Keywords:
Tissue characterisation, Cancer, Diagnostic procedure, Computer Applications-Detection, diagnosis, MR-Diffusion/Perfusion, MR, MR physics, Breast
Authors:
F. Borlinhas1, H. A. Ferreira2; 1Lisboa/PT, 2Lisbon/PT
DOI:
10.1594/ecr2013/C-2510
Conclusion
Le Bihan et al.
described for the first time the IVIM model and argued that it could distinguish molecular diffusion of water from the microcirculation of blood (perfusion) in brain tissue [2,4-6].
In this study ADC and D parameters were shown to be more robust in discriminating between lesion types and subtypes than D* and PF,
which reflect microcirculation.
- Benign lesions showed higher ADC and D mean values than malignant lesions.
This can be explained by the fact that malignant lesions have higher cellularity which restricts water movement.
In both lesions types,
ADC values are consistently higher than D values because ADC,
besides measuring water diffusion,
it is also sensitive to perfusion effects,
while D represents intrinsic tissue diffusion [5,6]. In addition,
it was observed that in general,
diferences in mean values and significances are higher when considering D rather than ADC.
This suggests that intrinsic tissue diffusion could be a better parameter for lesion characterization.
- IDC lesions showed lower mean values of D than DCIS lesions (although,
in this study,
not significantly different).
A possible explanation is that DCIS lesions are confined to mammary ducts and surrounding tissue architecture is more preserved,
translating into higher D values.
- D* values in benign lesions were shown to be higher than in malignant lesions.
This is contrary to what was expected since D* is considered to be a perfusion marker within each voxel and it is believed that malignant lesions are more perfused than benign lesions [4].
Patel et al.
study in cirrhotic and non-cirrhotic liver,
alerted for the fitting uncertainty in IVIM studies for D*,
due to fitting errors [2].
These results should,
therefore,
be compared to dynamic contrast enhanced perfusion results for better understanding of benign lesions behavior.
- While D* may have some limitations in perfusion quantification,
PF may have more success showing if there is any microperfusion effect for low b-values.
This is because it is easier and more robust to determine if signal intensity at low b-values is higher than expected,
based on diffusion alone (Fig. 7, Fig. 8 and Fig. 9) [5,
6].
In fact,
in this study,
benign lesions showed lower PF values than malignant lesions because the latter are more perfused at the capillary level.
Significantly higher PF mean values were observed for IDC in comparison to DCIS lesions.
This could translate higher perfusion of the invasive lesion,
as it is possible to observe on Fig. 3 (C) where DCIS lesion shows a low perfusion grade.
Relatively to IDC lesions grading,
it was observed that PF values in G2 and G3 stages are lower than in G1 which could suggest that fast growing tumors perhaps have an inefficient blood supply.
In the future,
the imaging protocol will include a larger number of b-values lower than 200 s/mm2,
because this b-value range is where pseudo-diffusion has more influence on signal decay.
This could probably improve the accuracy of the determination of IVIM parameters [2].
This study,
like the study done by Thakur et al., shows that diffusion-weighted imaging quantitative parameters are still more reliable than IVIM parameters,
namely in distinguishing between different breast lesion types [1].
Nevertheless,
it was also observed that PF could distinguish between IDC and DCIS lesion subtypes,
and as such IVIM parameters could have a role in more comprehensive lesion characterization.
Finally,
since the study sample is small and as many questions remain unsanswered,
more investigation on this topic is necessary.