Histopathology Reconstruction on Digital Imagery

Diagnosing cervical cancer in a woman is a multi-step procedure involving examination of the cervix, possible biopsy and follow-up. It is open to subjective interpretation and highly dependent upon the skills of cytologists, colposcopists, and pathologists. In an effort to reduce the subjectiveness of the colposcopist-directed biopsy and to improve the diagnostic accuracy of colposcopy, we have developed new colposcopic imaging systems with accompanying computer aided diagnostic (CAD) techniques to guide a colposcopist in deciding if and where to biopsy. If the biopsy's histopathology, the identification of the disease state at the cellular and near-cellular level, is to be used as the gold standard for CAD, then the location of the histopathologic analysis must match exactly to the location of the biopsy tissue in the digital image. Otherwise, no matter how perfect the histopathology and the quality of the digital imagery, the two data sets cannot be matched and the true sensitivity and specificity of the CAD cannot be ascertained. We report here on new approaches to preserving, continuously, the location and orientation of a biopsy sample with respect to its location in the digital image of the cervix so as to preserve the exact spatial relationship throughout the mechanical aspects of the histopathologic analysis. This new approach will allow CAD to produce a linear diagnosis and pinpoint the location of the tissue under examination.