Archana Vadiraj Malagi1, Virender Kumar2, Kedar Khare3, Chandan J. Das4, and Amit Mehndiratta1,5
1Centre for Biomedical Engineering, Indian Institute of Technology Delhi, New Delhi, India, 2Department of Nuclear Magnetic Resonance (NMR),, All India Institute of Medical Sciences Delhi, New Delhi, India, 3Department of Physics, Indian Institute of Technology Delhi, New Delhi, India, 4Department of Radiodiagnosis, All India Institute of Medical Sciences Delhi, New Delhi, India, 5Department of Biomedical Engineering, All India Institute of Medical Sciences Delhi, New Delhi, India
Synopsis
IVIM-DKI was used to evaluate clinical utility
and diagnostic performance of IVIM-DKI parameters in characterization of
prostate cancer(PCa) against benign prostatic hyperplasia(BPH) and healthy tissues.
Sixteen patients suffering from PCa and BPH were recruited and underwent
IVIM-DKI with routine MR sequences and biopsy. IVIM-DKI signal was modelled
using hybrid model with total variation (TV) penalty function which removes any
spurious values in parameters adaptively. HY with TV method produced high
diagnostic performance (AUC:D=0.94-0.98, f=0.82-0.85 and k=0.85-0.91) than HY
method for all tissues and high-quality parameter maps. IVIM-DKI
parameters (D,f,k) showed significant differences between tumor and BPH and
healthy tissue.
Introduction
Most
common prostatic disease comprises of benign prostatic hyperplasia(BPH) and
prostate cancer(PCa). BPH arises in transition zone and PCa is mostly
adenocarcinoma, which arises in peripheral zone(PZ) and less frequently in
transition zone and in about 20% of cases BPH and PCa can coexists1,2.
TRUS-guided biopsy is considered as the gold standard for confirmatory diagnosis. However, TURS biopsy has a low specificity for localization of small lesions and
its charecterisation3. Diffusion-weighted Imaging(DWI) plays a major
role in localization of PCa and BPH using apparent diffusion coefficient(ADC)
which measures water diffusivity in tissues4. Whereas IVIM-DKI altogether
captures fast and slow diffusion i.e. pseudo-diffusion and diffusion
coefficient(D & D*) as well as perfusion information and tissue heterogeneity
i.e. perfusion fraction and kurtosis(f & k)5-7. IVIM-DKI
parameters can be evaluated using a Hybrid model(HY) for analysis, which suffers
from inhomogeneity and thus is not widely used in clinical routine. This can be
corrected using a constrained reconstruction method i.e., total variation(TV)
penalty function which adaptively reduces sudden changes in the image8-10.
Objective of this study is to evaluate whether IVIM-DKI parameter using hybrid
model with TV can differentiate between heathy tissue, tumor and BPH.Methods
Patients
and MR acquisition: Sixteen patients with PCa (n=16, age: 61.94±5.47years) were recruited
and underwent TRUS biopsy. All patients were scanned in 1.5T MRI(Achieva;
Philips Healthcare, Netherlands), with a T2-weighted imaging(TR=3.863s and
TE=0.09s), including IVIM-DKI with
13b-values=0,25,50,75,100,150,200,500,800,1000,1250,1500,2000 s/mm2 using
phased-array surface coil(TR=5.774s,TE=0.081s).
Analysis: All parameters were estimated using Non-linear least square optimization
with in-house toolbox using MATLAB. ADC was calculated voxelwise using
monoexponential model, defined as below:
$$S⁄S_0=e^{-b.ADC}$$
IVIM-DKI parameters such as D, D*, f
and k were obtained using two methods i) HY model7 and ii) HY with
TV. HY model is
defined below:
$$S⁄S_0=fe^{-bD^*}+(1-f)e^{-bD+b^2 D^2 k/6}$$
where S and S0 are diffusion
signals with and without diffusion gradient b in s/mm2. In HY+TV
model, iteratively image gradient was calculated, and parameter values were
updated with TV parameters(alpha and beta) set to 0.005 and 0.99.
ROI localization: Multiple lesions were found in patients with high grade PCa, only
biopsy proven lesion was selected for tumor and BPH ROI. For tumor ROI, DWI at
b=2000 s/mm2(hyperintense) and ADC map(hypointense) was used to
localize tumor. Healthy PZ and BPH were drawn on b=0 s/mm2 image
as shown in Figure1. All ROIs were validated by radiologists with 10 years’
experience with prostate imaging.
Statistics: One-way ANOVA with Tukey's
multiple comparison test was used for computing significant
differences(p-value<0.05) between IVIM-DKI parameter values between healthy
PZ, tumor and BPH. Diagnostic performances of HY and HY+TV were evaluated using
receiver operating characteristics(ROC) analysis regression model with cut off
values, sensitivity and specificity(MedCalc, version 19.1.7).Results
Qualitative
and quantitative assessment of PCa, BPH and Healthy PZ: Figure2 shows a representative patient for qualitative comparison between Healthy PZ,
tumor and BPH having PCa with Gleason score of 6. Tumor appeared hypointense on
ADC, D, and f, whereas it appeared hyperintense on D* and k maps. For healthy
PZ and BPH, hyperintense region was observed on ADC, and D and hypointense on
b=2000 s/mm2, D*, f and k maps.
HY
results showed significant difference between for tumor against healthy
PZ(p-value<0.01) and BPH(p-value<0.01) only for D and f. HY+TV showed
significant results for D, f and k for tumor against healthy
PZ(p-value<0.01) and BPH(p-value<0.01). No significant results obtained
for D*. Mean values of IVIM-DKI parameters from HY and HY+TV were significantly
different(p<0.05) as shown in figure3.
Diagnostic
performance of HY and HY+TV model: For
tumor vs BPH, diagnostic performance of IVIM-DKI parameters improved from HY+TV
model with better AUC. D (HY+TV) showed highest AUC of 0.94 (cut off: 1.4x10-3
mm2/s) and other parameters showed poor AUC. For tumor vs healthy PZ
diagnostic performance improved by using HY+TV except for f parameter. D and k
showed highest AUC of 0.98 and 0.91 (cut off: D=1.4;k=0.7) respectively as shown
in figure4. Discussion
IVIM-DKI is not clinically reliable as HY model produces
non-physiological inhomogeneity in parameter map. This was corrected by total
variation penalty function along with HY model which removed spurious values. We
wanted to evaluate whether using IVIM-DKI parameter maps obtained from HY+TV
model can differentiate tumor from healthy PZ and BPH. Previous literature
has shown ADC showed good accuracy for differentiating tumor vs BPH compared
to IVIM parameters11-13. In this study, we were able to show
IVIM-DKI parameters of tumor region, such as D and f decreased whereas k
increased significantly against healthy tissue and BPH indicating that tumor
region consisted of poor cell differentiation with faster cell growth which
causes unobstructed water movement14. This also suggests that low f
and high k in tumor due to poor vasculature and increase in tissue
heterogeneity15. ROC analysis showed D(HY+TV) was able to better
differentiate between tumor against BPH(AUC:0.94) and healthy PZ(AUC:0.98) with high sensitivity(94%)
and specificity(88-94%). k and f improved in diagnostic performance with high
AUC, sensitivity(81-100%) and specificity(63-94%) using HY+TV.Conclusion
IVIM-DKI
parameters such as D, f and k from IVIM-DKI model with total variation method were
able to classify tumor against BPH and healthy tissue. Also, IVIM-DKI with TV
produced parameters with high diagnostic performance and clinically reliable
parameter map.Acknowledgements
This study was supported by IIT Delhi and AIIMS Delhi.
AVM was supported by research fellowship fund from the Ministry of Human Resource
Development, Government of India.References
- McVary KT. BPH: Epidemiology and comorbidities. Am J
Manag Care 2006;12 5 Suppl:S122-8.
-
McNeal JE. Normal histology of the prostate. Am J
SurgPathol 1988;12:619-33.
-
Hodge KK, McNeal JE, Terris MK, Stamey TA. Random
systematic versus directed ultrasound guided transrectal core biopsies of the
prostate. J Urol 1989;142(1):71–4.
- Morgan VA, Kyriazi S, Ashley SE et al. Evaluation of
the potential of diffusion-weighted imaging in prostate cancer detection. Acta
Radiol 2007;48(6):695–703.
-
Le Bihan D. et al. Separation of diffusion and
perfusion in intravoxel incoherent motion MR imaging.
Radiology.1988;168:497–505.
-
Jensen J H, Helpern J A, Ramani A. et al.
Diffusional Kurtosis Imaging : The Quantification of Non- Gaussian Water
Diffusion by Means of Magnetic Resonance Imaging. Magn Reson
Med.2005;1440:1432–1440.
-
Wu W C, Yang S C, Chen Y F. et al. Simultaneous
assessment of cerebral blood volume and diffusion heterogeneity using hybrid
IVIM and DK MR imaging: initial experience with brain tumors. Eur.
Radiol.2017;27:306–314.
-
Rudin L I, Osher S & Fatemi E. Nonlinear total
variation based noise removal algorithms. Phys. D nonlinear
Phenom.1992;60:259–268.
-
Kayal E B et al. Quantitative Analysis of Intravoxel
Incoherent Motion (IVIM) Diffusion MRI using Total Variation and Huber Penalty
Function. Med. Phys.2017;44:5849–5858.
-
Malagi, A. V. et al.
Effect of combination and number of b values in IVIM analysis with post-processing
methodology: simulation and clinical study. Magn. Reson. Mater. Physics, Biol.
Med. 2019;32, 519–527.
-
Zhang Y D. et al. The histogram analysis of
diffusion-weighted intravoxel incoherent motion (IVIM) imaging for
differentiating the gleason grade of prostate cancer. Eur.
Radiol.2015;25:994–1004.
-
Shinmoto H. et al. An intravoxel incoherent motion
diffusion-weighted imaging study of prostate cancer. Am. J.
Roentgenol.2012;199:496–500.
-
Tamura C. et al. Diffusion kurtosis imaging study of
prostate cancer: preliminary findings. J. Magn. Reson. Imaging.2014;40:723–729.
-
Döpfert J et al. Investigation of prostate cancer
using diffusion-weighted intravoxel incoherent motion imaging. Magn Reson
Imaging.2011;29:1053–1058.
-
Ding Y et al. Differentiating between malignant and
benign renal tumors: do IVIM and diffusion kurtosis imaging perform better than
DWI? Eur Radiol. 2019;6930-6939