Østerlind et al., 1988). The association with nevi is not valid for the LMM class of tumors. Interestingly, the density of regular melanocytes in the skin adjacent to a SSM, NM, or nevus appears higher than for normal skin (Mackie, 1982). Histological slides reveal more mitotic activity in melanomas that have invaded deeper into the skin and in melanomas with less pigment present. Nonpigmented melanomas are usually of the NM classification, although most NM are pigmented (Larsen, 1980).
An automated hybrid attention based deep convolutional capsule with weighted autoencoder approach for skin cancer classification
Published in The Imaging Science Journal, 2023
The performance of the proposed method is implemented and analyzed via the PYTHON platform. Some of the performance metrics like accuracy, Precision, recall and MCC are analyzed and compared with different existing approaches. The experiment’s dataset is collected from the ISIC 2019 [34] dataset of about 25,331 images. The dataset consists of melanoma in 4522 images, melanocytic nevi in 12, 875 images, BCC in 3323 images, benign keratosis in 2624 images, actinic keratosis in 867, vascular lesion in 253 and SCC in 628 images are used. The input image size is about pixels with red, green and blue channels (RGB). Tables 1 and 2 tabulate the hyperparameters and device configuration of the proposed method.
A multigrid Waveform Relaxation Method for solving the Pennes bioheat equation
Published in Numerical Heat Transfer, Part A: Applications, 2023
Cosmo D. Santiago, Gylles R. Ströher, Marcio A. V. Pinto, Sebastião R. Franco
As previously mentioned, several works in the literature suggest that dynamic behavior favors a more adequated detection of unhealthy skin regions than a steady state analysis. Iljaž et al. [44] showed that it is possible to obtain tumor parameters using exact static or dynamic measurement data. However, a dynamic approach proved better for inaccurate temperature data than the steady-state, which cannot capture minor temperature differences between healthy skin and the tumor. In Gomboc et al. [45], an experimental setup for an active cooling device for dynamic thermography is proposed to achieve a constant cooling temperature that induces deep cooling penetration and, therefore, better thermal contrast. Magalhaes et al. [7] collected and analyzed static and dynamic (cooling) thermal images of melanoma and melanocytic nevi lesions to retrieve thermal parameters particular to these skin lesions. The steady-state and dynamic variables were tested separately using different machine learning classifiers to verify whether the distinction between melanoma and nevi lesions was achievable. The differentiation of both skin tumors was doable, with an accuracy of 84.2% and a sensitivity of 91.3% after implementing a learner based on support vector machines and an input vector composed of static variables.
Evolving strategies for the development and evaluation of a computerised melanoma image analysis system
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2018
Aashish B. Katapadi, M. Emre Celebi, Shannon C. Trotter, Metin N. Gurcan
Analysing the misclassified images revealed the following misclassifications: 11.0% Blue Nevi, 16.2% Clark Nevi, 18.5% Dermal Nevi and 48.1% Reed-Spitz Nevi. The largest misclassification rate occurred for the Reed–Spitz nevi. In fact, while 48.1% of them are misclassified in the testing, only 32.7% of the data-set is actually composed of Reed–Spitz Nevi. This suggests that these nevi are the most difficult type of benign lesion to classify. This is consistent with clinical observation that Reed–Spitz Nevi tend to demonstrate characteristics of melanomas and are considered borderline lesions (Argenziano et al. 2002). This further suggests that when creating a classification system, it may be beneficial to develop a strategy reflecting the diversity of the categories of benign images.