Deriving image data from data received from an ultrasound probe

A method of producing image data representing a specimen being imaged using ultrasound, comprising the steps of:

• receiving input data representing an output from an ultrasound probe;
• obtaining a first matrix that is dependent upon attributes of said probe and upon the way in which said input data was produced;
• obtaining a second matrix that is the inverse of the sum of a positive scalar matrix having the same dimensions as said first matrix, and the product of a Hermitian transpose of the first matrix, and the first matrix;
• initializing an estimate of the mean of the distribution of said image data using said first matrix and said input data;
• initializing estimates of echogenicity of said image data and variance of noise of said image data;
• iteratively performing the following steps until a predetermined level of convergence is reached:
• (a) update said estimate of the mean of the distribution of said image data using:
• said first matrix, a respective said estimate of said echogenicity, a respective said estimate of said variance of said noise, and said input data, and
• (b) update said estimate of the echogenicity and said estimate of the variance of the noise using said updated estimate of the mean of the distribution of said image data,
• wherein at least one iteration of step (a) is performed using an algorithm that is preconditioned with said second matrix;
• and
• using a converged estimate of the mean of the distribution of said image data to produce output image data for display.

A non-transitory computer-readable medium encoded with:

• program instructions executable by a computer for processing input data representing an output from an ultrasound probe;
• a first matrix that is dependent upon attributes of said probe and upon the way in which said input data was produced;
• and
• a second matrix, that is the inverse of the sum of a positive scalar matrix having the same dimensions as said first matrix, and the product of a Hermitian transpose of the first matrix, and the first matrix;
• and
• wherein said program instructions, when executed by a the computer, cause the computer to perform steps of:
• receiving said input data from said ultrasound probe;
• obtaining said first matrix and said second matrix;
• initializing an estimate of the mean of the distribution of said image data using said first matrix and said input data;
• initializing estimates of the echogenicity of said image data and variance of noise of said image data;
• iteratively performing the following steps until a predetermined level of convergence is reached:
• (a) update said estimate of the mean of the distribution of said image data using:
• said first matrix, a respective said estimate of said echogenicity, a respective said estimate of said variance of said noise, and said input data, and
• (b) update said estimate of the echogenicity and said estimate of the variance of the noise using said updated estimate of the mean of the distribution of said image data,
• wherein each iteration of step (a) is performed using an algorithm that is preconditioned with said second matrix;
• and
• using said converged estimate of the mean of the distribution of said image data to produce output image data for a display.

A method of performing deconvolution on data received from an ultrasound probe, comprising the steps of:

• obtaining a blurring matrix (H) that encapsulates the ultrasound blur function associated with said ultrasound probe;
• obtaining a preconditioning matrix (P−1) that is the sum of:
• the product of a Hermitian transpose of said blurring matrix, and the blurring matrix;
• and a positive scalar matrix that is the product of a positive constant (η) and an identity matrix of the same dimension as the blurring matrix;
• i.e., P−1=(HHH+ηIH)−1;
• and
• evaluating an estimate of the mean of the distribution of the data received from the ultrasound probe by using an Expectation Maximization algorithm, in which at least one Expectation step thereof is performed using a Conjugate Gradients algorithm that is preconditioned using said inverse of the preconditioning matrix.