Automatic diagnosis of melanoma using hyperspectral data and GoogLeNet
(2020) In Skin Research and Technology 26(6). p.891-897- Abstract
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: 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... (More)
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: 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: 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.
(Less)
- author
- organization
- publishing date
- 2020-11
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- deep learning, GoogLeNet, hyperspectral imager, melanoma
- in
- Skin Research and Technology
- volume
- 26
- issue
- 6
- pages
- 7 pages
- publisher
- Wiley-Blackwell
- external identifiers
-
- scopus:85087212720
- pmid:32585082
- ISSN
- 0909-752X
- DOI
- 10.1111/srt.12891
- language
- English
- LU publication?
- yes
- id
- fa155e4d-70b5-4d49-9c63-c9f96dc5f086
- date added to LUP
- 2020-07-20 10:37:36
- date last changed
- 2024-09-05 01:36:10
@article{fa155e4d-70b5-4d49-9c63-c9f96dc5f086, abstract = {{<p>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: 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: 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.</p>}}, author = {{Hirano, Ginji and Nemoto, Mitsutaka and Kimura, Yuichi and Kiyohara, Yoshio and Koga, Hiroshi and Yamazaki, Naoya and Christensen, Gustav and Ingvar, Christian and Nielsen, Kari and Nakamura, Atsushi and Sota, Takayuki and Nagaoka, Takashi}}, issn = {{0909-752X}}, keywords = {{deep learning; GoogLeNet; hyperspectral imager; melanoma}}, language = {{eng}}, number = {{6}}, pages = {{891--897}}, publisher = {{Wiley-Blackwell}}, series = {{Skin Research and Technology}}, title = {{Automatic diagnosis of melanoma using hyperspectral data and GoogLeNet}}, url = {{http://dx.doi.org/10.1111/srt.12891}}, doi = {{10.1111/srt.12891}}, volume = {{26}}, year = {{2020}}, }