banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Flattening filter removal for improved image quality of megavoltage fluoroscopy
Rent this article for


Image of FIG. 1.
FIG. 1.

An example of the problem encountered when attempting clinical MV fluorscopy of implanted fiducial markers. Fiducial markers are highly visible in kV images but not in MV images due to decreased high-Z contrast at megavoltage energies. Left : A kV 4DCT image of a gold fiducial marker implanted in the liver. Center: A DRR created from one respiratory phase of the 4DCT. Right: One frame of a 6 MV treatment beam fluoroscopy image sequence of the same patient. The marker cannot be seen in the MV image sequence even when played in cine mode.

Image of FIG. 2.
FIG. 2.

The flattening filter causes attenuation, beam hardening, and scatter. The FFF beam is, therefore, more intense, has higher primary-to-scatter ratio, a smaller effective focal spot size, and its spectrum is softer with lower mean energy and better matched to the spectral response of the flat-panel detector array.

Image of FIG. 3.
FIG. 3.

The megavoltage energy spectra for a flattening filter attenuated beam (flat) and a flattening-filter-free (FFF) beam obtained by Monte Carlo simulation by Fadeggon (Ref. ). The FFF beam has much higher photon flux (approximately 100× greater) at low-energies (<100 kV) compared to the flat beam. This is also the energy range where the solid-state flat panel detector has maximum response per photon (bar on horizontal axis).

Image of FIG. 4.
FIG. 4.

(a) A 6 MV portal image of the QC3 image quality phantom is shown with ROIs over the high contrast regions used to compute the contrast to noise ratio. (b) The contrast-to-noise ratio for flat (circles) and FFF (square) beams plotted as a function of the dose rate (proportional to the MVF dose per image frame). Error bars are standard deviations. Note the FFF beam has 30% higher CNR than that expected due to photon fluence alone (curve plotted with dotted line). This points to other contributing factors: beam energy and scatter.

Image of FIG. 5.
FIG. 5.

Megavoltage portal images acquired using the flat beam (left) and FFF beam (right) are shown. The central structure is a gold domed disk, the outer four structures are inserts from an electron-density CT phantom equivalent to lung tissue (upper right) and other soft tissues. For the FFF versus flat beams, the attenuation is only slightly greater for soft tissue (+6%–9%), but is nearly doubled (+94%) for the high-Z disk. This demonstrates the impact on MV image contrast of the beam hardening caused by the flattening filter.

Image of FIG. 6.
FIG. 6.

The relative modulation transfer function (RMTF) characterizes the frequency response of an imaging system, which is the ability to transfer spatial frequencies from the target object to the output image. The RMTF of the flat and FFF beams were assessed using the QC3 phantom (Standard Imaging). The RMTF was increased at all spatial frequencies for the FFF beam, but especially at higher spatial frequencies. The inset figure shows the ratio RMTF(FFF)/RMTF(flat) as a function of spatial frequency.

Image of FIG. 7.
FIG. 7.

Top: Single frames from the fluoroscopy image sequences acquired using the flat (left) and flattening filter free (right) beams are shown after beam pulse artifact removal. The fiducials were positioned at isocenter sandwiched between 5 cm of solid water. Below: Intensity profiles through the image center are shown. From left to right, the fiducial markers are: (1) Visicoil 0.35 × 5 mm gold coil (0.1 mm center thickness), (2) Gold Anchor 0.2 × 20 mm gold notched wire, (3) SuperLock 0.8 × 4 mm gold bead, (4) SuperLock 0.9 × 5 mm gold coil, (5) Visicoil 1.1 × 10 mm gold coil, and (6) Civco 1.2 × 3 mm gold bead. To enable visualization of the smallest fiducial and more closely mimic what an observer sees when viewing a dynamic cine display, the images shown are the average over four frames. Bottom: Horizontal profiles through the image centers are shown; these were taken from individual image frames. The higher CNR in the FFF versus flat beam is clearly visible.

Image of FIG. 8.
FIG. 8.

The gain in contrast-to-noise-ratio (CNR) for FFF versus flat beams, defined as CNR(FFF) divided by CNR(Flat), is plotted as a function of fiducial marker diameter. Error bars represent the standard error. The FFF beam had substantially higher CNR for all fiducials and for all scatter material thicknesses. The CNR improvement increased with decreasing fiducial diameter under all scatter conditions consistent with the results shown in Fig. 6 . The thinnest Visicoil marker was excluded, because the extremely thin (0.05 mm) gold wire used to create the coil yielded low CNR with high uncertainty for the flat beam.


Generic image for table

Contrast and noise measurements from the MVF image sequences shown in Fig. 7 . Contrast and noise were measured across fiducial and background mask regions on each image frame; the mean and standard deviations were then computed across all frames in the sequence. Scatter material thickness was 5 cm.


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Flattening filter removal for improved image quality of megavoltage fluoroscopy