Advanced Microscopic Imaging

The primary scientific objective of this work is to explore the feasibility of imaging intrinsic optical signals in order to investigate the spatial and temporal dynamics of neuronal population activity.  Our approach is to characterize the spatial selectivity, temporal resolution, and spectral signature of several optical responses associated with the activation of neural tissue, including localized hemodynamic responses, metabolism-dependent changes in endogenous chromophores, and rapid changes in scattering and birefringence.  This work employs  high performance imagers based on  Los Alamos designs.  The system will be optimized to achieve the demanding mix of speed, sensitivity and dynamic range required for optical neurophysiological studies.

Optical techniques hold tremendous promise for the visualization of patterns of neural activation on a range of spatial and temporal scales.  A variety of molecular probes have been developed which respond to changes in cell membrane potential or of the cytoplasmic ionic environment with changes in their absorbance or fluorescence characteristics.  A number of laboratories have reported intrinsic optical signals associated with neural activity:  changes in birefringence, scattering, absorbance or reflection.  Some of the responses are relatively slow and are associated with metabolic or physiological changes over several seconds.  These include changes in regional cerebral blood flow (and corresponding changes in blood volume and oxygenation) that are the basis for functional neuroimaging using PET, SPECT, functional MRI and video difference imaging.

Other optical responses including birefringence changes and some scattering signals are observed on the millisecond time scale.  However, there are definite problems with the state of the art.  Both dye-dependent and intrinsic signals are small;  the responses associated with activity are generally  0.1-.001% of the background level.  In order to achieve the sensitivity and signal to noise performance required to capture such images, investigators have employed cooled, slow-scan CCD cameras.  Such devices typically provide temporal resolution on the order of hundreds of milliseconds to seconds. Because fast intrinsic signals or responses visualized with voltage sensitive dyes are small and transient, they are difficult to detect with slow-scan CCDs. Although most of the power in electrophysiological measurements of neural populations is at frequencies between 10 and 100 Hz, firing rates of individual neurons may approach 1 KHz.  Thus, existing technical approaches using slow scan or standard video devices (0.5-30 Hz) are adequate for studying the static functional organization of cortex, but are inadequate for monitoring the dynamics of neural transactions on their millisecond timescale.

Our work explores the feasibility of imaging intrinsic optical signals that closely track neuronal electrical responses. Such responses have been observed using single channel detectors and small photodiode arrays, but  have not previously been captured using electronic imaging technologies.  Imaging will be essential to investigate presumptive dynamic spatial/temporal patterns of activation, particularly for responses that are not time-locked to a stimulus or external event.  Given the growing interest in the role of cellular population dynamics in neural computation, this is a problem that is both very timely and is not adequately addressed by existing methodologies.  Our work will motivate and facilitate the development of advanced imaging strategies for macroscopic, microscopic and endoscopic applications. A flexible, general capability for high performance confocal and spectroscopic microscopic imaging will enable many other studies of cellular physiology and microanatomy, and will greatly enhance the utility of imaging techniques for a range of scientific and clinical applications.
 

Confocal Microscopy:  The conventional approach to confocal imaging involves the use of a pinhole aperture and a high quality imaging lens in both the illumination and imaging paths.  The acquisition of the image typically is by mechanical scanning of the specimen, or more commonly of the aperture(s).  Initial commercial systems appearing in the mid 1980s utilized laser illumination and photomultiplier detectors.  These were complex, inflexible, slow and expensive systems.  More recent designs based on spinning disks incorporating multiple pinhole or slit apertures have reduced system costs by a factor of two and allow video-rate acquisition or direct viewing, but suffer from reflected light from the disk surface and from insensitivity due to inefficient coupling of illumination light into the sample.

We have developed a novel approach for confocal microscopy that exploits the technical characteristics of available illumination and detection technologies to produce an imager with a number of advantages.  This system has two key components: 1) an area sensor such as a CCD imager, with a "virtual aperture" synthesized during or after sensor readout, and 2) an electronically scanned illumination subsystem.  The benefits of this design are reduced cost, faster image acquisition, improved efficiency and sensitivity, much improved reliability and ruggedness (since the system requires no moving parts).  The approach is very flexible; the same system can be used for confocal or standard video imaging (facilitating image field surveys or focusing) and different image contrast mechanisms (such as reflected light or darkfield) can be synthesized from the same dataset.  Work over the past three yars has produced a working prototype imager and has demonstrated that the system is indeed confeocal.  We have demonstrated the feasibility of the basic approach to image collection and reconstruction, and of a specific implementation (see the image at the top of this page).  Continuing work will push the technical limits of system performance.