Optical Coherence Tomography (OCT) Imaging and Computer Aided Diagnosis of Human Cervical Tissue Specimens

摘要

The keyword for management of cervical cancer is prevention. The present program within the UK, the 'National Health Service (NHS) cervical screening programme' (NHSCSP), is based on cytology. Although the program has reduced the incidence of cervical cancer, this program requires patient follow ups and relies on diagnostic biopsying. There is potential for reducing costs and workload within the NHS, and relieving anxiety of patients. In this study, Optical Coherence Tomography (OCT) was investigated for its capability to improve this situation. Our time domain bench top system used a superluminescent diode (Superlum), centre wave length ～1.3 μm, resolution (air) ～15 μm. Tissue samples were obtained according to the ethics approval by Gloucestershire LREC, Nr. 05/Q2005/123. 1387 images of 199 participants have been compared with histopathology results and categorized accordingly. Our OCT images do not reach the clarity and resolution of histopathology. Further, establishing and recognizing features of diagnostic significance seems difficult. Automated classification would allow one to take decision-making to move from the subjective appraisal of a physician to an objective assessment. Hence we investigated a classification algorithm for its ability in recognizing pre-cancerous stages from OCT images. The initial results show promise.