Tissue imaging and analysis using ultrasound waveform tomography

1. A computer-implemented method of analyzing a volume of tissue, comprising:

receiving, from a transducer, a set of acoustic signals derived from acoustic waveforms transmitted through the volume of tissue,

wherein the transducer comprises an array of ultrasound transmitters and an array of ultrasound receivers configured to surround the volume of tissue;

generating a sound speed model that characterizes sound attenuation to represent a distribution of sound speed across a region of the volume of tissue;

extracting a set of frequency components from the set of acoustic signals;

distributing a total number of iterations to the set of frequency components;

generating a simulated wavefield in terms of a frequency component of the set of frequency components and the sound speed model;

iteratively refining a value of the sound speed model with respect to the simulated wavefield for each of the set of frequency components until a threshold condition is satisfied, thereby producing a final value of the sound speed model, wherein the refining is performed for each of the frequency components for the number of iterations distributed to the frequency component, wherein a number of iterations distributed to a lower of the set of frequencies is no smaller than a number distributed to a higher of the set of frequencies;

computing an attenuation representation from the final value of the sound speed model; and

rendering one or more images for the volume of tissue based on the attenuation representation.

2. The method of claim 1 ,

wherein generating the simulated wavefield comprises modeling propagation of acoustic waves transmitted through the volume of tissue according to a Helmholtz operation expressed as [∇ 2 +ω 2 /c(r) 2 ]u(r,ω)=s(r,ω),

wherein ∇ 2 is the Laplacian operator, and [∇ 2 +ω 2 /c(r) 2 ] is a Helmholtz operator, including ω as a frequency component, c as the sound speed model, u as an expected numerical wavefield obtained at positions r of the transducer for the frequency component ω, and s as a spatial ultrasound source of the transducer.

3. The method of claim 1 , wherein generating the sound speed model comprises generating a set of acoustomechanical parameter slices associated with a set of coronal slices through the volume of tissue without considering out-of-plane acoustic waveform scattering.

4. The method of claim 1 , wherein the sound attenuation is intrinsic attenuation of a medium.

5. The method of claim 1 , wherein the refining is performed for the lower of the set of frequencies before the higher of the set of frequencies.

6. The method of claim 1 ,

wherein in the sound speed model is expressed as c=c R +ic I ,

wherein c R is a real portion corresponding to a phase velocity and c I is an imaginary portion proportional to the sound attenuation.

7. The method of claim 6 , wherein the rendering comprises generating an image based on a final value of real portion of the sound speed model.

8. The method of claim 6 , further comprising classifying different types of lesions in the volume of tissue based on a final value of real portion of the sound speed model.

9. The method of claim 8 , further comprising refining the classification using the attenuation representation.

10. The method of claim 6 , wherein the refining comprises applying a gradient of an error cost function comprising a difference between the simulated wavefield and a given wavefield.

11. The method of claim 10 ,

wherein the applying is expressed as c 2 =c 1 −β∇S(ω,c 1 ),

wherein c 2 is an updated value of the sound speed model, c 1 is a present value of the sound speed model, β is a step size, ∇S is the gradient of the error cost function, and ω is one of the set of frequency components.

12. The method of claim 10 , wherein the refining comprises:

applying the gradient with respect to real portion for one of the set of frequency components to obtain a value for the real portion;

applying the gradient with respect to imaginary portion for the one frequency components using the value for the real portion; and

proceeding to another of the set of frequency components.

13. The method of claim 10 , wherein the refining comprises:

applying the gradient with respect to real portion for each of the set of frequency components to obtain a value for the real portion; and

applying the gradient with respect to imaginary portion for each of the set of frequency components using the value for the real portion.

14. A non-transitory computer-readable storage medium with instructions stored thereon that, when executed by a computing system, cause the computing system to perform a method of analyzing a volume of tissue, the method comprising:

receiving, from a transducer, a set of acoustic signals derived from acoustic waveforms transmitted through the volume of tissue,

wherein the transducer comprises an array of ultrasound transmitters and an array of ultrasound receivers configured to surround the volume of tissue;

generating a sound speed model that characterizes sound attenuation to represent a distribution of sound speed across a region of the volume of tissue;

extracting a set of frequency components from the set of acoustic signals;

distributing a total number of iterations to the set of frequency components;

generating a simulated wavefield in terms of a frequency component of the set of frequency components and the sound speed model;

iteratively refining a value of the sound speed model with respect to the simulated wavefield for each of the set of frequency components until a threshold condition is satisfied, thereby producing a final value of the sound speed model, wherein the refining is performed for each of the frequency components for the number of iterations distributed to the frequency component, wherein a number of iterations distributed to a lower frequency is no smaller than a number distributed to a higher frequency;

computing an attenuation representation from the final value of the sound speed model; and

rendering an image for the volume of tissue based on the attenuation representation.

15. The non-transitory computer-readable storage medium of claim 14 ,

wherein the sound speed model comprises a real portion and an imaginary portion corresponding respectively to a phase velocity and the sound attenuation, and

wherein the refining comprises:

applying a gradient of an error cost function with respect to real portion for one of the set of frequency components to obtain a value for the real portion,

wherein the error cost function comprises a difference between the simulated wavefield and a given wavefield,

applying the gradient with respect to imaginary portion for the one frequency component using the value for real portion, and

proceeding to another of the set of frequency components.

16. A system for analyzing a volume of tissue, comprising:

a transducer comprising an array of ultrasound transmitters and an array of ultrasound receivers and configured to surround the volume of tissue,

wherein the array of ultrasound transmitters emits acoustic waveforms toward the volume of tissue to be received by the array of ultrasound receivers; and

wherein the transducer converts received acoustic waveforms to a set of acoustic signals;

a processor, comprising:

a generating unit configured to generate a sound speed model that characterizes sound attenuation to represent a distribution of sound speed across a region of the volume of tissue;

an identifying unit configured to identify a set of frequencies from the set of acoustic signals;

a distributing unit configured to distribute a total number of iterations to the set of frequencies;

a creating unit configured to create a simulated wavefield in terms of a frequency of the set of frequencies and the sound speed model;

a refining unit configured to iteratively refine a value of the sound speed model with respect to the simulated wavefield for each of the set of frequencies until a stop condition is satisfied, thereby producing a final value of the sound speed model, wherein refining is performed for each of the frequencies for the number of iterations distributed to the set of frequencies, wherein a number of iterations distributed to a lower of the set of frequencies is no smaller than a number distributed to a higher of the set of frequencies;

a computing unit configured to compute an attenuation representation from the final value of the sound speed model; and

a rendering unit configured to render one or more images for the volume of tissue based on the attenuation representation; and

a display configured to display the one or more images.