2013 Volume 51 Issue 2 Pages 103-111
This paper describes our attempt to distinguish cancer and normal tissues using fast Fourier transform (FFT) and its differentiation. Pathological diagnosis is mostly performed based on the observation of sectioned sample tissue, which sometimes leads to unsuccessful diagnosis. In this study, we compared and analyzed resected normal and cancer tissues from 26 cases of colorectal cancer at the Cancer Institute Hospital of Japanese Foundation for Cancer Research. Each extracted tissue was divided into two pieces that were compared after processing as follows. One was pathologically diagnosed with hematoxylin and eosin (HE) staining. The other was suspended/liquefied and fluorescently stained for approximately ten minutes with automatic cell isolation equipment and frozen dry reagent. DNA ploidy analysis was performed on the processed tissues with a flow cytometer. Previous studies reported histograms were classified into just diploidy and aneuploidy;unclassified others were not analyzed because they assumed a problem with the measuring equipment, the cell isolation method or the samples themselves. The current study discusses histogram analysis including what were excluded in previous studies. We classified the previously unclassified histograms based on their characteristics. We also developed an algorithm which uses the minimum value of differentiated FFT patterns and the frequency of oscillation for amplitudes which exceed a set value. The result of discrimination by this new algorithm, with reference to the result of pathological diagnosis, showed 100% sensitivity and around 90% specificity. In addition, we identified debris in the tube in HE stained tissue of which other half showed abundant debris in its histogram. The cases showing no diploidy in normal tissues had reduced specificity. These results suggest that our newly developed DNA ploidy and FFT analysis may be an efficacious diagnosis method.
