ESTRO 36 Abstract Book

S918

ESTRO 36 2017

_______________________________________________________________________________________________

Results

Once the dUVH was created in terms of uptake values

(Fig.1.bottom left) the histogram was normalized to the

uptake value of the very first peak assuming to correspond

to the normal rectal tissue glucose uptake. Subsequently,

the maximum S/B ratios were sampled for all patients

(Fig.1.bottom right). Table 1 represents results of two-

sample paired t-test indicating that patients with

complete response to radiation (pT0) have higher and

significantly different S/Bmax when compared to non-

responders. While there are some contradictory data in

the literature in regards to SUVmax and clinical outcomes,

our results are in agreement with notion that more

aggressive tumors (that spend longer periods of time in M

phase, known to be more radiosensitive) have better

response to

radiation.

Conclusion

While the dUVH method was initially developed to extract

different biological sub-volumes (glucose phenotypes)

within tumors, results presented here suggest that the

same method can help in defining the background uptake

on PET images. The method described here shows an

alternative to sampling the background (normal) uptake

within contralateral (healthy) tissue in the case of paired

organs (lung, brain, etc). Furthermore, reconstructed

S/Bmax values may prove to represent a prognostic factor

of tumor response to radiation and hence allow for

tailoring of more patient specific treatment strategies.

EP-1702 Evaluation of radiation induced MRI intensity

change in vertebral bodies after proton beam scanning

L. Placidi

, R. Poel

, A.J. Lomax

, D.C. Weber

, M. Peroni

Paul Scherrer Institute PSI, Centre for Proton Therapy,

Villigen PSI, Switzerland

Purpose or Objective

This retrospective study aims at evaluating the magnetic

resonance intensity bone change (MRiBC) [Gensheimer et

al. 2009] induced by pencil beam scanning (PBS) proton

treatment. Two fundamental aspects are tackled: (i) the

MRI signal reproducibility, especially if the imaging is

performed at different time points, with the goal of

generating

differential

maps

which

would

enhance/visualize the signal changes; (ii) correlation of

MRiBC with the dose distribution, for the purpose of

identifying a cumulative dose threshold.

Material and Methods

Data of three patients treated at PSI where target was

involving or immediately next the spinal cord resulting

in an evident bone change (fatty replacement) after the

treatment was selected. All patients received a pre- and

a post-treatment MRI and a subset of the acquired

sequences (T2 space transversal,T1 vibe dixon transversal

with contrast media) were included into this

analysis. Rigid registration on the bony structure has been

performed in the region of interest of the bone change

between the planning CT and the MRIs (T1 and T2

sequences), both pre and post treatment. To generate a

differential map enhancing only the fatty replacement, it

is important that same tissues have the same signal across

the different time points. Therefore, we evaluated the

stability of MRI signal pre- and post- treatment on

vertebral bodies outside the treatment area. Finally, the

post treatment MR Intensity was correlated with dose

distribution.

Results

Variations in signal intensity of the same ROI and in

different acquisition days show a mean value of the mean

signal intensity difference of 50 and a max-min difference

of 130 (grey intensity scale). Therefore, generating a

reliable subtraction map is extremely difficult with

conventional sequences. For the considered 3 cases, the

correlation between the MRiBC and dose distribution looks

good for all 3 patients (Fig. 1). In this case, the bone

change lies between the 20%-30% isodose line.

Conclusion

MRI signals are not quantitative enough to use and it is

difficult to manipulate MRI settings to make it more

quantitative. Image processing, as normalisation,

smoothing, etc., can lead to a substantial variation of the

original data set and shows relevant signal histogram

differences. A possible common solution could be a

quantitative MRI sequences (as a T2 relaxation time map),

that is already under investigation and could reduce signal

variation of an order of magnitude. As the dataset at our

disposal is consisting of pre- and post-treatment MRIs only,

it is difficult to say whether bone marrow change is more

like a threshold or if there is in fact a useable gradient. In

order to monitor the onset of the fatty replacement, track

it during the entire process and better correlate it to the

delivered dose, periodic MRI acquisition during treatment

is necessary. Nevertheless, for these patients, it is clear

that there is a correlation between MRiBC and dose.

Electronic Poster: Physics track: Images and analyses