Multiparameter breast cancer cell image analysis for objective estimation of nuclear grade: Comparison with light microscopic observational data

We performed a multiparameter image analysis assessment of breast cancer cell population nuclear grade (NG), which is regarded as one of the main prognostic factors for treatment efficacy and survival of the patients and compared it with light microscopic estimation of NG. Cytological imprint slides from 20 ductal carcinomas (15 infiltrating and 5 noninfiltrating) were stained according to Leischmann-AzureII-eosine method, and NG was estimated by light microscopic observation according to Black in Fisher's modification. Simultaneously, using specially elaborated software, in each patient 100 cancer cells were analyzed for nuclear perimeter, diameter, area, nucleolar area, and average intensity of staining. The chromatin structure was assessed using mean diameter of chromatin grains and relative chromatin area within the nucleus. Light microscopic estimation revealed 4/15 grade 2 and 7/15 grade 3 tumours out of 15 infiltrating ductal carcinomas, with 4/15 classified as intermediate between grade 2-3. Multifactorial linear correlation coefficient r = 0.39, p<0.001 for ductal cancer, higher NG comes with increasing nucleolar area, nuclear roundness factor (polymorphism), nuclear area, and chromatin area within the cell nucleus. Image analysis may yield precise information on NG as a prognostic factor in breast cancer patients.