Automated analysis of Indirect Immunofluorescence images is significant in the computerized detection of Autoimmune Diseases (AIDs). The recognition of particular shapes in Indirect Immunofluorescence (IIF) images is clinically associated with specific AIDs. In this work, an attempt to differentiate Homogeneous and Speckled shapes in IIF images using Neutrosophic Sets (NS) segmentation and a neural network-based classification is performed. The characteristics of NS to handle the edge boundary information of the cells is utilized. The IIF specimen images belonging to the two shapes are obtained from the public dataset. The images are subjected to illumination correction using Top-Hat transform, denoising by Split Bregman Anisotropic Total Variation and contrast enhancement with image normalization. Segmentation of cell boundaries is performed using indeterminate subset of NS. Geometric features are extracted from cell edges to assess its morphology. Multilayer Perceptron (MLP) network is employed to classify the two patterns. Results show that the indeterminacy of NS is able to segment cell edges. The geometric features are obtained to be statistically highly significant (p<0.001) between the two patterns. MLP is found to perform better with average Recall, Accuracy, and Area under the Receiver Operating Characteristic measures of 98.6%, 98.7%, and 99.8% respectively. The proposed work is found to provide better results as compared to the existing methods. Hence, this study appears to be clinically significant in the morphological investigation of specimen-level IIF pattern classification for AID detection.