Abstract — In this paper we address the issue of classifying the fluorescent intensity of a sample in Indirect Immuno-Fluorescence(IIF). Since IIF is a subjective, semi-quantitative test in its verynature, we discuss a strategy to reliably label the image data set byusing the diagnoses performed by different physicians. Then, wediscuss image pre-processing, feature extraction and selection.Finally, we propose two ANN-based classifiers that can separateintrinsically dubious samples and whose error tolerance can beflexibly set. Measured performance shows error rates less than 1%,which candidates the method to be used in daily medical practiceeither to perform pre-selection of cases to be examined, or to act as asecond reader. Keywords — Artificial neural networks, computer aided diagnosis,image classification, indirect immuno-fluorescence, patternrecognition. I. I NTRODUCTION ONNECTIVE tissue diseases (CTD) are autoimmunedisorders of unknown aetiology characterized by achronic inflammatory process involving connective tissues. Acommon marker of CTD, although it occurs at a variable ratein the different forms, is the presence of serum antinuclearautoantibodies (ANA) [1]. The recommended method forANA testing is indirect immunofluorescence (IIF) [2], [3]. InIIF a serum sample is tested with a substrate containing aspecific antigen. Fluorochrome conjugated anti humanimmunoglobulin antibodies reveal the antigen antibodyreaction, and the slide is examined at fluorescencemicroscope.The readings in IIF are subjected to interobserver variabilitythat limits the reproducibility of the method. To date, thehighest level of automation in IIF tests is the preparation ofslides with robotic devices performing dilution, dispensationand washing operations [4], [5]. The development of a systemthat can offer a support to physician decision is therefore anevident medical demand [3].In this paper we focus on the development of a system thatwould be able to classify the fluorescent intensity of IIF