NMR studies have indicated that choline (Cho) is elevated in rapidly growing tissues such as tumors. 1H MRS can be used to image the location of tumors within the prostate gland. One differentiates cancer from benign tissue by using the ratio (Cho+Cr)/Cit [Cr=creatine; Cit=citrate] of respective peaks in the MR spectrum. The ratios are calculated on a spatial grid covering the prostate tissue. A voxel transformation (morphing) algorithm is designed for mapping biological points of interest from MRS images to US/CT images for target dose-escalation treatment design in implants and in IMRT. Robustness of the algorithm is tested, and plan quality and biological significance are evaluated.
Materials/Methods
Biological information was acquired via a GE Signa 1.5 Tesla MR Scanner. Radiofrequency excitation was achieved by using a whole body birdcage resonator. The NMR signal was received using a 4 element phased array antenna combined with an expandable MRInnervu endorectal RF probe. Images of the gland were taken. Mapping of metabolites over a 50-mm field of view was performed using chemical shift imaging to encode (6.25}6.25}6.25) mm3 voxels. Images and spectral data were then processed. Peak areas of Cho, Cr and Cit were calculated by numerical integration over the spectral ranges corresponding to each metabolite.
The morphing algorithm takes as input the MRS prostate voxels with outlined pockets of high tumor cell proliferation, and US/CT images of the prostate voxels for treatment. The morphing involves a non-rigid body point matching algorithm and simulated annealing routines to learn and develop a non-rigid point transformation which maps the MRS voxels onto the US/CT. The resulting transformation is used to map the MRS-identified tumor pockets onto the US/CT images. Cross validation is performed to gauge robustness. The resulting US/CT images are used for target dose-escalation. Using real patient data, optimal plans are obtained via integer programming. Associated plan quality and tumor control probability (TCP) are compared to guage the benefit of the biological information.
Results
The morphing procedure ran within 5 CPU mins. Cross validation indicates less than 0.1% error, illustrating reliability. Dose escalation was performed with PTV dose prescribed at 144Gy using 125I seeds, and 75.6Gy for IMRT. An escalated dose of >= 120% was placed on MRS-identified tumor pockets. Several sites of dose escalation are tested. Even when a tumor pocket is near the urethra, low urethral dose can be maintained while achieving good tumor dose escalation. Based on a biological model, TCP improves from 65% to 95%.
Conclusions
With advances in biological and functional imaging, there is an urgent need to incorporate radiobiological parameters within the planning process. The voxel transformation algorithm allows for biologically-enhanced treatment, which facilitates targeted delivery of escalated dose and the potential to improve overall clinical outcome. Expected improvements in tumor control, and organs-at-risk dose reduction can be significant. Clinical studies are needed to validate its importance, and to measure potential gain in clinical outcome. The morphing algorithm has been adapted for 4D IMRT planning in which organ motion and tumor shrinkage information is incorporated.
Author Disclosure: E.K. Lee, None; M. Zaider, None.
