Impeded diffusion fraction for quantitative imaging diagnostic assay
Methods and systems are provided for analyzing diffusion weighted images (DWI) using impeded diffusion fraction models for quantitative imaging diagnostic assay of cancer, such as glandular tissue cancers. The Impeded diffusion fraction models are tissue and cancer independent and generate a single score representative of multi-compartment diffusion fractions occurring within each voxel of a DWI image.
1 . A computer-implemented method of analyzing diffusion weighted images, the computer-implemented method comprising:
receiving, at one or more processors, diffusion weighted image (DWI) data corresponding to a sample of an imaged tissue and identifying within the DWI data a region of interest comprising one or more voxels, the region of interest spanning at least partially a tissue region within the sample;
applying, using the one or more processors, the DWI data corresponding to the region of interest to an impeded diffusion fraction model and, from applying the impeded diffusion fraction model, determining an impeded diffusion fraction for each of the one or more voxels in the region of interest, the impeded diffusion fraction model being a multi-compartment model of water coordination by macromolecules agnostic to tissue compartment origin and applicable across multiple tissue compartments using a fixed rate for free diffusion and order of magnitude scaling constraints for vascular diffusion and coordinated diffusion;
from the impeded diffusion fraction of the one or more voxels, identifying the presence and/or severity of a pathology in the imaged tissue; and
generating, using the one or more processors and the identified presence and/or severity of the pathology in the imaged tissue, a pathology determination image indicating a location of the presence and/or the severity of the pathology within the region of interest.
2 . The computer-implemented method of claim 1 , wherein the multiple tissue compartments comprise a cellular compartment, a free water compartment, and a vascular compartment.
3 . The computer-implemented method of claim 1 , wherein the impeded diffusion fraction model comprises an expression for nonlinear impeded diffusion fraction (IDF):
S
b
S
0
≈
F
f
E
f
+
(
1
-
F
p
-
F
f
)
E
c
;
E
f
=
exp
(
-
b
D
f
)
;
E
c
≈
exp
(
-
b
F
fc
D
f
)
;
IDF
=
1
-
F
p
-
F
f
-
F
f
c
where Sb is the DWI signal intensity as a function of b-value, S0 is signal intensity at b=0, b are three or more b-values between 0.1 and 2 ms/μm2 (one or more>1 ms/μm2) Df is free diffusion rate, Fp is a pseudo diffusion fraction corresponding to a vascular compartment, Ff is a free water fraction, and Ffc is an uncoordinated water fraction in a cellular compartment.
4 . The computer-implemented method of claim 1 , wherein the impeded diffusion fraction model comprises an expression for a linear impeded diffusion fraction (IDFL):
log
(
S
b
S
0
)
=
C
1
+
C
2
·
b
;
F
fc
=
-
C
2
/
D
f
;
F
c
=
exp
(
C
1
)
;
IDF
L
=
exp
(
C
1
)
+
C
2
/
D
f
where Sb is the DWI signal intensity as a function of b-value, S0 is signal intensity at b=0, C1 is the linear fit intercept, C2 is a linear fit slope, b are two or more b-values between 0.6 and 2 ms/μm2 (one or more>1 ms/μm2) linear fit intercept, Df is free diffusion rate, and Fc is cellular fraction, for each of the one of more voxels.
5 . The computer-implemented method of claim 1 , wherein the impeded diffusion fraction model comprises an nonlinear fit impeded diffusion fraction.
6 . The computer-implemented method of claim 5 , wherein the nonlinear fit impeded diffusion fraction comprises fit constraints.
7 . The computer-implemented method of claim 1 , wherein the impeded diffusion fraction model comprises a linear fit impeded diffusion fraction.
8 . The computer-implemented method of claim 7 , wherein the linear fit impeded diffusion fraction comprises fit constraints.
9 . The computer-implemented method of claim 1 , wherein the DWI data comprises image data of the sample and comprising a plurality of voxels, the method further comprising:
prior to applying the DWI data to the impeded diffusion fraction model, identifying from the plurality of voxels, voxels satisfying a threshold voxel signal threshold condition and applying, as the DWI data, those voxels satisfying the threshold voxel signal threshold condition.
10 . The computer-implemented method of claim 1 , wherein the DWI data comprises image data of the sample and comprising a plurality of voxels, the method further comprising:
prior to applying the DWI data to the impeded diffusion fraction model, identifying from the plurality of voxels, voxels corresponding to a boundary region between normal tissue and lesion indicating tissue, and excluding, from the DWI data, those voxels corresponding to the boundary region.
11 . The computer-implemented method of claim 1 , wherein the DWI data comprises a plurality of voxels and wherein identifying the presence of a pathology in the imaged tissue comprises:
determining the impeded diffusion fraction for each of a plurality of voxels, determining a statistical summary metric of impeded diffusion fraction from the plurality of impeded diffusion fractions, and identifying the presence of the pathology from the summary metric for impeded diffusion fraction.
12 . The computer-implemented method of claim 11 , wherein the statistical summary metric for the plurality of voxels within region of interest is any mathematical histograms characteristic (moment), or percentiles.
13 . The computer-implemented method of claim 11 , wherein identifying the presence of a pathology in the imaged tissue comprises: determining if the impeded diffusion fraction is above a threshold value of the statistical summary metric, the threshold value corresponding to the presence of the pathogen.
14 . The computer-implemented method of claim 1 , wherein identifying the presence of a pathology in the imaged tissue-comprises: determining if the impeded diffusion fraction is above a threshold value, the threshold value corresponding to the presence of the pathogen.
15 . The computer-implemented method of claim 14 , wherein the threshold value is determined from retrospective DWI analysis and adjusted for acquisition protocol bias in T1 and T2 weighting and b-range.
16 . The computer-implemented method of claim 1 , wherein the pathogen is cancer.
17 . The computer-implemented method of claim 1 , wherein the pathogen is cancer severity.
18 . The computer-implemented method of claim 1 , wherein the pathology is cancer and the sample is taken from disease tissue, the method further comprising, after performing a treatment on the subject:
receiving, at one or more processors, subsequent DWI data corresponding to a subsequent sample taken from the disease tissue, and identifying within the subsequent DWI data one or more voxels corresponding to the region of interest;
applying, using the one or more processors, the subsequent DWI data to the impeded diffusion fraction model and determining the presence of a change in the impeded diffusion fraction from the impeded diffusion fraction determined from the DWI data; and
from the presence of the change in the impeded diffusion fraction, determining an efficacy of the treatment on the subject.
19 . The computer-implemented method of claim 16 , wherein the cancer is prostate cancer, breast cancer, or pancreatic cancer.