Material and Methods

The experimental study was performed at St. George Hospital Cancer Care Cen- tre, Kogorah, NSW. The objective of this study is to validate the feasibility of HDR BrachyView design in terms of its sensitivity and experimental uncertainty, in partic- ular to assess the reliability of determining the CoM of the source projection (no 3D image reconstruction being performed at this stage; this will be explored in Section 7.3. Figure 7.1 shows the experimental apparatus, which consists of a 4 mm thick planar tungsten collimator with seven double-cone pinholes positioned 9 mm above a single Timepix detector (as measured from the top surface of detector to the centre plane of collimator). The planar collimator is used in this experiment rather than the cylindrical collimator as it permits more convenient repositioning of the detector for preliminary experimental characterisation of the design, and is considerably easier and cheaper to manufacture.

A catheter was placed inside a solid water prostate phantom, in a channel which is parallel to the collimator pinholes and positioned 45 mm above the central axis of the collimator. This height was selected as the worst case in terms of source tracking

Figure 7.1: Experimental apparatus.

accuracy and uncertainty which are proportional to the source to detector distance. The source positions in a close proximity of the detector plane may also be interesting to study. However, due to single Timepix was used in this experiment, it is difficult to obtain multiple full size source projections for most source depths in the catheter placed most close to the detector (5 mm above the collimator surface). Measurement of source placed to a close proximity will be interesting to investigate the FoV of HDR BrachyView system in the future study with a prototype of full length Timepix detec-

tor array. A Flexitron afterloader was used to deliver a 192Ir source into the phantom

via the catheter. At the time of the experiment, the source activity is approximately 217.6 GBq; this source was stepped through the catheter in 1 mm increments, with a dwell time of 1.7 s for each source position. This dwell time is equivalent to 1 s for a standard 370 GBq source, which is calculated by afterloader system.

The Timepix detector was coupled to the Fitpix USB data acquisition module and connected to a dedicated data acquisition laptop. The detector was operated with a frame duration of 0.5 seconds per frame, with each frame displayed in real time via the Pixelman user interface. The index of the frame which registered the first source projection was recorded as the start frame. Since the source dwell time is 1.7 s, the 3 frames following the start frame correspond approximately to one particular source position. The position of the CoM for each projection in these images was then calculated as per the methods discussed in Chapter 4. As shown in Figure 7.2,

h h D D sp pd _p c Source

Figure 7.2: Schematic showing distance between CoMs

the distance between the CoMs of any two adjacent source projections (Dc) can be

expressed as:

Dc=

hpd+ hsp

hsp

× Dp (7.1)

where hsp is the distance between the source and collimator, hpd is the distance

between the detector and the collimator, and Dp is the distance between adjacent

pinholes.

Because the value of hsp, hpd and Dp were kept constant as 45 mm, 9 mm and

6 mm, respectively, throughout the experiment, Dcon the detector plane also remains

constant and is expected to be 7.2 mm. This quantity was measured and used to estimate the uncertainty of the HDR BrachyView probe in locating the projection’s CoM.

7.2.2

Results

Two projection frames in which the source has been projected through adjacent pin- holes on the detector are shown in Figure 7.3. The position of the centres of mass

15mm 15mm 0 50 100 150 200 250 50 100 150 200 250 0 100 200 300 400 500 600 (a) Position 1, hsp= 43 mm 15mm 15mm 0 50 100 150 200 250 50 100 150 200 250 0 100 200 300 400 500 (b) Position 2, hsp= 43 mm

Figure 7.3: Projection images of the source through two pinholes on the detector plane for two different source positions; exposure time is 0.5 s.

180 190 200 210 220 230 240 250 260 0 5 10 15 Frame number

Centre positions on detector plane (mm) in direction along source movement

Figure 7.4: Positions of the source CoM on the detector plane

of the source projections through two adjacent pinholes for all source locations were plotted and are shown in Figure 7.4. The movement of the projection on the detector plane along the direction of the source movement in constant increments is clearly vis- ible. The measured distance between the centres of mass of two adjacent projections on the detector plane is also shown in Figure 7.5. The average distance between the

two projections in same frame was measured to be 7.1±0.1 mm and are shown for all

180 190 200 210 220 230 240 250 260 6 6.2 6.4 6.6 6.8 7 7.2 7.4 7.6 7.8 8 Frame Number

Distance between two source image centres(mm)

Figure 7.5: Distance between CoMs of projections on detector plane

7.2.3

Discussion

The experiment serves to validate several basic concepts used in the HDR BrachyView design. Although the experiment does not yet provide an estimate of the error of the estimated source position in three dimensions, it is a necessary preliminary step in quantifying the error in the detector plane (which is directly related to the final positioning error via the geometry of the detector and collimator arrangement).

The plot of the CoMs (Figure 7.3) demonstrates the capability of HDR BrachyView to resolve the projection CoMs when the source was placed 43 mm above the collimator. The Timepix has not been masked or equalised, and the raw response of the detector can be seen in the images, which show an interesting concentric artifact resulting from non-uniformities in the doping of the detector’s semiconductor substrate (bias voltage of 100 V was applied to the sensor). The appearance of a minority of samples in an adjacent position are due to the difference between source dwell time and interval of samples. Since the frame duration is 0.5 s per frame and every third frame was plotted, the interval between samples is 1.5 s which is 0.2 s shorter than the source dwell time (1.7 s). This makes it possible that two adjacent plotted frames are obtained when

source is in same position. The value of hsp of 43 mm corresponds to the most distant

to be able to to track with the specified sub-1 mm accuracy, which represents the worst case of ‘minification’ seen within the treatment volume.

As shown in Figure 7.5, the consistency of the distance between the CoMs of two projections in each frame is a confirmation of the ability of the system to precisely track the CoM of the source image independently of the angular view of the source. The distance is estimated with a maximum error of 0.2 mm on the detector plane, which includes the uncertainty of both CoMs. Therefore, for the source locations evaluated (43 mm above the plane of the centre of the collimator), the maximum error in CoM location determined by the HDR BrachyView probe is approximately 0.1 mm