Analysis and development of computer-based solutions for IIF applications

The antinuclear antibodies (ANAs) can be found in rheumatic and non rheumatic conditions, but their importance depends strongly on the clinical suspect of ANA Associated Rheumatic Diseases (AARDs). The indirect immunofluorescence (IIF) on human epithelial cells (HEp-2) is considered the gold standard for ANA testing however, it is strongly influenced by several issues. This manuscript aims to provide significant contributions in the IIF research, in the biomedical field, which is a bridge between engineering and medicine. For this reason will be presented engineering solutions designed for medical applications. We discuss different issues that should be considered in a systematic way to provide effective and viable solution in IIF research: image segmentation; inter-observer reading IIF variability study, fluorescence intensity classification. With respect to the first issue, this manuscript tackles with HEp-2 cells segmentation, i.e. one of the open issue in HEp-2 image analysis, providing contributions into four main directions: we survey the literature on HEp-2 image segmentation dividing the works in the literature into three main categories. We introduce a segmentation pipeline employing a deformable model whose energy minimization, differently from the the classical snakes and active contour models, it can detect objects with boundaries not necessarily defined by gradient since the stopping term is related to a particular segmentation of the image rather than on a derivative quantity. We make available new 24 HEp-2 images containing 985 cells, thus enlarging the reference image dataset publicly available that now contains 1964 cells. We experimentally benchmark the proposed segmentation method with most representative methods in the literature using the enlarged image repository, revealing that a segmentation pipeline using active contour better copes with the variability in cell appearance due to different fluorescence intensities and staining patterns as well as to irregular illumination. Let's now turn the attention to inter-observer reading IIF variability study, the results archived in our analysis show that is necessary and crucial a standardization of ANA testing and how it is far from being completed. In fact the IIF method is labor intensive, subjective and prone to reader bias. A proposal of tight procedures both for users (working protocol) and for manufacturers (assay kits production) and a large use of international standards and independent calibrators could make easily the standardization process. Therefore, an international standardization of the HEp-2 assay kit (fixative used, conjugate etc.) and the introduction of a CAD system may represent two important elements of this process reducing operator time of work and improving the diagnostic efficacy. CAD systems may represent one of the most important novel elements of harmonization in the autoimmunity field, reducing intra- and inter-laboratory variability in a new vision of the diagnostic autoimmune platform. Furthermore, our study demonstrates that there can be marked discordance between ANA results drawn from different laboratories mostly for intensity and multiple patterns classification. Laboratories agreement improves using digital images and comparing each single human evaluation to a potential reference (gold standard) data that is essential for use CAD systems in routine work lab. Finally, with respect to fluorescence intensity classification, the results archived by the proposed system show that the method of Maximum Observation allows a gold standard that maximizes the agreement between expert physicians minimizing the probability of incorrect label on training sample and also that the system archives performances comparable to those archived by human experts and superior, to those obtained by method at the state-of-art.