The protocol describes an optimized workflow for preclinical PET-based radiotherapy in a rat glioblastoma model using in-house developed algorithms. The optimized approach favors automation and is less time consuming when compared to the previously developed workflow. After anesthetizing an F98 glioblastoma rat harboring a seven to eight millimeter diameter tumor, inject 10 to 12 megabecquerel of fluorine-18 FET dissolved in 200 microliters of saline in the lateral tail vein one hour before PET acquisition.
Let the animal regain consciousness while the tracer gets distributed throughout the body. Meanwhile, fix a capillary filled with the MRI PET agent for easier co-registration. Then anesthetize the animal again and place it on a multi-modality bed.
Secure the animal using hook and loop fasteners to maintain a fixed position during the imaging and micro-irradiation, then wrap the animal in bubble wrap to preserve its body temperature during the multi-modality imaging and therapy. One hour after the injection of the PET tracer, perform a PET scan. Reconstruct the PET scan into a 3D volume with a 0.4 millimeter voxel size using 30 iterations of the maximum likelihood expectation maximization algorithm.
Next, inject an MRI contrast agent into the tail vein, then place the rat still fixed on the multi-modality bed in the animal holder of the MRI scanner. Perform a localizer scan followed by a contrast enhanced T1 weighted spin echo sequence. Then place the animal still fixed on the multi-modality bed on a plastic holder secured onto the four-axis robotic positioning table on the micro-irradiator.
Perform a high-resolution treatment planning cone beam CT requiring a total of 360 projections over 360 degrees. Reconstruct the CT images with an isotropic voxel size of 0.275 millimeters. For image co-registration, place the three image modalities into one folder, then import the converted images into MATLAB.
Next, run the dose painting co-registration MATLAB script which converts the DICOM images to the NIfTI format, filters the PET image with a one millimeter full width half max Gaussian filter, crops the PET image and moves the image centers close to each other and performs the actual rigid body co-registration using statistical parametric mapping. Evaluate the result of the automatic co-registration before proceeding to treatment planning. To apply method one, run the dose painting radiation planning MATLAB script and load the three different imaging modalities into the MATLAB app.
Next, place a generous bounding box around the contrast enhancement on the transverse, sagittal and frontal views of the T1 weighted MRI scan. Save the location of the bounding box, then finalize the box. Determine the contrast enhanced volume using a threshold.
If multiple regions have been selected, select only the largest volume, the center of which is considered as the first isocenter to deliver a prescribed dose for radiation therapy. Expand the previously determined MRI contrast enhancement by 10 pixels in each direction. If multiple regions are detected, retain only the largest PET volume, the center of which is considered the second isocenter to deliver a prescribed dose for radiation therapy.
For the first isocenter, deliver a prescribed dose of 2, 000 centigray using three non-coplanar arcs at couch positions zero, negative 45 and negative 90 degrees with a gantry rotation of 120, 120, and 60 degrees respectively. Use a fixed collimator size of 10 by 10 millimeters. However, for smaller tumors, use an appropriate size such as five by five millimeters.
For the second isocenter, deliver a prescribed dose of 800 centigray using three non-coplanar arcs at couch positions zero, negative 45 and negative 90 degrees with a gantry rotation of 120, 120, and 60 degrees respectively. Use a fixed collimator size of one millimeter. Calculate the dose distribution within the animal and the beam delivery parameters.
To apply method two, load the three different imaging modalities into the MATLAB app as demonstrated earlier, then place a generous bounding box around the contrast enhancement on the transverse, sagittal and frontal views of the fluorine-18 FET PET image and save the locations of the bounding box. After finalizing the bounding box, use the appropriate MATLAB script to determine V50, V60, V70, V80, and V90 in the isocenters and the jaw dimensions for each beam required to guide the motorized variable collimator. To deliver a prescribed dose of 2, 000 centigray distributed over 16 beams for V50 and a dose of 800 centigray distributed over 40 beams for V60 to V90, select the output file generated by the MATLAB script and import the 56 beams into the treatment planning software.
After verifying that all 56 beams have been imported correctly, calculate the dose distribution within the animal and the beam delivery parameters. Both methods of PET-based dose painting radiation therapy were applied to three different cases. Case one has a spherical homogenous PET uptake while cases two and three have a ring-shaped uptake where the reduced PET uptake is most likely necrotic tissue.
The dose volume histograms for method two are systematically closer to the ideal dose distribution than those for method one. A substantial tumor volume receives insufficient irradiation in cases two and three when treated with method one. The D90 and D50 values are considerably lower for method one than for method two.
Ideally, Q volume histograms make a sharp drop at a Q value equal to one. Method two always results in dose distributions that are closer to the dose objective than method one. Furthermore, the overall Q factors for method two were superior to those for method one.
This methodology is a crucial step towards inverse planning which is generally used in clinical routine and further narrows the gap between preclinical radiation research and the clinic.
