Automatic diagnosis of melanoma using hyperspectral data and GoogLeNet.
GoogLeNet
deep learning
hyperspectral imager
melanoma
Journal
Skin research and technology : official journal of International Society for Bioengineering and the Skin (ISBS) [and] International Society for Digital Imaging of Skin (ISDIS) [and] International Society for Skin Imaging (ISSI)
ISSN: 1600-0846
Titre abrégé: Skin Res Technol
Pays: England
ID NLM: 9504453
Informations de publication
Date de publication:
Nov 2020
Nov 2020
Historique:
received:
23
03
2020
accepted:
30
05
2020
pubmed:
26
6
2020
medline:
14
8
2021
entrez:
26
6
2020
Statut:
ppublish
Résumé
Melanoma is a type of superficial tumor. As advanced melanoma has a poor prognosis, early detection and therapy are essential to reduce melanoma-related deaths. To that end, there is a need to develop a quantitative method for diagnosing melanoma. This paper reports the development of such a diagnostic system using hyperspectral data (HSD) and a convolutional neural network, which is a type of machine learning. HSD were acquired using a hyperspectral imager, which is a type of spectrometer that can simultaneously capture information about wavelength and position. GoogLeNet pre-trained with Imagenet was used to model the convolutional neural network. As many CNNs (including GoogLeNet) have three input channels, the HSD (involving 84 channels) could not be input directly. For that reason, a "Mini Network" layer was added to reduce the number of channels from 84 to 3 just before the GoogLeNet input layer. In total, 619 lesions (including 278 melanoma lesions and 341 non-melanoma lesions) were used for training and evaluation of the network. The system was evaluated by 5-fold cross-validation, and the results indicate sensitivity, specificity, and accuracy of 69.1%, 75.7%, and 72.7% without data augmentation, 72.3%, 81.2%, and 77.2% with data augmentation, respectively. In future work, it is intended to improve the Mini Network and to increase the number of lesions.
Sections du résumé
BACKGROUND
BACKGROUND
Melanoma is a type of superficial tumor. As advanced melanoma has a poor prognosis, early detection and therapy are essential to reduce melanoma-related deaths. To that end, there is a need to develop a quantitative method for diagnosing melanoma. This paper reports the development of such a diagnostic system using hyperspectral data (HSD) and a convolutional neural network, which is a type of machine learning.
MATERIALS AND METHODS
METHODS
HSD were acquired using a hyperspectral imager, which is a type of spectrometer that can simultaneously capture information about wavelength and position. GoogLeNet pre-trained with Imagenet was used to model the convolutional neural network. As many CNNs (including GoogLeNet) have three input channels, the HSD (involving 84 channels) could not be input directly. For that reason, a "Mini Network" layer was added to reduce the number of channels from 84 to 3 just before the GoogLeNet input layer. In total, 619 lesions (including 278 melanoma lesions and 341 non-melanoma lesions) were used for training and evaluation of the network.
RESULTS AND CONCLUSION
CONCLUSIONS
The system was evaluated by 5-fold cross-validation, and the results indicate sensitivity, specificity, and accuracy of 69.1%, 75.7%, and 72.7% without data augmentation, 72.3%, 81.2%, and 77.2% with data augmentation, respectively. In future work, it is intended to improve the Mini Network and to increase the number of lesions.
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
891-897Informations de copyright
© 2020 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.
Références
Koh D, Wang H, Lee J, Chia KS, Lee HP, Goh CL. Basal cell carcinoma, squamous cell carcinoma and melanoma of the skin: analysis of the Singapore Cancer Registry data 1968-97. Br J Dermatol. 2003;148:1161-1166.
Meyskens FL, Berdeaux DH, Parks B, Tong T, Loescher L, Moon TE. Cutaneous malignant melanoma (Arizona cancer center experience). I. Natural history and prognostic factors influencing survival in patients with stage I disease. Cancer. 1988;62(6):1207-1214.
Nomura K. Frequency and characteristic feature of metastatic brain tumors in Japan. Jpn J Neurosurg. 2003;12(5):323-329.
Rigel DS, Carucci JA. Malignant melanoma: prevention, early detection, and treatment in the 21st century. Cancer J Clin. 2000;50(4):215-236.
Uong A, Zon LI. Melanocytes in development and cancer. J Cell Physiol. 2010;222(1):38-41.
Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature. 2000;407(6801):249-257.
Rosenberg LE, Rosenberg DD. Single-Gene De-fects. In: Human Genes and Genomes. Massachusetts: Academic Press; 2012: 193.
Stolz W, Riemann A, Cognetta AB, et al. ABCD rule of dermatoscopy: a new practical method for early recognition of malignant melanoma. Eur J Dermatol. 1994;4(7):521-527.
Argenziano G, Soyer HP, Chimenti S, et al. Dermoscopy of pigmented skin lesions: results of a consensus meeting via the internet. J Am Acad Dermatol. 2003;48(5):679-693.
Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis. Med Image Anal. 2017;42:60-88.
Li Y, Zhang H, Shen Q. Spectral-spatial classi-fication of hyperspectral imagery with 3D convolutional neural network. Remote Sens. 2017;9(1):67.
Kingma DP, Welling M.Auto-encoding variational bayes. arXiv preprint, arXiv:1312.6114, 2013.
Goodfellow IJ, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets. Proc of Advances in Neural Information Processing Systems 27 (NIPS 2014). pp. 2672-2680, 2014.
Fujisawa Y, Otomo Y, Ogata Y, et al. Deep-learning-based, computer-aided classifier developed with a small dataset of clinical images surpasses board-certified dermatologists in skin tumour diagnosis. Br J Dermatol. 2019;180(2):373-381.
Esteva A, Kuprel B, Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):115-118.
Takemura K, Hirano G, Nagaoka T. Automatic diagnosis of melanoma using HSD and machine learning. IEICE Tech Rep. 2018; 117(518): 103-104 (in Japanese).
Hirano G, Nemoto M, Kimura Y, et al. Automatic diagnosis of melanoma using hyperspectral data and GoogLeNet. Proc JBMES. 2019;1:1A-6A.
Kato K, Nemoto M, Kimura Y, et al. Performance improvement of automated melanoma diagnosis system by data augmentation. Proc JBMES. 2019;1:1A-7A.
Nagaoka T, Nakamura A, Okutani H, et al. Hyperspectroscopic screening of melanoma on acral volar skin. Skin Res Technol. 2012;18(3):301-310.
Moller MF. A scaled conjugate gradient algorithm for fast supervised learning. Neural Netw. 1993;6(4):525-533.
Barlow HB. Unsupervised learning. Neural Comput. 1989;1(3):295-311.
Kaelbling LP, Littman ML. Reinforcement learning a survey. J Artif Intell Res. 1996;4:237-285.
Zhang GP. Neural networks for classification: a survey. IEEE T Syst Man Cy C. 2000;30(4):451-462.
Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Proc of the 25th International Conference on Neural Information Processing Systems. 1, pp. 1097-1105, 2012.
Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions. Proc of the IEEE conference on computer vision and pattern recognition. pp. 1-9, 2015.
He K, Zhang X, Ren S, Sun J.Deep residual learning for image recognition. Proc of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 770-778, 2016.
Deng J, Dong W, Socher R, Li L, Li K, Fei-Fei L. ImageNet: a large-scale hierarchical image data-base. 2009 IEEE Conference on Computer Vision and Pattern Recognition. pp. 248-255, 2009.
Yoshida T, Iyatomi H. Alignment of major axis for automated melanoma diagnosis with deep learning approach, 31st fuzzy system symposium, 2015 (in Japanese).