Digital Holographic Microscopy: Screening Tool for Cervix Cancer. Preliminary Study
Nazim Benzerdjeb, Christian Garbar, Philippe Camparo, Henri P Sevestre. CHU Picardie, Amiens, France; Institut Jean Godinot, Reims, France; Centre de Pathologie, Amiens, France
Background: Liquid-based cytology Pap test (LBC) replaces the conventional Pap test as LBC is more efficient, partly owing to computer assistance. However the diagnosis still depends on human eye, sensitivity is not perfect, and LBC requires complex technology. Holocyt Intelligence System Corp. (HIS) develops of a digital holographic microscopy (DHM) instrument and software, which provides a three dimensional image reconstruction of uterine cervical cells, without delay nor staining. The purpose of this preliminary study was to identify reproducible optic criteria of cells considered normal or abnormal.
Design: Analysis was performed with the DHM (model c-Line), vial “c-Box”, computer and software (Ovizio®) lent by HIS). Residual material from the three labs, randomly chosen, was analysed after LBC (Thinprep® (UHA) and Turbitec® (IJG, PCA). Residual material poured into a “c-Box” was subjected to DHM. At least 20 “normal” or “abnormal” looking cells reconstructed by software were selected per vial. The following were automatically measured and extracted: maximum height nuclear (MHN), normalized nuclear to cytoplasm height (NHN), nuclear diameter (ND) and nucleo-cytoplasmic ratio (NCR). Low and high grade cells were grouped, due to small samples figures. Values are expressed as mean ± SD. Statistical analysis of the data were performed using Mann–Whitney and ROC curve. Differences are considered significant when p < 0.05.
Results: In 32 specimens (12 in UHA, 15 in IJG, 5 in PCA) with normal Pap test, 1333 cells analyzed with DHM. In 21 specimens (12 in UHA, 4 in IJG, 5 in PCA) with abnormal Pap test, 494 cells were analyzed with DHM. Available histologic correlation was: 8 CIN1 (6 LSIL, 2 ASC-US), 2 CIN3 (1 LSIL, 1 ASC-H), 3 normal. with pooled three or one observer (except for MHN in PCA p=0.56), a significant difference between normal and abnormal specimens was clear (three observers: NHN:0.45±0.14 vs 0.59±0.21, MHN:0.19±0.08 vs 0.28±0.12, ND:9.13±2.2 vs 18.6±9.8, NCR:0.04±0.04 vs 36±0.32). Globally, the areas under the ROC curves (95% confidence intervals) are 0.82[0.79-0.84] for NHN, 0.85[0.82-0.87] for MHN, 0.94[0.93-0.96] for ND, 0.97[0.96-0.98] for NCR.
Conclusions: This preliminary study demonstrates that the criteria analyzed with DHM had good area under the ROC. We feel encouraged to perform large scale validation study. Hopefully holograhic analysis will be performed automatically, will provide quickly clear delineation of normal and abnormal cells, on a closed vial, without alteration of fixed cells, allowing ancillary studies.
Monday, March 4, 2013 1:00 PM
Poster Session II # 81, Monday Afternoon