Little attention has been paid so far to physiological signals for emotion recognition compared to audiovisual emotion channels such as facial expression or speech. This paper investigates the potential of physiological signals as reliable channels for emotion recognition. All essential stages of an automatic recognition system are discussed, from the recording of a physiological dataset to a feature-based multiclass classification. In order to collect a physiological dataset from multiple subjects over many weeks, we used a musical induction method which spontaneously leads subjects to real emotional states, without any deliberate lab setting. Four-channel biosensors were used to measure electromyogram, electrocardiogram, skin conductivity and respiration changes. A wide range of physiological features from various analysis domains, including time/frequency, entropy, geometric analysis, subband spectra, multiscale entropy, etc., is proposed in order to find the best emotion-relevant features and to correlate them with emotional states. The best features extracted are specified in detail and their effectiveness is proven by classification results. Classification of four musical emotions (positive/high arousal, negative/high arousal, negative/low arousal, positive/low arousal) is performed by using an extended linear discriminant analysis (pLDA). Furthermore, by exploiting a dichotomic property of the 2D emotion model, we develop a novel scheme of emotion-specific multilevel dichotomous classification (EMDC) and compare its performance with direct multiclass classification using the pLDA. Improved recognition accuracy of 95\% and 70\% for subject-dependent and subject-independent classification, respectively, is achieved by using the EMDC scheme.

Cells infected with mammalian reoviruses often contain large perinuclear inclusion bodies, or "factories," where viral replication and assembly are thought to occur. Here, we report a viral strain difference in the morphology of these inclusions: filamentous inclusions formed in cells infected with reovirus type 1 Lang (T1L), whereas globular inclusions formed in cells infected with our laboratory's isolate of reovirus type 3 Dearing (T3D). Examination by immunofluorescence microscopy revealed the filamentous inclusions to be colinear with microtubules (MTs). The filamentous distribution was dependent on an intact MT network, as depolymerization of MTs early after infection caused globular inclusions to form. The inclusion phenotypes of T1L ؋ T3D reassortant viruses identified the viral M1 genome segment as the primary genetic determinant of the strain difference in inclusion morphology. Filamentous inclusions were seen with 21 of 22 other reovirus strains, including an isolate of T3D obtained from another laboratory. When the 2 proteins derived from T1L and the other laboratory's T3D isolate were expressed after transfection of their cloned M1 genes, they associated with filamentous structures that colocalized with MTs, whereas the 2 protein derived from our laboratory's T3D isolate did not. MTs were stabilized in cells infected with the viruses that induced filamentous inclusions and after transfection with the M1 genes derived from those viruses. Evidence for MT stabilization included bundling and hyperacetylation of ␣-tubulin, changes characteristically seen when MT-associated proteins (MAPs) are overexpressed. Sequencing of the M1 segments from the different T1L and T3D isolates revealed that a single-amino-acid difference at position 208 correlated with the inclusion morphology. Two mutant forms of 2 with the changes Pro-208 to Ser in a background of T1L 2 and Ser-208 to Pro in a background of T3D 2 had MT association phenotypes opposite to those of the respective wild-type proteins. We conclude that the 2 protein of most reovirus strains is a viral MAP and that it plays a key role in the formation and structural organization of reovirus inclusion bodies.

Cells infected with mammalian orthoreoviruses contain large cytoplasmic phase-dense inclusions believed to be the sites of viral replication and assembly, but the morphogenesis, structure, and specific functions of these "viral factories" are poorly understood. Using immunofluorescence microscopy, we found that reovirus nonstructural protein NS expressed in transfected cells forms inclusions that resemble the globular viral factories formed in cells infected with reovirus strain type 3 Dearing from our laboratory (T3D N ). In the transfected cells, the formation of NS large globular perinuclear inclusions was dependent on the microtubule network, as demonstrated by the appearance of many smaller NS globular inclusions dispersed throughout the cytoplasm after treatment with the microtubule-depolymerizing drug nocodazole. Coexpression of NS and reovirus protein 2 from a different strain, type 1 Lang (T1L), which forms filamentous viral factories, altered the distributions of both proteins. In cotransfected cells, the two proteins colocalized in thick filamentous structures. After nocodazole treatment, many small dispersed globular inclusions containing NS and 2 were seen, demonstrating that the microtubule network is required for the formation of the filamentous structures. When coexpressed, the 2 protein from T3D N also colocalized with NS, but in globular inclusions rather than filamentous structures. The morphology difference between the globular inclusions containing NS and 2 protein from T3D N and the filamentous structures containing NS and 2 protein from T1L in cotransfected cells mimicked the morphology difference between globular and filamentous factories in reovirusinfected cells, which is determined by the 2-encoding M1 genome segment. We found that the first 40 amino acids of NS are required for colocalization with 2 but not for inclusion formation. Similarly, a fusion of NS amino acids 1 to 41 to green fluorescent protein was sufficient for colocalization with the 2 protein from T1L but not for inclusion formation. These observations suggest a functional difference between NS and NSC, a smaller form of the protein that is present in infected cells and that is missing amino acids from the amino terminus of NS. The capacity of NS to form inclusions and to colocalize with 2 in transfected cells suggests a key role for NS in forming viral factories in reovirus-infected cells.The replication and assembly of viruses are often concentrated in specific locations within infected cells, such as on the actin cytoskeleton for human parainfluenza virus type 3 (13), on the outer mitochondrial membranes for flock house virus (24), in cytoplasmic inclusions for vaccinia virus (39), and in nuclear inclusions for herpes simplex virus (32). The nonfusogenic mammalian orthoreoviruses (reoviruses) are believed to replicate and assemble in cytoplasmic phase-dense inclusions in infected cells (31). These inclusions contain viral doublestranded RNA (34), viral proteins (9, 31), partially and fully assembled viral particles (...