Submitted for Publication

Kang, X. J., T. J. Herron, et al. (2009). "Validation of the Anisotropy Index Ellipsoidal Area Ratio (EAR) in Diffusion Tensor Imaging."

A new diffusion anisotropy index, Ellipsoidal Area Ratio (EAR), was described recently and proved to be less noise-sensitive than fractional anisotropy (FA) by theory and simulation. Here we show that EAR has higher signal to noise ratios than FA in average DTI data from 40 normal subjects. EAR was also more sensitive than FA in detecting white matter abnormalities in a patient with widespread diffuse axonal injury. Monte Carlo simulation showed that EAR’s mean values are more biased by noise than FA when anisotropy is small, both for single fiber tracts and when fiber tracts cross. However, the improved signal to noise ratio of EAR relative to FA suggests that EAR may be a superior measure of anisotropy both in quantifying both deep white matter with relatively uniform fiber tracts and pericortical white matter structure with relatively low anisotropy and fiber crossings.

Yund, E. W. and D. L. Woods (2009). "Content and Procedural Learning in Repeated Sentence Tests of Speech Perception."

Objectives. Repeated testing of speech perception is critical in evaluating the benefits of hearing aids and auditory rehabilitation, but learning the test can masquerade as improved perception. A large learning effect was reported in a previous study of the speech reception threshold (SRT) in quiet, and, surprisingly, the performance improvement occurred not only on sentences that were repeated but also on sentences that the listeners had never heard before. The objective of the first experiment was to determine the stability of repeated measures with the Hearing in Noise Test (HINT) when some sentences were repeated and others were not. The objective of the second experiment was to determine the stability of the HINT and the QuickSIN when no sentences were repeated across test sessions.

Design. In Exp.1, eight normal-hearing participants completed five HINT sessions on separate days that included repeated sentence lists, while the first and last sessions also included unique sentence lists. The intensity of the noise was 65 dB (A). In Exp. 2, 23 normal-hearing participants completed three sessions of the HINT and the QuickSIN with unique sentence lists used on each day of testing. The HINT noise and the QuickSIN sentences were presented at 70 dB (A).

Results. In Exp. 1, SRTs for repeated sentence lists improved by an average of 2.7 dB, while unique lists showed insignificant change across sessions. Content learning was vulnerable to forgetting at a rate equivalent to -0.3 dB per day. These results demonstrate content learning but not procedural learning with the HINT and suggest that the HINT provides stable measures of speech perception in noise when sentences are not repeated. In Exp. 2, small but significant improvements were seen for HINT SRTs (0.2 dB per session), while QuickSIN SRTs showed an even smaller improvement (0.1 dB per session) that failed to reach statistical significance.

Conclusions. Both the HINT and the QuickSIN provide stable and sensitive measures of speech perception over time. The results with the HINT and QuickSIN at moderate noise levels differ from previous results with another sentence test in quiet where SRTs improved 6-9 dB over five sessions for both repeated and unique sentences. Differences in test procedures appear to account for the difference in SRT stability. In particular, testing in quiet may have facilitated thresholds gains due to procedural learning for quiet listening, but similar gains were not possible at moderate noise intensities.

Woods, D. L. and E. W. Yund (2009). "Consonant identification in nonsense syllables and words."

The California Syllable Test (CST) was designed to assess the ability of individual subjects to identify consonants in consonant-vowel-consonant (CVC) syllables presented in variable-intensity speech-spectrum noise. Sixteen young subjects with normal hearing identified 20 different initial and final consonants in 720 CVCs, randomly sampled from a corpus 9,600 CVC tokens that included both words and nonsense syllables. Signal-to-noise ratios (SNRs) were set to produce mean d’ of 2.20 for each consonant, with noise levels adjusted during the mid-vowel segment to provide appropriate SNRs for initial and final consonants differing in intrinsic identifiability. The 1-hr CST provided measures of consonant identification ability that were precise to approximately 0.5 dB. Consonants were identified more accurately in words than nonsense syllables, particularly at high SNRs. There were also significant interactions in the processing of initial and final consonants. Small learning effects were observed over the three days of testing. Moreover, performance on the CST accurately predicted sentence reception thresholds measured with two tests, the HINT and QuickSIN. The CST appears to be a promising tool for evaluating the effects of audiological rehabilitation.

Turken, A., T. J. Herron, et al. (2009). "Multimodal surface-based morphometry reveals diffuse cortical atrophy in traumatic brain injury: a case study."

While post-mortem investigations and animal studies suggest that traumatic brain injury (TBI) produces widespread cortical atrophy, TBI patients with significant cognitive deficits often present with negative brain imaging finding. Here, we applied automated surfaced-based morphometry (SBM) to quantify cortical abnormalities in a chronically impaired TBI patient. By combining information from high-resolution structural MRI and diffusion tensor imaging we found widespread but regionally specific co-localized abnormalities in cortical thickness, neuropil density and pericortical white matter integrity which were not evident on standard clinical MRI scans. The findings were replicated across imaging sessions. SBM is a promising new tool for quantifying cortical pathology in neurological disease.

Woods, D. L., E. W. Yund, et al. (2009). "Quantifying consonant identification in speech spectrum noise."

Sixteen young subjects with normal hearing identified 20 different initial and final consonants of consonant-vowel-consonant (CVC) syllables randomly sampled from a corpus 9,600 tokens. Psychometric functions were obtained for each consonant at three signal-to-noise ratios (SNRs) spanning a range of 12 dB and centered at a baseline SNR adjusted to produce a d’ of 2.20 for each consonant. Noise levels adjusted during the mid-vowel segment to provide appropriate SNRs for initial and final consonants. The SNRs needed to equate the d’ measures for different consonants varied by more than 40 dB and the slopes of d’ psychometric functions varied over a 4-fold range and. Large differences were also found in response criteria for different consonants. Confusion analysis showed frequent confusions in place of articulation, followed by place + manner, manner, and voicing, with disproportionate reductions in voicing and combined-feature errors as SNRs increased. With the exception of liquids, most consonants were more easily perceived in the initial than final syllable position. Different patterns of confusion were also found for initial and final consonants, with a greater incidence of place confusions found for final consonants.

Kang, X. J., T. J. Herron, et al. (2009). "Diffusion Properties Of Cortical Gray Matter And Pericortical White Matter: Regional Variations, Hemispheric Asymmetries, And Methodological Issues."

The characterization of average diffusion properties of cortical gray and white matter in 3D space is complicated by the variability introduced by intersubject differences in the relative locations of gyri and sulci. Here we introduce a new method is to visualize the average diffusion properties of cortical gray matter and pericortical white matter on cortical surface maps. Mean diffusivity (MD) and fractional anisotropy (FA) data were gathered from 60 subjects and co-registered with high-resolution T1 images that had been inflated and co-registered to a spherical coordinate system using FreeSurfer and then realigned to a hemispherically unified coordinate system. Five cortical surfaces in parallel with the white/gray junction were defined. Diffusion properties were analyzed on each of the surfaces and displayed using equal-area Mollweide projections. The intra-subject reliability of FA and MD measures appears to be sufficiently good relative to a healthy population’s intersubject variation to provide reasonable power for group studies. The differences between the left and right hemispheres for these measures are also discussed as are lobar level differences. The influence of CSF susceptibility and cortical thickness were investigates. We find that FA is mostly unaffected by these confounds but MD needs to be analyzed using these confounds as covariates or using previous suggested techniques.