Integration of Fractal Dimension in Vision Transformer for Skin Cancer Classification
Keywords:machine learning, skin cancer, skin lesion classification, Vision Transformer, fractal dimension, classification tasks
In order to classify skin lesions, many efforts have been made to create various automated diagnostic systems. For that purpose many efforts have been put in creating various automated diagnostics systems Nowadays, with the rapid advancements in deep learning, Vision Transformers have emerged as powerful models for image processing and analysis purposes. This type of model has already proved useful for cancer detection and classification tasks in particular. However, the complexity and variability of skin lesions present significant challenges in accurately classifying them. Integrating the concept of fractal dimension into Vision Transformers can potentially improve their performance by capturing the intricate structural patterns of skin lesions. This paper aims to explore the integration of fractal dimension metrics into a Vision Transformer for skin cancer classification. The problem at hand is to investigate the integration of fractal dimension metrics into the existing Vision Transformer architecture for the accurate classification of skin lesions as cancerous or non-cancerous. Fractal dimensions provide a measure of the complexity and irregularity of an object, which can be informative in characterizing skin cancer lesions. We aim to research possability and ways of incorporating fractal dimension metrics into the Vision Transformer model for results improvements.
Cancer facts & figures 2023 (no date) American Cancer Society. Available at: https://www.cancer.org/research/cancer-facts-statistics/all-cancer-facts-figures/2023-cancer-facts-figures.html (Accessed: 10 June 2023).
(No date) Adjusted rates 2018 melanoma of skin С43 table 1 – general rates, 2018. Available at: http://www.ncru.inf.ua/publications/BULL_21/PDF_E/38-39-mel.pdf (Accessed: 10 June 2023).
A. Dosovitskiy, et al. An image is worth 16x16 words: Transformers for image recognition at scale, (2021), arXiv.org. Available at: https://doi.org/10.48550/arXiv.2010.11929 (Accessed: 10 June 2023).
A. Vaswani, et al. Attention is all you need, (2017), arXiv.org. Available at: https://doi.org/10.48550/arXiv.1706.03762 (Accessed: 10 June 2023).
K. J. Falconer, (2003). Fractal geometry : Mathematical foundations and applications : K. J. Falconer, onlinelibrary.wiley.com. Available at: https://doi.org/10.1002/0470013850 (Accessed: 10 June 2023).
M. Kirkby, (1983). The fractal geometry of nature. Benoit B. Mandelbrot. W. H. Freeman and co., San Francisco, 1982. No. of pages: 460. Price: £22.75 (hardback). Earth Surface Processes and Landforms, 8(4), 406. https://doi.org/10.1002/esp.329008041 Accessed: 10 June 2023).
P. Tschandl, (2023). The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions, Harvard Dataverse. Available at: https://doi.org/10.7910/DVN/DBW86T (Accessed: 10 June 2023).
J. F. Canny, (1986). A computational approach to edge detection – researchgate. Available at: http://dx.doi.org/10.1109/TPAMI.1986.4767851 (Accessed: 10 June 2023).
V. Nikitin, and N. Shapoval, (2023). Vision Transformer for skin cancer classification, Scientific Collection ‘InterConf+’. Available at: https://doi.org/10.51582/interconf.19-20.05.2023.039 (Accessed: 10 June 2023).
C.-F. Chen, Q. Fan, and R. Panda, (2021). Crossvit: Cross-attention multi-scale vision transformer for Image Classification, arXiv.org. Available at: https://arxiv.org/abs/2103.14899 (Accessed: 10 June 2023)
Authors who publish with this journal agree to the following terms:
Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).