We present low-resolution 5.5-35 μm spectra for 103 galaxies from the 12 μm Seyfert sample, a complete unbiased 12 μm flux limited sample of local Seyfert galaxies selected from the IRAS Faint Source Catalog, obtained with the Infrared Spectrograph (IRS) on-board Spitzer Space Telescope. For 70 of the sources observed in the IRS mapping mode, uniformly extracted nuclear spectra are presented for the first time. We performed an analysis of the continuum emission, the strength of the polycyclic aromatic hydrocarbon (PAH) and astronomical silicate features of the sources. We find that on average, the 15-30 μm slope of the continuum is α15–30 = –0.85 ± 0.61 for Seyfert 1s and –1.53 ± 0.84 for Seyfert 2s, and there is substantial scatter in each type. Moreover, nearly 32% of the Seyfert 1s, and 9% of the Seyfert 2s, display a peak in the mid-infrared spectrum at 20 μm, which is attributed to an additional hot dust component. The PAH equivalent width decreases with increasing dust temperature, as indicated by the global infrared color of the host galaxies. However, no statistical difference in PAH equivalent width is detected between the two Seyfert types, 1 and 2, of the same bolometric luminosity. The silicate features at 9.7 and 18 μm in Seyfert 1 galaxies are rather weak, while Seyfert 2s are more likely to display strong silicate absorption. Those Seyfert 2s with the highest silicate absorption also have high infrared luminosity and high absorption (hydrogen column density NH > 1023 cm–2) as measured from the X-rays. Finally, we propose a new method to estimate the active galactic nucleus contribution to the integrated 12 μm galaxy emission, by subtracting the "star formation" component in the Seyfert galaxies, making use of the tight correlation between PAH 11.2 μm luminosity and 12 μm luminosity for star-forming galaxies.
The first high resolution Spitzer IRS 9-37um spectra of 29 Seyfert galaxies (about one quarter) of the 12um Active Galaxy Sample are presented and discussed. The high resolution spectroscopy was obtained with corresponding off-source observations. This allows excellent background subtraction, so that the continuum levels and strengths of weak emission lines are accurately measured. The result is several new combinations of emission line ratios, line/continuum and continuum/continuum ratios that turn out to be effective diagnostics of the strength of the AGN component in the IR emission of these galaxies. The line ratios [NeV]/[NeII], [OIV]/[NeII], already known, but also [NeIII]/[NeII] and [NeV]/[SiII] can all be effectively used to measure the dominance of the AGN. We extend the analysis, already done using the 6.2um PAH emission feature, to the equivalent width of the 11.25um PAH feature, which also anti-correlates with the dominance of the AGN. We measure that the 11.25um PAH feature has a constant ratio with the H_2 S(1) irrespective of Seyfert type, approximately 10 to 1. Using the ratio of accurate flux measurements at about 19um with the two spectrometer channels, having aperture areas differing by a factor 4, we measured the source extendness and correlated it with the emission line and PAH feature equivalent widths. The extendness of the source gives another measure of the AGN dominance and correlates both with the EWs of [NeII] and PAH emission. Using the rotational transitions of H$_2$ we were able to estimate temperatures (200-300K) and masses (1-10 x 10^6 M_sun), or significant limits on them, for the warm molecular component in the galaxies observed.
We observed the far-IR fine-structure lines of 26 Seyfert galaxies with the Herschel-PACS spectrometer. These observations are complemented with Spitzer Infrared Spectrograph and Herschel SPIRE spectroscopy. We used the ionic lines to determine electron densities in the ionized gas and the [C I] lines, observed with SPIRE, to measure the neutral gas densities, while the [O I] lines measure the gas temperature, at densities below ~104 cm–3. Using the [O I]145 μm/63 μm and [S III]33/18 μm line ratios, we find an anti-correlation of the temperature with the gas density. Various fine-structure line ratios show density stratifications in these active galaxies. On average, electron densities increase with the ionization potential of the ions. The infrared lines arise partly in the narrow line region, photoionized by the active galactic nucleus (AGN), partly in H II regions photoionized by hot stars, and partly in photo-dissociated regions. We attempt to separate the contributions to the line emission produced in these different regions by comparing our observed emission line ratios to theoretical values. In particular, we tried to separate the contribution of AGNs and star formation by using a combination of Spitzer and Herschel lines, and we found that besides the well-known mid-IR line ratios, the line ratio of [O III]88 μm/[O IV]26 μm can reliably discriminate the two emission regions, while the far-IR line ratio of [C II]157 μm/[O I]63 μm is only able to mildly separate the two regimes. By comparing the observed [C II]157 μm/[N II]205 μm ratio with photoionization models, we also found that most of the [C II] emission in the galaxies we examined is due to photodissociation regions.
Objectives: To evaluate the accuracy of a data-driven approach, such as machine learning classification, in predicting disability progression in MS.Methods: We analyzed structural brain images of 163 subjects diagnosed with MS acquired at two different sites. Participants were followed up for 2–6 years, with disability progression defined according to the expanded disability status scale (EDSS) increment at follow-up. T2-weighted lesion load (T2LL), thalamic and cerebellar gray matter (GM) volumes, fractional anisotropy of the normal appearing white matter were calculated at baseline and included in supervised machine learning classifiers. Age, sex, phenotype, EDSS at baseline, therapy and time to follow-up period were also included. Classes were labeled as stable or progressed disability. Participants were randomly chosen from both sites to build a sample including 50% patients showing disability progression and 50% patients being stable. One-thousand machine learning classifiers were applied to the resulting sample, and after testing for overfitting, classifier confusion matrix, relative metrics and feature importance were evaluated.Results: At follow-up, 36% of participants showed disability progression. The classifier with the highest resulting metrics had accuracy of 0.79, area under the true positive versus false positive rates curve of 0.81, sensitivity of 0.90 and specificity of 0.71. T2LL, thalamic volume, disability at baseline and administered therapy were identified as important features in predicting disability progression. Classifiers built on radiological features had higher accuracy than those built on clinical features.Conclusions: Disability progression in MS may be predicted via machine learning classifiers, mostly evaluating neuroradiological features.