Respiratory-induced prostate motion: characterization and quantification in dynamic MRI.

BACKGROUND AND PURPOSE: To investigate prostate movement during deep breathing and contraction of abdominal musculature by means of dynamic MRI and analyze implications for image-guided radiotherapy of prostate cancer. PATIENTS AND METHODS: A total of 43 patients and 8 healthy volunteers were examined with MRI. Images during deep respiration and during contraction of abdominal musculature (via a coughing maneuver) were obtained with dynamic two-dimensional (2D) balanced SSFP; 3 frames/s were obtained over an acquisition time of 15 s. Images were acquired in sagittal orientation to evaluate motion along both the craniocaudal (cc)-axis and anteroposterior (ap)-axis. Prostate motion was quantified semi-automatically using dedicated software tools. RESULTS: Respiratory induced mean cc-axis displacement of the prostate was 2.7 ± 1.9 (SD) mm (range, 0.5-10.6 mm) and mean ap-axis displacement 1.8 ± 1.0 (SD) mm (range, 0.3-10 mm). In 69% of the subjects, breathing-related prostate movements were found to be negligible (< 3 mm). The prostate displacement for abdominal contraction was significantly higher: mean cc-axis displacement was max. 8.4 ± 6.7 (SD) mm (range, 0.6-27 mm); mean anteroposterior movement was 8.3 ± 7.7 (SD) mm (range, 0.7-26 mm). CONCLUSION: Dynamic MRI is an excellent tool for noninvasive real-time imaging of prostate movement. Further investigations regarding possible applications in image-guided radiotherapy, e.g., for individualized planning and in integrated linac/MRI systems, are warranted.

Pulmonary functional imaging: qualitative comparison of Fourier decomposition MR imaging with SPECT/CT in porcine lung.

PURPOSE: To compare unenhanced lung ventilation-weighted (VW) and perfusion-weighted (QW) imaging based on Fourier decomposition (FD) magnetic resonance (MR) imaging with the clinical reference standard single photon emission computed tomography (SPECT)/computed tomography (CT) in an animal experiment. MATERIALS AND METHODS: The study was approved by the local animal care committee. Lung ventilation and perfusion was assessed in seven anesthetized pigs by using a 1.5-T MR imager and SPECT/CT. For time-resolved FD MR imaging, sets of lung images were acquired by using an untriggered two-dimensional balanced steady-state free precession sequence (repetition time, 1.9 msec; echo time, 0.8 msec; acquisition time per image, 118 msec; acquisition rate, 3.33 images per second; flip angle, 75°; section thickness, 12 mm; matrix, 128 × 128). Breathing displacement was corrected with nonrigid image registration. Parenchymal signal intensity was analyzed pixelwise with FD to separate periodic changes of proton density induced by respiration and periodic changes of blood flow. Spectral lines representing respiratory and cardiac frequencies were integrated to calculate VW and QW images. Ventilation and perfusion SPECT was performed after inhalation of dispersed technetium 99m ((99m)Tc) and injection of (99m)Tc-labeled macroaggregated albumin. FD MR imaging and SPECT data were independently analyzed by two physicians in consensus. A regional statistical analysis of homogeneity and pathologic signal changes was performed. RESULTS: Images acquired in healthy animals by using FD MR imaging and SPECT showed a homogeneous distribution of VW and QW imaging and pulmonary ventilation and perfusion, respectively. The gravitation-dependent signal distribution of ventilation and perfusion in all animals was similarly observed at FD MR imaging and SPECT. Incidental ventilation and perfusion defects were identically visualized by using both modalities. CONCLUSION: This animal experiment demonstrated qualitative agreement in the assessment of regional lung ventilation and perfusion between contrast media-free and radiation-free FD MR imaging and conventional SPECT/CT.

Software-guided standardization of manual landmark data in medical images.

It is crucial to evaluate registration algorithms in order to make them available in clinical practice. Several evaluation strategies have been proposed in the past, and one approach is to evaluate these algorithms with intrinsic anatomical landmarks identified by a health professional. The acquisition and handling of large amounts of these landmark data is a time-consuming task for the health professional, and it is vulnerable to errors and inconsistencies. Additionally, limited access to appropriate tools makes dealing with landmark data considerably more difficult. We introduce a strategy for the acquisition of landmarks for the landmark-based evaluation of registration algorithms and we present an ontology-driven software tool that assists the different partners involved to act according to that strategy. This tool provides the user with intrinsic knowledge of the registration problems, the possibility to conveniently make the acquired data available to further processing, and an easy-to-use graphical interface.