| dc.description.abstract | Dental caries is the most common chronic disease in the world. An early diagnosis is required in order to initiate prompt treatment strategies. Radiographic imaging (X-Ray) is the most common detection method, but uses ionization radiation and isn’t ideally used during pregnancy or in early childhood development. Optical coherence tomography (OCT) imaging uses no ionising radiation and may be used in place of X-rays in some clinical settings. We present development of a Convolutional Neural Network (CNN) for quantification of caries lesions in human dental tissues imaged by OCT. Using high definition high contrast time delay integration x-ray microtomography (XMT) as a gold standard for mineralization measurements, we are training the CNNs to apply weightings derived from XMT to OCT data. We show results using Local Laplacian, Wiener, BM3D denoising and sharpening, to clean low SNR OCT data followed by automatic identification of normal structure and pathology. | en_US |
