A hardware approach for histological and histopathological digital image stain normalization.
Mircea-Sebastian Şerbănescu, Iancu Emil Pleşea · PubMed · 2015
Advances in technology made the migration of pathological diagnosis to digital slides possible. As the need for objectivity and automation emerged, new computer software algorithms were proposed. Computer algorithms demand accurate color and intensity values in order to provide reliable results. The tissue samples undergo several processing steps from histological preparation to digitalization, which cannot be completely standardized. Thus, non-standardized input data generates unreliable output data. In this article, we discuss a new computational normalization algorithm for histopathological stained slides that uses a hardware color marker. The marker is added to the glass slide together with the tissue section, exposed to all the processing steps and altered in the same manner as the biological material of interest, thus becoming a solid color marker for image normalization. The results of the proposed method are numerically and perceptually tested in order to prove the advantages of the method. We conclude that our combined hardware-software technique for staining normalization of digital slides is superior to the existing methods based on only software normalization, and that its implementation will tackle not only the acquisition errors but also the technical errors that may occur during the staining process.