Identification of regions of interest and extraction of time value curves in imaging procedures

• US 8,315,449 B2
• Filed: 06/24/2008
• Issued: 11/20/2012
• Est. Priority Date: 06/24/2008
• Status: Expired due to Fees

• Current Assignee: Bayer Healthcare LLC (Bayer AG)
• Original Assignee: Medrad Incorporated (Bayer AG)
• Inventors: Kemper, Corey, Kalafut, John F.
• Primary Examiner(s): Laballe, Clayton E
• Assistant Examiner(s): Rhodes, Jr., Leon W
• Application Number:
  US12/144,811
• Publication Number:
  US 20090316970A1
• Time in Patent Office: 1,610 Days
• Field of Search: 382/128, 382/130
• US Class Current: 382/130
• CPC Class Codes: G06F 18/23213 with fixed number of clusters, e.g. K-means clustering
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/0016 involving temporal comparison
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06V 10/763 Non-hierarchical techniques, e.g. based on statistics of modelling distributions
• CPC Class Codes: G06V 2201/031 of internal organs

Abstract

A system and method of extracting at least one time-value curve enables determination of a protocol for a patient in an imaging procedure. The method includes the step of determining a series of 0 through T M-dimensional data sets of pixel values of an imaged portion of the patient acquired using an imaging system. M and T are integers, and the 0 and T data sets correspond to sets at times t=0 and t=T, respectively. Other steps include: computing a predetermined number of correlated segments of the imaged portion corresponding to a number of regions of interest by computing a similarity metric of a time series of pixel values; computing the at least one time-value curve for at least one of those regions; and determining a protocol for a diagnostic scan using the imaging system based at least in part upon data from the at least one time value curve.

Claims

1. A method of extracting at least one time-value curve to determine a protocol for a patient in an imaging procedure, the method comprising the steps of:

(a) determining a series of 0 through T M-dimensional data sets of pixel values of an imaged portion of the patient acquired using an imaging system, wherein M is an integer, T is an integer, the 0 data set corresponds to a data set at a time t=0, the T data set corresponds to a data set at a time t=T, and wherein a pixel at a location represented by a variable x has an enhancement profile vector defined as;

y=[y_x(0)y_x(1) . . . y_x(T−

1)y_x(T)],

and an enhancement level y of the pixel at location x is defined over time as;

y_x(t)=y_x(0)+s_x(t)+η

_x(t),

wherein y_x(0) y_x(0) is a baseline enhancement level for the pixel, s is a change in signal due to a flow of fluid in the patient, and η

is a noise term;

(b) computing a predetermined number of correlated segments of the imaged portion corresponding to a predetermined number of regions of interest of the patient by computing a similarity metric of a time series of pixel values;

(c) computing the at least one time-value curve for at least one of the regions of interest; and

(d) determining a protocol for a diagnostic scan using the imaging system based at least in part upon data from the at least one time value curve.

2. The method of claim 1 wherein M=2, thereby rendering the M-dimensional data sets as two-dimensional data sets.

3. The method of claim 1 wherein M=3, thereby rendering the M-dimensional data sets as three-dimensional data sets.

4. The method of claim 1 wherein the predetermined number of correlated segments are computed using K-means clustering.

5. The method of claim 4 wherein K clusters are computed by minimizing a sum of squares of distances between data points and corresponding K cluster centroids, wherein K is an integer corresponding to the predetermined number of correlated segments.

6. The method of claim 5 wherein the pixel values are randomly assigned to the K clusters and the K cluster centroids are a component-wise mean of the pixel values in a corresponding K cluster after centering and normalizing the pixel values to a zero mean and unit standard deviation as follows:

7. The method of claim 6 wherein each of the K cluster centroids is determined using the following formula:

8. The method of claim 7 wherein a distance from the centered and normalized pixel values to each of the K cluster centroids y_c is determined using the formula:

9. The method of claim 8 wherein each of the centered and normalized pixel values is assigned to one of the K clusters to which it exhibits the minimum computed distance to a K cluster centroid.

10. The method of claim 9 wherein the steps of claims 7 through 9 are repeated until convergence, resulting in a segmented image having K segments corresponding to the K clusters.

11. The method of claim 10 further comprising filtering the segmented image to eliminate pixels that are not well correlated with neighboring pixels.

12. The method of claim 11 wherein each pixel of the segmented image is compared with all eight of its neighboring pixels in filtering.

13. The method of claim 11 further comprising morphologically opening the segmented image.

14. The method of claim 13 further comprising overlaying the segmented image upon a contrast enhanced bolus data set and computing enhancement profiles for each of the K segments.

15. The method of claim 14 further comprising semantically labeling at least one of the enhancement profiles with a semantic label corresponding to a region of interest of the patient based upon at least one characteristic of the labeled enhancement profile as compared to at least one other computed enhancement profile.

16. The method of claim 10 further comprising overlaying the segmented image upon a contrast enhanced bolus data set and computing enhancement profiles for each of the K segments.

17. The method of claim 16 further comprising semantically labeling at least one of the enhancement profiles with a semantic label corresponding to a region of interest of the patient based upon at least one characteristic of the labeled enhancement profile as compared to at least one other computed enhancement profile.

18. A system for extracting at least one time-value curve to determine a protocol for a patient in an imaging procedure, the system comprising:

(a) an input system for input of data output from at least one imaging system, the data comprising a series of 0 through T M-dimensional data sets of pixel values of an imaged portion of the patient acquired using the at least one imaging system, wherein M is an integer, T is an integer, the 0 data set corresponds to a data set at a time t=0, the T data set corresponds to a data set at a time t=T, and wherein a pixel at a location represented by a variable x has an enhancement profile vector defined as;

y_x=[y_x(0)y_x(1) . . . y_x(T−

1)y_x(T)],

and an enhancement level y of the pixel at the location x is defined over time as;

y_x(t)=y_x(0)+s_x(t)+η

_x(t),

wherein y_x(0) is a baseline enhancement level for the pixel, s is a change in signal due to a flow of fluid in the patient, and η

is a noise term;

(b) at least one processor in communicative connection with the input system and adapted to compute (i) a predetermined number of correlated segments of the imaged portion corresponding to a predetermined number of regions of interest of the patient by computing a similarity metric of a time series of pixel values and (ii) the at least one time value curve for at least one of the regions of interest; and

(c) at least one parameter generator system to determine a protocol for a diagnostic scan using the at least one imaging system based at least in part upon data from the at least one time value curve.

19. The system of claim 18 further comprising a memory system in communicative connection with the at least one processor, the memory system having stored therein an algorithm executable by the at least one processor to compute the predetermined number of correlated segments of the imaged portion corresponding to the predetermined number of regions of interest of the patient.

20. The system of claim 18 wherein the system is in communicative connection with the at least one imaging system.

21. The system of claim 18 wherein the system is in communicative connection with at least one injector system.

22. The system of claim 18 wherein at least a portion of the system is integrated with the at least one imaging system.

23. The system of claim 18 wherein at least a portion of the system is integrated with at least one injector system.