Granted Patent B2

US 12,648,817 · App. 18/559,312 · Granted Jun 9, 2026

Inventors: Danas Sutula (Munich, DE); Max Langhof (Munich, DE)

Assignee: BRAINLAB SE

A61B34/10G06T7/0012A61B2034/104A61B2034/105G06T2207/30004

Abstract

Claims (107)

1 . Computer-implemented method of predicting a deformation of at least one part of a patient body, the method comprising:

providing a computer simulatable biomechanical model of the at least one part of a patient body

wherein the biomechanical model is a parametrized, patient-specific biomechanical model for predicting the deformation of the at least one part of the patient body,

the method further comprising:

carrying out S>1 forward simulations of the biomechanical model thereby calculating S resulting deformation vectors;

calculating by at least one processor, an approximated biomechanical model based on the S resulting deformation vectors;

acquiring pre-operative data;

optimizing, by at least one processor, a deformation predicted by the approximated biomechanical model, which results in an optimized, predicted deformation vector; and

deforming, by at least one processor, the pre-operative data by applying the optimized, predicted deformation vector to the pre-operative patient data.

2 . The computer-implemented method according to claim 1 ,

wherein the approximated biomechanical model is a parametrized, patient-specific biomechanical model and is configured to predict a deformation of the at least one part of a patient body in the form of a deformation vector as an output.

3 . The computer-implemented method according to claim 1 , the method further comprising the step: building the computer simulatable biomechanical model based on the acquired pre-operative data.

4 . The computer-implemented method according to claim 1 ,

wherein the biomechanical model is configured to receive as an input values for P different parameters p 1 , . . . , p p ; and

wherein the biomechanical model is configured to output a deformation vector based on the received input values p 1 , . . . , p p .

5 . The computer-implemented method according to claim 4 ,

wherein the optimization optimizes the values of said P parameters, and

wherein the number P parameters is less than number S forward simulations.

6 . The computer-implemented method according to claim 4 ,

wherein the P different parameters comprise at least one of a level of cerebrospinal fluid, an amount of a tumour-induced tissue deformation and/or an amount of deformation after tumour resection, an amount of fluid drainage, and an amount of pressure applied by a surgeon on a brain surface with an ultrasound probe during acquisition of ultrasound images.

7 . The computer-implemented method according to claim 4 ,

wherein the P different parameters p 1 , . . . , p p define a parameter space R P ; and

wherein the step S 3 of calculating the approximated biomechanical model based on the S resulting deformation vectors is an approximation carried out in the parameter space R P defined by the P different parameters p 1 , . . . , p p .

8 . The computer-implemented method according to claim 4 ,

wherein the S calculated deformation vectors are N-dimensional vectors of a deformation space R N with N deformation degrees of freedom,

wherein the step of calculating the approximated biomechanical model comprises the step of

determining a multivariate interpolation function for the S resulting deformation vectors, and

wherein the multivariate interpolation function is a function from parameter space R P to deformation space R N .

9 . The computer-implemented method according to claim 8 ,

wherein the step of determining the multivariate interpolation function comprises the step of

carrying out a radial basis function interpolation.

10 . The computer-implemented method according to claim 9 ,

wherein carrying out the radial basis function interpolation comprises the step of

determining an S×N weight matrix W, and

wherein the determined weight matrix W comprises information derived from the S deformation vectors resulting from the S forward simulations of the biomechanical model.

11 . The computer-implemented method according to claim 9 ,

wherein the step of carrying out the radial basis function interpolation comprises

choosing a radial basis function and optionally choosing a shape parameter,

choosing S sets of P parameter values as center points of the radial basis function in parameter space,

carrying out the S>1 forward simulations, and

calculating the radial basis function weights thereby determining the weight matrix W.

12 . The computer-implemented method according to any of the claim 1 ,

wherein the step of calculating an approximated biomechanical model comprises the step

carrying out a deformation mode decomposition of the calculated S deformation vectors, which were calculated in step, thereby determining M principal deformation modes of the S deformation vectors and/or M principal frequency modes of the S deformation vectors.

13 . The computer-implemented method according to claim 12 , the method further comprising:

approximating the biomechanical model by a linear combination of the M determined principal deformation modes using M respective mode weights for each principal deformation mode, and/or approximating the biomechanical model by a linear combination of the M principal frequency modes using M respective mode weights for each principal frequency mode.

14 . The computer-implemented method according to claim 13 ,

wherein during the step of optimizing the deformation predicted by the approximated biomechanical model said M mode weights are optimized.

15 . The computer-implemented method according to claim 14 ,

wherein the biomechanical model is configured to receive as an input values for P different parameters p 1 , . . . , p p ; and

wherein the biomechanical model is configured to output a deformation vector based on the received input values p 1 , . . . , p p ; and

wherein M>P and M<S hold true.

16 . The computer-implemented method according to claim 15 ,

wherein the P different parameters comprise at least one of a level of cerebrospinal fluid, an amount of mass effect, i.e., a tumour-induced tissue deformation and/or an amount of deformation after tumour resection, an amount of fluid drainage, and an amount of pressure applied by a surgeon on a brain surface with an ultrasound probe during acquisition of ultrasound images.

17 . The computer-implemented method according to claim 13 , the method further comprising:

determining a probability value representing a physical plausibility of the approximated biomechanical model based on a comparison of a projection from a point in mode space to parameter space.

18 . The computer-implemented method according to claim 13 , the method further comprising:

determining a respective range for each mode based on projecting each of the S initially computed deformation vectors into mode space thereby obtaining S sets of mode weights, and for each mode weight, setting an allowed range based on a minimum value and a maximum value of the S weight values of that mode thus obtained, and

constraining each mode weight to a range between the minimum and the maximum value obtained for said mode.

19 . The computer-implemented method according to claim 13 wherein during the step of carrying out the deformation mode decomposition of the S deformation vectors a Principal Component Analysis of the S deformation vectors is applied thereby identifying said M principal deformation modes of the S deformation vectors; and/or wherein during the step of carrying out the deformation mode decomposition of the S deformation vectors a Fourier decomposition of the S deformation vectors is applied thereby identifying said M principal deformation modes of the S deformation vectors.

20 . The computer-implemented method according to claim 12 ,

wherein the S calculated deformation vectors are N-dimensional vectors of a deformation space R N with N deformation degrees of freedom,

wherein in the step of carrying out the deformation mode decomposition of the S deformation vectors N deformation modes are calculated, the method further comprising the step

selecting a subset of M deformation modes considered as the principal deformation modes, and

wherein M<N holds true and wherein M<S holds true.

21 . The computer-implemented method according to claim 20 ,

wherein the selection of the subset of M deformation modes is based on a measure of mode relevance.

22 . The computer-implemented method according to claim 21 ,

wherein the measure of mode relevance is a relative magnitude of a corresponding singular value; or in case of a Fourier decomposition of the S deformation vectors modes above a pre-defined threshold frequency are ignored; and/or in case of a Fourier decomposition of the S deformation vectors modes below a pre-defined threshold amplitude are ignored.

23 . The computer-implemented method according to claim 1 ,

wherein the step of optimizing the deformation predicted by the approximated biomechanical model further comprises:

using the approximated biomechanical model for simulating a deformation vector based on P input parameter values of the P different parameters p 1 . . . , p p ;

applying the simulated deformation vector to the acquired pre-operative data resulting in deformed pre-operative data;

calculating a measure of agreement between the deformed pre-operative data and intra-operative patient data, and

optimizing a cost function, which optimizes said measure of agreement between the deformed pre-operative data and the intra-operative patient data, by repeating the optimizing steps, thereby determining the optimized, predicted deformation vector.

24 . The computer-implemented method according to claim 23 ,

wherein the cost function uses at least one of an Euclidian distance for landmark pairs; average surface distance for point clouds/point surfaces; an image similarity measure, between the intra-operative patient data and the deformed pre-operative patient data; or any combination thereof.

25 . The computer-implemented method according to claim 24 ,

wherein the cost function uses an Euclidian distance for landmark pairs,

wherein each landmark pair comprises one landmark in the pre-operative data and one landmark in the intra-operative data, and

wherein the cost function measures a distance between these two landmarks after the deformation vector has been applied, i.e., after e.g., the landmark in the pre-operative data, has been deformed.

26 . The computer-implemented method according to claim 23 ,

wherein the cost function is mathematically differentiable with respect to the deformation vector, and the method further comprising the steps

mathematically differentiating the cost function with respect to the deformation vector thereby determining gradient information, and

using said determined gradient information in the optimizing steps.

27 . The computer-implemented method according to claim 1 , further comprising:

augmenting the approximated biomechanical model with additional translational degrees of freedom involving adding three additional parameters or weights, which each control global translation of the deformation vector in one particular axis direction.

28 . The computer-implemented method according to claim 1 , the method further comprising:

displaying the deformed pre-operative data to a user.

29 . A non-transitory computer readable medium comprising instructions which when executed by at least once processor, cause at least one processor to:

provide a biomechanical model of at least one part of a patient body, wherein the biomechanical model is a parametrized, patient-specific biomechanical model for predicting a deformation of the at least one part of the patient body,

the method further comprising:

carry out S>1 forward simulations of the biomechanical model thereby calculating S resulting deformation vectors;

calculate by at least one processor, an approximated biomechanical model based on the S resulting deformation vectors;

acquire pre-operative data;

optimize, by at least one processor, a deformation predicted by the approximated biomechanical model, which results in an optimized, predicted deformation vector; and

deform, by at least one processor, the pre-operative data by applying the optimized, predicted deformation vector to the pre-operative patient data.

30 . A medical system comprising:

at least one computer having at least one processor operable to:

provide a biomechanical model of at least one part of a patient body, wherein the biomechanical model is a parametrized, patient-specific biomechanical model for predicting a deformation of the at least one part of the patient body,

the method further comprising:

carry out S>1 forward simulations of the biomechanical model thereby calculating S resulting deformation vectors;

calculate by at least one processor, an approximated biomechanical model based on the S resulting deformation vectors;

acquire pre-operative data;

optimize, by at least one processor, a deformation predicted by the approximated biomechanical model, which results in an optimized, predicted deformation vector; and

deform, by at least one processor, the pre-operative data by applying the optimized, predicted deformation vector to the pre-operative patient data at least one electronic data storage device storing the computer simulatable biomechanical model; and

at least one communication interface configured for acquiring intra-operative patient data.

Assignments (2)

Priority Claims (1)

Continuity (1)

References Cited (14)