A Robust Computer-Aided Approach for Scoring and Quantitative Evaluation of Whole Slide Digital Images of HER2 Immunohistochemistry in Breast Cancer

Articles

4 MIN

HER2 is recognized as an essential prognostic parameter in breast cancer. Usually interpretation of HER2 negative cases which has no membrane staining or has faint and weak incomplete cells and also positive cases (3+) that consist of strong and complete cell are easy but cells with moderate and weak HER2 staining (2+ or equivocal) require further analysis with fluorescence in situ hybridization (FISH) technique which is expensive and time consuming. In order to minimize the number of equivocal 2+ cases, we presented an automated approach for scoring HER2 in whole slide images (WSI) which would reduce the number of cases that need to be further processed by the FISH method. This approach is based on morphological techniques which discriminate between amplified and nonamplified cases with high accuracy and decreases the equivocal cases. Experimental results on real datasets demonstrate the wide applicability and high accuracy of our approach in validating of IHC analysis in WSI.

Download Full Text

Keywords

Digital pathology, HER2, Immunohistology, Computer-aided diagnosis, Whole-slide imaging.

Quantitative output variables

The detailed results correspond to an almost perfect agreement between visual IHC and automated image analysis which allows us to simplify the comparison as indicated in TABLE 1. Among the 300 ROI, the 140 (46%) slides were evaluated as IHC negative (score 0/1+) by the proposed method, however, 127 (42%) slides evaluated as IHC negative by visual IHC whereas all FISH tests were negative. Among 50 equivocal cases which are scored 2+ by visual IHC, 24 of them are categorized as 0/1+ and 3+ by the proposed method. In 26 (8%) cases both approaches agreed in equivocal cases (score 2+). All the 123 (41%) cases which considered as 3+ cases by the pathologist are also positive in our method and the FISH results were also positive. Furthermore, 11 (3%) slides with positive FISH scores are counted as positive cases which are evaluated as equivocal cases in pathologist visual scoring. By the presented method only 26 slides should be further analyzed by the FISH test. Cohon's kappa is 0.86, that shows a good agreement between visual IHC and the proposed method.

Workflow

References

[1] Robert A Smith, Kimberly S Andrews, Durado Brooks, Stacey A Fedewa, Deana Manassaram-Baptiste, Debbie Saslow, Otis W Brawley, and Richard C Wender. Cancer screening in the united states, 2017: a review of current american cancer society guidelines and current issues in cancer screening. CA: a cancer journal for clinicians, 67(2):100–121, 2017.
[2] Magali Lacroix-Triki, Simone Mathoulin-Pelissier, Jean-Pierre Ghnassia, Gaetan Macgrogan, Anne Vincent-Salomon, Véronique Brouste, Marie-Christine Mathieu, Pascal Roger, Frédéric Bibeau, Jocelyne Jacquemier, et al. High inter-observer agreement in immunohistochemical evaluation of her-2/neu expression in breast cancer: a multicentre gefpics study. European Journal of Cancer, 42(17):2946–2953, 2006.
[3] Melissa Cregger, Aaron J Berger, and David L Rimm. Immunohistochemistry and quantitative analysis of protein expression. Archives of pathology & laboratory medicine, 130(7):1026–1030, 2006.
[4] Farzad Ghaznavi, Andrew Evans, Anant Madabhushi, and Michael Feldman. Digital imaging in pathology: whole-slide imaging and beyond. Annual Review of Pathology: Mechanisms of Disease, 8:331–359, 2013.
[5] Patrick L Fitzgibbons, David L Page, Donald Weaver, Ann D Thor, D Craig Allred, Gary M Clark, Stephen G Ruby, Frances O'Malley, Jean F Simpson, James L Connolly, et al. Prognostic factors in breast cancer: College of american pathologists consensus statement 1999. Archives of pathology & laboratory medicine, 124(7):966–978, 2000.
[6] DJ Slamon, G Clark, S Wong, W Levin, A Ullrich, and W McGuire. Human breast cancer: correlation of relapse and. science, 3798106(177):235, 1987.
[7] Jeffrey S Ross and JA Fletcher. Her-2/neu (c-erb-b2) gene and protein in breast cancer. American journal of clinical pathology, 112(1 Suppl 1):S53–67, 1999.
[8] Salar Razavi, Gökhan Hatipo˘glu, and Hülya Yalçın. Automatically diagnosing her2 amplification status for breast cancer patients using large fish images. In Signal Processing and Communications Applications Conference (SIU), 2017 25th, pages 1–4. IEEE, 2017.
[9] Antonio C Wolff, M Elizabeth H Hammond, David G Hicks, Mitch Dowsett, Lisa M McShane, Kimberly H Allison, Donald C Allred, John MS Bartlett, Michael Bilous, Patrick Fitzgibbons, et al. Recommendations for human epidermal growth factor receptor 2 testing in breast cancer: American society of clinical oncology/ college of american pathologists clinical practice guideline update. Journal of clinical oncology, 31(31):3997–4013, 2013.
[10] Henrik O Helin, Vilppu J Tuominen, Onni Ylinen, Heikki J Helin, and Jorma Isola. Free digital image analysis software helps to resolve equivocal scores in her2 immunohistochemistry. Virchows Archiv, 468(2):191–198, 2016.
[11] M Milagro Fernández-Carrobles, Gloria Bueno, Marcial García- Rojo, Lucía González-López, Carlos López, and Oscar Déniz. Automatic quantification of ihc stain in breast tma using colour analysis. Computerized Medical Imaging and Graphics, 61:14– 27, 2017.