Statistics-based Classification Approach for Hyperspectral Dermatologic Data Processing

Hyperspectral Imaging (HSI) for dermatology applications lacks a physical model to differentiate between cancerous or non-cancerous pigmented skin lesions. In this paper the statistical properties of a set of HSI data are exploited as an alternative to this limitation. The hyperspectral dermatologic database employed in the experiments is composed by 40 noncancerous and 36 cancerous pigmented skin lesions (PSLs) obtained from 61 patients. The preliminary experiments suggest the potential of a simple statistics metrics, such as the coefficient of variation, to distinguish between cancerous and non-cancerous PSLs using hyperspectral data. A sensitivity result of 100% was achieved in the test set providing an overall accuracy classification of 80%.

[1]  M. Mihm,et al.  The performance of MelaFind: a prospective multicenter study. , 2011, Archives of dermatology.

[2]  Hannes Kazianka,et al.  Segmentation and Classification of Hyper-Spectral Skin Data , 2007, GfKl.

[3]  Philippe Burlina,et al.  Cross validating hyperspectral with Ultrasound-based skin thickness estimation , 2014, 2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS).

[4]  Janis Spigulis,et al.  Smartphone snapshot mapping of skin chromophores under triple-wavelength laser illumination , 2017, Journal of biomedical optics.

[5]  Roberto Sarmiento,et al.  Dermatologic Hyperspectral Imaging System for Skin Cancer Diagnosis Assistance , 2019, 2019 XXXIV Conference on Design of Circuits and Integrated Systems (DCIS).

[6]  Reda Kasmi,et al.  Classification of malignant melanoma and benign skin lesions: implementation of automatic ABCD rule , 2016, IET Image Process..

[7]  Baowei Fei,et al.  In-Vivo and Ex-Vivo Tissue Analysis through Hyperspectral Imaging Techniques: Revealing the Invisible Features of Cancer , 2019, Cancers.

[8]  Vandana Jagtap,et al.  Computer Aided Melanoma Skin Cancer Detection Using Image Processing , 2015 .

[9]  G. Argenziano,et al.  The impact of dermoscopy on melanoma detection in the practice of dermatologists in Europe: results of a pan‐European survey , 2017, Journal of the European Academy of Dermatology and Venereology : JEADV.

[10]  Mia K Markey,et al.  Segmentation of diffuse reflectance hyperspectral datasets with noise for detection of Melanoma , 2012, 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[11]  A. Jemal,et al.  Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries , 2018, CA: a cancer journal for clinicians.

[12]  Jeremy S Bordeaux,et al.  Early detection of melanoma: reviewing the ABCDEs. , 2015, Journal of the American Academy of Dermatology.

[13]  Tim K. Lee,et al.  Automatic detection and segmentation of vascular structures in dermoscopy images using a novel vesselness measure based on pixel redness and tubularness , 2015, Medical Imaging.

[14]  S. Kozlov,et al.  In vivo hyperspectral imaging of skin malignant and benign tumors in visible spectrum , 2018 .

[15]  Guolan Lu,et al.  Medical hyperspectral imaging: a review , 2014, Journal of biomedical optics.

[16]  A. Marghoob,et al.  High-dynamic-range dermoscopy imaging and diagnosis of hypopigmented skin cancers. , 2015, JAMA dermatology.

[17]  Ferdi van der Heijden,et al.  Statistical analysis of spectral data: a methodology for designing an intelligent monitoring system for the diabetic foot , 2013, Journal of biomedical optics.

[18]  Fred Godtliebsen,et al.  Recent advances in hyperspectral imaging for melanoma detection , 2019, WIREs Computational Statistics.

[19]  Eduardo Quevedo,et al.  Non-Invasive Skin Cancer Diagnosis Using Hyperspectral Imaging for In-Situ Clinical Support , 2020, Journal of clinical medicine.