Identification of prostheses in dental images: avoiding ionizing images using hyperspectral information
Abstract
Hyperspectral images, images in multiples wavelengths of the electromagnetic spectrum of light that go beyond the visible spectrum, have been widely used for medical diagnosis and focused mainly on classification of healthy or malignant tissues, due to its ability to detect changes in some biological markers, such as oxygen saturation. However, segmentation using hyperspectral images, which consists of grouping regions of interest in the image, such as organs, tissues and lesions, is still little studied in the literature, despite its great potential. Therefore, the objective of this present work was to study whether hyperspectral images of oral and dental reflectance help in the differentiation between prosthesis and enamel (tooth). Segmentation experiments using a public dataset of oral and dental images (ODSI-DB) showed better performance of a UNet trained with hyperspectral images, when compared to the one trained with RGB images. A qualitative assessment also showed that there are a large number of inconsistent manual segmentation masks for the prosthesis class that are inconsistent (41%), possibly indicating that the performance of the U-Net is much better than the measured Dice.
References
M. Paoletti, J. Haut, J. Plaza, and A. Plaza, “Deep learning classifiers for hyperspectral imaging: A review,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 158, p. 279–317, Dec. 2019.
M. Akewar and M. Chandak, “Hyperspectral imaging algorithms and applications: A review,” Jan. 2024.
G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” Journal of Biomedical Optics, vol. 19, no. 1, p. 010901, Jan. 2014.
S. Panasyuk, S. Yang, D. Faller, D. Ngo, R. Lew, J. Freeman, and et al., “Medical hyperspectral imaging to facilitate residual tumor identification during surgery,” Cancer Biol Ther, vol. 6, no. 3, p. 439–446, Mar. 2007.
J. Ma, Y. He, F. Li, L. Han, C. You, and B. Wang, “Segment anything in medical images,” Nature Communications, vol. 15, no. 1, Jan. 2024.
Y. Shen, Z. Fang, Y. Gao, N. Xiong, C. Zhong, and X. Tang, “Coronary arteries segmentation based on 3d fcn with attention gate and level set function,” IEEE Access, vol. 7, p. 42826–42835, 2019.
U. Khan, S. Paheding, C. P. Elkin, and V. K. Devabhaktuni, “Trends in deep learning for medical hyperspectral image analysis,” IEEE Access, vol. 9, pp. 79 534–79 548, 2021.
O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in LCNS. Cham: Springer International Publishing, 2015, pp. 234–241.
J. Hyttinen, P. Fält, H. Jäsberg, A. Kullaa, and M. Hauta-Kasari, “Oral and dental spectral image database—odsi-db,” Applied Sciences, vol. 10, no. 20, p. 7246, Oct. 2020.
G. Yang, C. V. Stewart, M. Sofka, and C.-L. Tsai, “Registration of challenging image pairs: Initialization, estimation, and decision,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 11, p. 1973–1989, Nov. 2007.
H. Huang, L. Lin, R. Tong, H. Hu, Q. Zhang, Y. Iwamoto, X. Han, Y.-W. Chen, and J. Wu, “Unet 3+: A full-scale connected unet for medical image segmentation,” in ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, May 2020.
K. Zou, S. Warfield, A. Bharatha, C. Tempany, M. Kaus, S. Haker, and et al., “Statistical validation of image segmentation quality based on a spatial overlap index,” Acad. Radiol., vol. 11, no. 2, pp. 178–189, 2004.
