Applying artificial neural network models to clinical decision making.

Because psychological assessment typically lacks biological gold standards, it traditionally has relied on clinicians' expert knowledge. A more empirically based approach frequently has applied linear models to data to derive meaningful constructs and appropriate measures. Statistical inferences are then used to assess the generality of the findings. This article introduces artificial neural networks (ANNs), flexible nonlinear modeling techniques that test a model's generality by applying its estimates against "future" data. ANNs have potential for overcoming some shortcomings of linear models. The basics of ANNs and their applications to psychological assessment are reviewed. Two examples of clinical decision making are described in which an ANN is compared with linear models, and the complexity of the network performance is examined. Issues salient to psychological assessment are addressed.

Hemispheric language dominance in children demonstrated by functional magnetic resonance imaging.

The objective of this study was to demonstrate hemispheric language dominance in normal children. Fifteen normal children were evaluated with functional magnetic resonance imaging (MRI) using an age-related silent word spelling paradigm. The data were analyzed with the cross-correlation method, and lateralization indices were calculated in language regions as determined by Talairach coordinates. Activation foci were detected in the left inferior frontal area and were strongly lateralized, with language lateralization indices of 0.74 +/- 0.21 (age 7-12 years, nine subjects), and 0.79 +/- 0.18 (13-18 years, six subjects). The indices were similar to those for adults (0.83 +/- 0.21, four subjects). Our study established that language is strongly lateralized to the left hemisphere in children as young as 7 years of age.

A half-Fourier gradient echo technique for dynamic MR imaging.

Recently we developed the simultaneous dual FLASH (SDFLASH) pulse sequence that simultaneously obtains sequential images from the brain and the internal-carotid arteries in the neck with 1-sec temporal resolution using a standard MR scanner. The high temporal resolution (1 sec) of the SDFLASH technique was achieved partly by using a low number of phase-encoding views which thereby limited our in-plane spatial resolution to 6.25 x 3.12 mm pixels. To overcome this limitation we have developed a calibration technique which corrects distortions in signal intensity and object shape when using gradient echo half-Fourier spin warp imaging. Using this calibration technique, the operator can use the 41% decrease in scan time to either double the spatial or temporal resolution. We have successfully used this technique to acquire SDFLASH images of the head and neck with 1.0 sec temporal resolution and 3.12 x 1.6 mm spatial resolution.