Two-photon images reveal unique texture features for label-free identification of ovarian cancer peritoneal metastases.

For cancer patients, treatment selection fundamentally relies on staging, with "under-staging" considered a common problem. Imaging modalities that can complement conventional white-light laparoscopy are needed to detect more accurately small metastatic lesions in patients undergoing operative cancer care. Biopsies from healthy parietal peritoneum and ovarian peritoneal metastases obtained from 8 patients were imaged employing a two-photon laser scanning microscope to generate collagen-second harmonic generation (SHG) and fluorescence images at 755 nm and 900 nm excitation and 460 ± 20 nm and 525 ± 25 nm emission. Forty-one images were analyzed by automated image processing algorithms and statistical textural analysis techniques, namely gray level co-occurrence matrices. Two textural features (contrast and correlation) were employed to describe the spatial intensity variations within the captured images and outcomes were used for discriminant analysis. We found that healthy tissues displayed large variations in contrast and correlation features as a function of distance, corresponding to repetitive, increased local intensity fluctuations. Metastatic tissue images exhibited decreased contrast and correlation related values, representing more uniform intensity patterns and smaller fibers, indicating the destruction of the healthy stroma by the cancerous infiltration. The textural outcomes resulted in high classification accuracy as evaluated quantitatively by discriminant analysis.