Neural network and conventional classifiers for fluorescence-guided laser angioplasty.

In laser angioplasty, fluorescence spectra of targeted tissue may be used to classify the tissue as atherosclerotic or normal and guide selective laser ablation of atherosclerotic plaque. Here, the ability of the back-propagation and K-nearest neighbors techniques to classify arterial fluorescence spectra is investigated. Both methods are competitive with other classification schemes. The relative performance of variations on both techniques is used to make inferences about the geometry of the classification task.

Design and evaluation of a fiberoptic fluorescence guided laser recanalization system.

Current angioplasty techniques for recanalization of totally occluded arteries are limited by the inability to cross the occlusion and by the risk of perforation. A fiberoptic fluorescence guided laser recanalization system was developed and evaluated in vitro for recanalization of 17 human femoral or tibial totally occluded arterial segments (length 1.9-6.8 cm, diameter 2.5-6.0 mm). A 400 or 600 micron silica fiber was coupled to a helium-cadmium laser (lambda = 325 nm) for fluorescence excitation and to a holmium: YAG laser (lambda = 2.1 micron) for tissue ablation. Fluorescence was recorded during recanalization after every other holmium laser pulse. During recanalization, each arterial segment was bent 30-90 degrees with respect to the fiber to simulate arterial tortuosity. Ablation continued with fiber advancement as long as the fluorescence confirmed that the target tissue was atherosclerotic. Arterial spectra were classified as normal or atherosclerotic by an on-line computerized fluorescence classification algorithm (sensitivity 93%, specificity 95%). Normal fluorescence necessitated redirection of the fiber greater than 30 times per segment to continue recanalization. Fifteen of 17 totally occluded arteries had multiple recanalization channels created following total energy delivery of 40-1,016 Joules per segment with no angiographic or histologic evidence of laser perforation. Two heavily calcified arterial occlusions were not recanalized due to inhibition of holmium: YAG laser ablation by the recording of normal fluorescence spectra. Therefore, this fluorescence guided laser recanalization system appears safe and effective for recanalization of totally occluded arteries and merits in vivo evaluation. However, the lower sensitivity of fluorescence detection of heavily calcified plaques may limit the efficacy (but not safety) of fluorescence guided recanalization of heavily calcified occlusions.

Characterization of the site dependency of normal canine arterial fluorescence.

The difference in fluorescence between normal and atherosclerotic artery has been proposed as a feedback mechanism to guide selective laser ablation of atherosclerotic plaque. This fluorescence difference is due to the relative difference in collagen:elastin content of normal artery and atherosclerotic plaque. However, normal arteries have site-dependent variation in collagen: elastin content which may affect their fluorescence spectra. To evaluate the site dependency of normal arterial fluorescence, helium-cadmium (325 nm) laser-induced fluorescence spectra were analyzed in vitro from the ascending aorta, abdominal aorta, and carotid, femoral, renal, and coronary arteries (N = 57) of 12 normal mongrel dogs. Elastin and collagen contents were determined for a subset of these arteries (N = 15). The spectral width of normal arterial Fluorescence varied by site and correlated with the measured collagen:elastin content at each site (r = -0.84, P less than 0.005). Fluorescence spectra were decomposed into collagen and elastin spectral components by using a linear model with a least-squared error criterion fit. The derived collagen and elastin spectral coefficients correlated with the measured collagen and elastin tissue content (r = 0.75 and 0.83 respectively, P less than 0.005). Thus, the fluorescence spectra of normal arteries is site dependent and correlates with the collagen:elastin content. Therefore, spectral feedback algorithms for laser angioplasty guidance must be site specific.

Videographic tomography. I. Reconstruction with parallel-beam projection data.

An approach to video-rate tomographic image reconstruction using standard filtered backpropagation is presented. Several methods of filtering are reported along with an optical system for backprojection. The focus is on reconstruction of data acquired in a parallel-beam geometry. The methods of implementing the filtering step are discussed, and the implementation of the backpropagation step is described. Results demonstrating the feasibility of the approach are presented.

Biochemical basis for the difference between normal and atherosclerotic arterial fluorescence.

The observation that laser-induced fluorescence (LIF) spectra of atherosclerotic and normal artery are different has been proposed as the basis for guiding a "smart" laser angioplasty system. The purpose of this study was to investigate the causes of this difference in LIF. Helium-cadmium laser-induced (325 nm) fluorescence was recorded from pure samples of known constituents of normal and atherosclerotic artery including collagen, elastin, calcium, cholesterol, and glycosaminoglycans. Similarities between the LIF spectra of atherosclerotic plaque and collagen and normal aorta and elastin were noted. LIF spectroscopy was then performed on specimens of atherosclerotic aortic plaque (n = 9) and normal aorta (n = 13) and on their extracted lipid, collagen, and elastin. Lipid extraction did not significantly alter atherosclerotic plaque or normal aortic LIF, suggesting a minor contribution of lipid to arterial LIF. The LIF spectra of normal aorta wall was similar to the spectra of the extracted elastin, whereas the LIF spectra of atherosclerotic aortic plaque was similar to the spectra of the extracted collagen. These observations are consistent with the reported relative collagen-to-elastin content ratio of 0.5 for normal arterial wall and 7.3 for atherosclerotic plaque. A classification algorithm was developed to discriminate normal and atherosclerotic aortic spectra based on an elastin and collagen spectral decomposition. A discriminant score was formed by the difference of elastin and collagen (E-C) coefficients and used to classify 182 aortic fluorescence spectra. The mean E-C value was +0.83 +/- 0.04 for normal and -0.48 +/- 0.07 for atherosclerotic aorta (p less than 0.001). Classification accuracy was 92%.(ABSTRACT TRUNCATED AT 250 WORDS)

Development and evaluation of spectral classification algorithms for fluorescence guided laser angioplasty.

Laser angioplasty, or the ablation of atherosclerotic plaque using laser energy, has tremendous potential to expand the scope of nonsurgical treatment of obstructive vascular disease. Clinical laser angioplasty, however, has been hindered by an unacceptable risk of vessel perforation. Laser-induced fluorescence spectroscopy can discriminate atherosclerotic from normal artery and may therefore be capable of guiding selective plaque ablation. To assess the feasibility of utilizing spectral information to discriminate arterial tissue type, several classification algorithms were developed and evaluated. Arterial fluorescence spectra from 350 to 700 nm were obtained from 100 human aortic specimens. Seven spectral classification algorithms were developed with the following techniques: multivariate linear regression, stepwise multivariate linear regression, principal components analysis, decision plane analysis, Bayes decision theory, principal peak ratio, and spectral width. The classification ability of each algorithm was evaluated by its application to the training set and to a validation set containing 82 additional spectra. All seven spectral classification algorithms prospectively classified atherosclerotic and normal aorta with an accuracy greater than 80 percent (range: 82-96 percent). Laser angioplasty systems incorporating spectral classification algorithms may therefore be capable of detection and selective ablation of atherosclerotic plaque.

Change in laser-induced arterial fluorescence during ablation of atherosclerotic plague.

Analysis of the change in arterial fluorescence during plaque ablation may provide the basis for developing a fluorescence-guided ablation system capable of selective plaque ablation without risk of vessel perforation. Accordingly, fluorescence spectra were recorded from 91 normal and 91 atherosclerotic specimens of cadaveric human aorta. The ratio of the laser-induced fluorescence intensity at 382 nm to 430 nm (LIF ratio) was capable of classifying these specimens with an 89% accuracy with a threshold value of 1.8 (atherosclerotic greater than or equal to 1.8, normal less than 1.8). To characterize the change in fluorescence during plaque ablation, mechanical plaque ablation with a cold microtome was performed on 16 atherosclerotic aortic specimens. Fluorescence spectra were recorded serially after each 100 microns of plaque ablation; recordings revealed a change in fluorescence spectra from atherosclerotic to a normal pattern. With an LIF ratio of 1.8 to signal termination of plaque ablation, 15 of the atherosclerotic plaques had a residual plaque thickness less than 200 microns; one specimen had a residual plaque thickness of 300 microns. No specimen demonstrated ablation of the media. There was a statistically significant correlation between LIF ratio and plaque thickness (r = .73, P less than .001), but considerable variation in LIF ratio existed at each thickness. Therefore, laser-induced fluorescence spectroscopy is capable of discriminating atherosclerotic from normal aorta and of signaling completion of plaque ablation.

Fluorescence spectroscopy guidance of laser ablation of atherosclerotic plague.

Laser-induced fluorescence (LIF) spectroscopy can only be used for laser angioplasty guidance if high-power laser ablation does not significantly alter the pattern of tissue fluorescence. Although the spectra of normal and atherosclerotic arteries differ, the change in fluorescence spectra following laser angioplasty has not been well studied. Therefore, the purpose of this study was to assess whether laser-induced fluorescence spectroscopy could guide selective laser ablation of atherosclerotic plaque and, if so, to develop a quantitative LIF score that could be used to control a "smart" laser angioplasty system. Baseline LIF spectroscopy of 50 normal and 50 atherosclerotic human aortic specimens was performed using an optical fiber coupled to a He-Cd laser and optical multichannel analyzer. LIF was then serially recorded during erbium:YAG laser ablation of 27 atherosclerotic specimens. Laser ablation was terminated when the arterial LIF spectrum visually appeared normal. Histologic analysis revealed a mean initial plaque thickness of 1,228 +/- 54 microns and mean residual plaque thickness of 198 +/- 27 microns. Ablation of the media occurred in only three specimens. A discriminant function was derived to discriminate atherosclerotic from normal tissue for computer guidance of laser angioplasty. The LIF score, derived from stepwise multivariate linear regression analysis of the LIF spectra, correctly classified 93% of aortic specimens. The spectra obtained from the atherosclerotic specimens subjected to fluorescence-guided laser revealed a change in score from "atherosclerotic" to "normal" following plaque ablation. Seven atherosclerotic specimens were subjected to laser angioplasty with on-line computer control using the LIF score. Mean initial plaque thickness was 1,014 +/- 86 microns, and mean residual plaque thickness was 78 +/- 29 microns. There was no evidence of ablation of the media. Therefore, LIF guidance of laser ablation resulted in minimal residual plaque without arterial perforation. These findings support the feasibility of an LIF-guided laser angioplasty system for selective atherosclerotic plaque ablation.

Statistical performance of outer-product associative memory models.

A figure of merit, the probability a bit is correct after update, is used to evaluate the performance of randomly coded outer-product associative memory models. Networks with bipolar binary states and nonzero diagonal connections are shown to yield the best performance with respect to this figure of merit. A surprising result is that an all-positive network, one with binary states and positive connections, is superior to a standard Hopfield style network with binary states and bipolar connections. A prescription for the optimal threshold point for the all-positive network is given.

Hopfield model associative memory with nonzero-diagonal terms in memory matrix.

The discrete-valued neural network proposed by Hopfield requires zero-diagonal terms in the memory matrix so that the net evolves toward a local minimum of an energy function. For a version of this model with bipolar nodes and positive terms along the diagonal, the net evolves so that only updates that lower the energy by a sufficient amount are accepted. For a net programmed as an outer-product associative content-addressable memory, the version with nonzero-diagonal elements performs nearly identically to one with zero-diagonal terms, and the dropping of the zero-diagonal requirement is advantageous for optical implementation.

Imaging with rotating slit apertures and rotating collimators.

The statistical quality of conventional nuclear medical imagery is limited by the small signal collected through low-efficiency conventional apertures. Coded-aperture imaging overcomes this by employing a two-step process in which the object is first efficiently detected as an "encoded" form which does not resemble the object, and then filtered (or "decoded") to form an image. We present here the imaging properties of a class of time-modulated coded apertures which, unlike most coded apertures, encode projections of the object rather than the object itself. These coded apertures can reconstruct a volume object nontomographically, tomographically (one plane focused), or three-dimensionally. We describe a new decoding algorithm that reconstructs the object from its planar projections. Results of noise calculations are given, and the noise performance of these coded-aperture systems is compared to that of conventional counterparts. A hybrid slit-pinhole system which combines the imaging advantages of a rotating slit and a pinhole is described. A new scintillation detector which accurately measures the position of an event in one dimension only is presented, and its use in our coded-aperture system is outlined. Finally, results of imaging test objects and animals are given.

Coded aperture imaging of gamma-ray sources with an off-axis rotating slit.

The rotating slit is a time-varying coded aperture that can be used advantageously with a scintillation camera in imaging of static radioactive objects. The coded image obtained from an on-axis slit conveys no tomographic information, but when the slit is displaced away from the axis of rotation, tomographic information can be obtained. From one set of coded images, each plane of a 3-D object can be brought in focus by shift operations and a computed-tomography algorithm. Inverse filtering is necessary to remove the side lobes of the point spread function. The intrinsic resolution of the camera, the slit width, the filter function, and the display all affect the lateral resolution, while the object distance, the off-axis distance of the slit, and the lateral resolution determine the longitudinal resolution of the system.