Biomedical subjects

M I Miller

Publications and source records attributed to M I Miller.

5 recordsLinked to original sources

PubMed1992

A personal computer based implementation of the maximum-likelihood method of analysis of electron microscope autoradiographs.

The maximum-likelihood (ML) method for the quantitative analysis of electron-microscopic autoradiographs has been shown to be substantially superior to the conventional crossfire (CF) method. It can generate reliable and accurate tracer concentration estimates with far fewer micrographs and produce valid estimates even at counts low enough to preclude the use of the crossfire method while eliminating the need for special ad hoc treatment of narrow membranous structures as well as the secondary verification of the tracer concentration estimates. Despite these significant advantages, the large computational requirements of the ML method has to date hampered its widespread use. In this paper, we present a new line-integration method that allows us to reduce the computational requirements of the ML method to a point where it becomes feasible to implement it on a small computer system of the type typically available to a laboratory user of EM autoradiography. We present the complete line-integration method for the particular case of EM autoradiography with tritium, and show how it can be adapted to other isotopes. We have constructed a software package that implements the complete maximum-likelihood method on the IBM PC class of machines using our line-integration method. Features of this software package which are of particular importance to the research community are device independence, which makes it usable with a large variety of currently available laboratory equipment, and easy portability of the software and data between different computer systems.

Algorithms↗

PubMed1992

Bayesian model selection and minimum description length estimation of auditory-nerve discharge rates.

Auditory-nerve fiber discharges are modeled as self-exciting point processes with intensity given by the product of a stimulus-related function and a refractory-related function. Previous methods of estimating these two functions, based on the maximum-likelihood principle, have the problem of estimating more parameters than the data can support. A new procedure, based on a Bayes criterion for choosing the complexity of the model in addition to estimating the parameters, solves the over-parametrization problem. This procedure is seen to relate asymptotically to Rissanen's minimum description length (MDL) criterion. A performance comparison of the MDL procedure with previous maximum-likelihood algorithms promotes the adoption of the MDL procedure for simultaneous estimation of the stimulus and recovery properties of auditory-nerve discharge.

Algorithms↗

PubMed1992

A statistical study of cochlear nerve discharge patterns in response to complex speech stimuli.

Cochlear nerve discharge patterns in response to the synthesized consonant-vowel stimulus /da/ were collected from a population of 223 auditory-nerve fibers from a single cat. For each nerve fiber, discharges were measured from multiple, independent stimulus presentations, with the means and variances of the post-stimulus time histograms and Fourier transforms of response generated from the ensemble of stimulus presentations. The statistics were not consistent with those predicted via an inhomogeneous Poisson counting process model. Specifically, the synchronized components as measured by the Fourier transforms of post-stimulus time histogram responses have variances that are as much as a factor of 3 times lower than the predicted by the Poisson model. To account for the non-Poisson nature of the statistics, the Markov process model of Siebert/Gaumond was adopted. Using the maximum-likelihood and minimum description length algorithms, introduced by Miller [J. Acoust. Soc. Am. 77, 1452-1464 (1985)] and Mark and Miller [J. Acoust. Soc. Am. 91, 989-1002 (1992)], estimates of the stimulus and recovery functions were computed for each nerve fiber. Then, Markov point processes were simulated with the stimulus and recovery functions generated from these nerve fibers. The statistics of the simulated Markov processes are shown to have almost identical first- and second-order statistics as those measured for the population of auditory-nerve fibers, and demonstrates the effectiveness of the Markov point process model in accounting for the correlation effects associated with the discharge history-dependent refractory properties of auditory nerve response.

Acoustic Stimulation↗

PubMed1992

Regularized linear method for reconstruction of three-dimensional microscopic objects from optical sections.

The inverse problem involving the determination of a three-dimensional biological structure from images obtained by means of optical-sectioning microscopy is ill posed. Although the linear least-squares solution can be obtained rapidly by inverse filtering, we show here that it is unstable because of the inversion of small eigenvalues of the microscope's point-spread-function operator. We have regularized the problem by application of the linear-precision-gauge formalism of Joyce and Root [J. Opt. Soc. Am. A 1, 149 (1984)]. In our method the solution is regularized by being constrained to lie in a subspace spanned by the eigenvectors corresponding to a selected number of large eigenvalues. The trade-off between the variance and the regularization error determines the number of eigenvalues inverted in the estimation. The resulting linear method is a one-step algorithm that yields, in a few seconds, solutions that are optimal in the mean-square sense when the correct number of eigenvalues are inverted. Results from sensitivity studies show that the proposed method is robust to noise and to underestimation of the width of the point-spread function. The method proposed here is particularly useful for applications in which processing speed is critical, such as studies of living specimens and time-lapse analyses. For these applications existing iterative methods are impractical without expensive and/or specially designed hardware.

Computer Simulation↗

PubMed1991

Bayesian image reconstruction for emission tomography incorporating Good's roughness prior on massively parallel processors.

Since the introduction by Shepp and Vardi [Shepp, L. A. & Vardi, Y. (1982) IEEE Trans. Med. Imaging 1, 113-121] of the expectation-maximization algorithm for the generation of maximum-likelihood images in emission tomography, a number of investigators have applied the maximum-likelihood method to imaging problems. Though this approach is promising, it is now well known that the unconstrained maximum-likelihood approach has two major drawbacks: (i) the algorithm is computationally demanding, resulting in reconstruction times that are not acceptable for routine clinical application, and (ii) the unconstrained maximum-likelihood estimator has a fundamental noise artifact that worsens as the iterative algorithm climbs the likelihood hill. In this paper the computation issue is addressed by proposing an implementation on the class of massively parallel single-instruction, multiple-data architectures. By restructuring the superposition integrals required for the expectation-maximization algorithm as the solutions of partial differential equations, the local data passage required for efficient computation on this class of machines is satisfied. For dealing with the "noise artifact" a Markov random field prior determined by Good's rotationally invariant roughness penalty is incorporated. These methods are demonstrated on the single-instruction multiple-data class of parallel processors, with the computation times compared with those on conventional and hypercube architectures.

Algorithms↗