Faculty Bibliography

Multimedia lives with images; medical images are born from digital imaging. A physician's multimedia workstation cannot exist without tools for manipulating images, performing measurements and, generally speaking, extracting and collecting pieces of information from the available data. Image analysis and computer vision constitute a wide and rapidly evolving field. This paper is intended as an introductory document for medical imaging researchers and practitioners wishing an overview of recent developments and trends. The major lines of activities in the domain are presented, under the common framework of a processing pipeline. After a presentation of the various stages of this pipeline, current subjects of research are indicated.

Neuroanatomical and histological findings from post-mortem brains, as well as in vivo findings from magnetic resonance imaging (MRI) studies, suggest the presence of morphologic temporal lobe abnormalities in schizophrenia. To determine whether or not sulco-gyral pattern abnormalities in the temporal lobe could be detected in vivo, we applied computerized surface rendering techniques to MR data sets in order to make both qualitative and quantitative analyses of three-dimensional reconstructions of the temporal and frontal cortex in 15 schizophrenic patients and 15 normal controls. The qualitative analysis, based on a visual classification of the temporal lobe sulco-gyral pattern by 4 raters blind to diagnosis, showed that in schizophrenics there was a more vertical orientation to the sulci in the left temporal lobe, with an interrupted course of sulci due to gyri coursing across the sulci. Normal controls, in contrast, showed a more horizontal orientation with no interruptions. These findings were supported by the quantitative analysis, where more sulcal lines, representing an interrupted course of sulci, were observed in the temporal lobes (more pronounced on the left) in schizophrenics than in normal controls. These data suggest that some of the abnormalities observed in schizophrenia may have their origin in alterations occurring during the course of neurodevelopment when the sulco-gyral pattern is determined.

In contrast to acquisition-based noise reduction methods a postprocess based on anisotropic diffusion is proposed. Extensions of this technique support 3-D and multiecho magnetic resonance imaging (MRI), incorporating higher spatial and spectral dimensions. The procedure overcomes the major drawbacks of conventional filter methods, namely the blurring of object boundaries and the suppression of fine structural details. The simplicity of the filter algorithm permits an efficient implementation, even on small workstations. The efficient noise reduction and sharpening of object boundaries are demonstrated by applying this image processing technique to 2-D and 3-D spin echo and gradient echo MR data. The potential advantages for MRI, diagnosis, and computerized analysis are discussed in detail.

A computerized system for processing spin-echo magnetic resonance (MR) imaging data was implemented to estimate whole brain (gray and white matter) and cerebrospinal fluid volumes and to display three-dimensional surface reconstructions of specified tissue classes. The techniques were evaluated by assessing the radiometric variability of MR volume data and by comparing automated and manual procedures for measuring tissue volumes. Results showed (a) the homogeneity of the MR data and (b) that automated techniques were consistently superior to manual techniques. Both techniques, however, were affected by the complexity of the structure, with simpler structures (eg, the intracranial cavity) showing less variability and better spatial correlation of segmentation results between raters. Moreover, the automated techniques were completed for whole brain in a fraction of the time required to complete the equivalent segmentation manually. Additional evaluations included interrater reliability and an evaluation that included longitudinal measurement, in which one subject was imaged sequentially 24 times, with reliability computed from data collected by three raters over 1 year. Results showed good reliability for the automated segmentation procedures.