3 – Warsaw University of Technology, Warsaw, Poland 4 – Central Clinical Hospital of the Ministry of the Interior and Administration, Warsaw, Poland.

MATERIALS AND METHODS

RESULTS

DISCUSSION AND CONCLUSIONS

Anna Korzyńska1, Łukasz Roszkowiak1, Jakub Żak1, Dorota Pijanowska1, Krzysztof Siemion1,4,

Wojciech Kozłowski2, Renata Maryniak2, Tomasz Markiewicz2,3

BACKGROUND

Methods: The MetpiKi67 (METhod of Proliferation Index of Ki67) method quantifies proliferating

cells' index in samples from Diffuse Large B-cell Lymphoma (DLBCL) patients. This method

quantifies the percentage of immunopositive nuclei (proliferating cancer cells nuclei) to the whole

number of nuclei in selected region.

The proliferation index is the most important feature used in benign and malignant diagnosis. It quantifies the percentage of immunopositive nuclei (proliferating cancer cells nuclei) to the whole number of nuclei in selected region with Ki67 stains. The localization of the antigens are visualized with 3,3'-Diaminobenzidine and Haematoxylin (DAB&H). The automated quantification of this index is dependent of the architecture of tissue section and type of diseases what results in several such algorithms developed so far [1-4]. In the case of this study the method is adjusted to the samples from the patients diagnosed towards Diffuse Large B-cell Lymphoma (DLBCL), that is very aggressive and can cause high mortality. The samples are mostly biopsies of various lymph nodes, but also skin, tonsils, breast and so on. The lymph node architecture is not always present in samples. The B-cell nuclei are varying in the shape, size and its chromatin is arranged differently: sometimes is uniformly distributed, but mostly is distributed peripherally with dark brown spots on brighter brown background or opposite. The samples of lymph node and affected tissue in various magnification are presented below.

Proposed MetpiKi67 method performs well

according to visual inspection but it should be

evaluated in comparison to the gold standard.

We plan to compare the results with other

automatic methods [1-4] and with results of

human evaluation based on objects counting.

The higher the cancer grade the less homogeneous nuclei structures are observed. Their size is increasing and shape is varying. This essentially complicates the identification of individual portions of chromatin and combining them into one object.

Acknowledgements:

References:

This study was supported by the National Centre

for Research and Development, Poland

(grant PBS2/A9/21/2013).

Automated evaluation of the region

PI = 93% and

objects area ratio = 0.90

Pathologist evaluation of WSI = 90%

Automated evaluation of the region

PI = 80% and

objects area ratio = 0.73

Pathologist evaluation of WSI = 60%

Automated evaluation of the region

PI = 70% and

objects area ratio = 0.88

Pathologist evaluation of WSI = 90%

Automated evaluation of the region PI =

70% and

objects area ratio = 0.64

Pathologist evaluation of WSI = 90%

Automated evaluation of the region PI

= 89% and

objects area ratio = 0.95

Pathologist evaluation of WSI = 90%

1. Markiewicz T, Grala B, Kozlowski W, Osowski S. Computer system for cell counting in selected brain tumors at Ki-67 immunohistochemical staining. Anal Quant

Cytol Histopathol 2010;32:323–32.

2. Markiewicz T, Wisniewski P, Osowski S, Patera J, Kozlowski W, Koktysz R. Comparative analysis of the methods for accurate recognition of cells in the nuclei

staining of the Ki- 67 in neuroblastoma and ER/PR status staining in breast cancer. Anal Quant Cytol Histopathol 2009;31:49–62.

3. Marcial Gracia Rojo, Gloria Bueno, Janina Słodkowska, 2009, “Review of imaging solutions for integrated quantitative immunohistochemicastry in the Pathology

daily practice”, Folia Histochmica et cytobiologica, vol 47(3),2009 str.: 349-354

4. Singh SS, Qaqish B, Johnson JL, Ford OH 3rd, Foley JF, Maygarden SJ, Mohler JL. Samplingstrategy for prostate tissue microarrays for Ki-67 and androgen

receptor biomarkers. Anal Quant Cytol Histol Aug;2004 26:194–200

5. Tomasz Markiewicz, Anna Korzynska, Andrzej Kowalski, Zaneta Swiderska-Chadaj, Piotr Murawski, Bartlomiej Grala, Malgorzata Lorent, Marek Wdowiak,

Jakub Zak, Lukasz Roszkowiak, Wojciech Kozlowski, Dorota Pijanowska, 2016, „MIAP - web-based platform for the computer analysis of microscopic images to

support the pathological diagnosis”, Biocybernetics & Biomedical Engineering 36(4): 597-609, DOI 10.1016/S0208-5216;,

6. Łukasz Roszkowiak, Anna Korzyńska, Marylene Lejeune, Ramon Bosch and Carlos Lopez, 2016,„Improvements to segmentation method of stained lymphoma

tissue section images”, Ed.: Burduk Robert, Jackowski Konrad, Kurzynski Marek, Woźniak Michał, Zolnierek Andrzej “Proceedings of the 9th International

Conference on Computer Recognition Systems CORES 2015”, Advances in Intelligent Systems and Computing 403: 609-617 Springer International

PublishingDOI: 10.1007/978-3-319-26227-7_57;

7. Swiderska Z, Korzynska A, Markiewicz T, Lorent M, Zak J, Wesolowska A, et al. Comparison of the manual, semiautomatic, and automatic selection and leveling

of hot spots in whole slide images for Ki-67 quantification in meningiomas. Anal Cell Pathol 2015;2015:1–15.

11 digital slides from Pathology Department of the

Military Institute of Medicine in the form of .mrxs files

(scanned by 3DHistech slide scanners) with diagnosis

and proliferation index estimated by experienced

pathologist was used to verify proposed method.

Skin of the left mental

region.

Lymph node of the right axillary

region.

The MetpiKi67 of Proliferation Index estimation in samples

from patients with Diffuse Large B-cell Lymphoma

1 – Nalecz Institute of Biocybernetics and Biomedical Engineering PAS, Warsaw, Poland. 2 – Military Institute of Medicine, Warsaw, Poland.

It is complicated to segment nuclei with peripheral chromatin than to select nuclei of cell not affected by cancer.