The quantification of DNA damage, both in vivo and in vitro, can be very time consuming, since large amounts of samples need to be scored. Additional uncertainties may arise due to the lack of documentation or by scoring biases. Image analysis automation is a possible strategy to cope with these difficulties and to generate a new quality of reproducibility. In this communication we collected some recent results obtained with the automated scanning platform Metafer, covering applications that are being used in radiation research, biological dosimetry, DNA repair research and environmental mutagenesis studies. We can show that the automated scoring for dicentric chromosomes, for micronuclei, and for Comet assay cells produce reliable and reproducible results, which prove the usability of automated scanning in the above mentioned research fields.

The micronucleus assay (MNT) in human lymphocytes is frequently used to assess chromosomal damage as a consequence of environmental mutagen exposure, to assess the effect of mutagens or to search for reduced DNA repair capacity after a mutagenic challenge. We have established an automated scoring procedure for the cytokinesis blocked MNT based on computerized image analysis (Metasystems Metafer 4 version 2.12). To evaluate the results we used the reproducibility of counts, established a dose-response curve for gamma-irradiation and used the ability of the system to differentiate between breast cancer patients and controls as a biological reference, a difference which we had observed before by visual counting. Blood cultures were irradiated with gamma-rays (2 Gy) at the beginning and treated with cytochalasin B during the last 24 h. The slides were stained with Giemsa for visual counting and with DAPI for automated analysis. Our test sample consisted of 73 persons (27 with breast cancer and 26 female and 20 male controls). A comparison between visual counting (controls, mean MN frequency 313) and automated counting (mean MN frequency 106) in slides from the same culture revealed a large drop for the automated counts. However, the automated counts were as reproducible as the visual counts [coefficient of variation (CV) on the sample approximately 20%; CV on repeated counts of the same slides approximately 5%] and both counts were highly correlated. Furthermore, the discrimination between cases and controls improved for automated counting of slides from the same cultures [visual odds rato (OR) < or = 4.0, P = 0.009; automated OR > 16, P < 0.0001], with a strong dependence on the set of parameters used. This improvement was confirmed in a validation sample of an additional 21 controls and 20 cases (OR = 11, P = 0.0018) performed as a prospective or diagnostic test.

Background: A new chimerism analysis based on automated interphase fluorescence in situ hybridization (FISH) evaluation was established to detect residual cells after allogene sexmismatched bone marrow or blood stem-cell transplantation.

BACKGROUND AND AIMS Routine pathological diagnostics in kidneys are mainly based on semi-quantitative eyeballing. In own former studies, we showed predictive value of precise immune cell quantification in allografts using digital semi-automated techniques. We now aim to achieve fully automated segmentation workflow with CNNs. METHOD Standard routine stains (immuno/histochemistry, immunofluorescence) were digitized (20×) with Metafer, a commercial scanning/imaging platform. Diagnostically relevant anatomical compartments (cortex, medulla, glomeruli, tubuli [proximal/distal/collecting duct], glomerular/peritubular capillaries and nuclei) were manually annotated by use of immunomarkers to generate large data sets on human renal biopsies and nephrectomies. Data were used to train multi-class semantic segmentation CNNs with broad data augmentation to achieve a robustness against staining variances. RESULTS Using Jones-HE stains for multi-class segmentation, a cortex-medulla-extrarenal CNN revealed pixel based hit rates above 97.9%, detection of glomeruli had a pixel based hit rate above 99%, a multi-class CNN for tubules, tubular membranes and peritubular capillaries resulted in a hit rate of 91.5%, and nuclear-based cell detection shows pixel based hit rates above 98%. Identification of cell location in interstitium, tubuli, glomeruli, peritubular and glomerular capillaries reached very high hit rates: Glomerular endothelial cells actually result in 83% true positives, 13% false negatives and 4% false positives. Additionally, a tubulus classifier (proximal tubulus, distal tubulus, collecting duct and atrophic tubulus) with an accuracy &gt;90% was developed. CONCLUSION Automated structure segmentation by CNNs can complement and specify classical nephropathological diagnostics, especially for spatial risk marker evaluation in early transplant biopsies.