Miniaturised Low-Cost Gamma Scanning Platform for Contamination Identification, Localisation and Characterisation: A New Instrument in the Decommissioning Toolkit

Round 1

Reviewer 1 Report

In this paper, the authors reported a low-cost, miniaturized and rapidly deployable gamma-ray scanning system to allow the examination of enclosed (internal) or outdoor (external) spaces for radioactive hot-spots. The device is controlled by two stepper motors for pitch and yaw rotations. By controlling gimble, the device scans the environment and reports the dense laser point cloud. Each point has a 4096-dimension radiation feature descriptor measured by a radiation Kromek GR1 sensor. The device was successfully field tested in 2 case studies in Ukraine.

My comments:

1) The paper emphasizes on the scanning equipment, and its contribution to theory and technical method is slightly weak. It is recommended to strengthen the description of detailed theory and method and to further clarify the theoretical contribution.

2) Eq.6 is incorrect.

Author Response

The authors would like to thank the reviewer for their remarks, which we have addressed as follows:

1) We agree that the theoretical contribution was weak and the emphasis of the manuscript was too much on field deployment of the prototype and case studies. We have added theoretical modeling of the prototype, including calculations for field of view, overlap, critical angles, and scan pixel size. We have also added typical values for scans with different parameters and expanded on the integration time problem.

2) We apologise for the errors and oversight in equation 6, this has been corrected.

Reviewer 2 Report

I think this is a good and interesting paper that is almost ready for publication. The only thing I miss is how good is the device compared with other comercial ones? I mean, how good is in terms of quality of results (e.g. resolution), size, handling, cost (at least an estimation of coast) when compared with other similar devices? Such data would be very interesting and at least a comparison with other comercial products could be very interesting and useful for readers. This could be included in the discussion.

Author Response

The authors would like to thank the reviewing expert for their thoughts and recommendations. As the presented prototype was designed with the express objective to fit through small diameter access ports in the nuclear industry, its design was optimised to keep dimensions as small as possible rather than to be as accurate as possible. We have added a complete breakdown of the theoretical model of the prototype, indicating the trade-off between size and spatial resolution for a colimated detector. Additional details regarding size, power management, and deployment have also been added to the data processing sections and to the conclusion, as suggested.

Reviewer 3 Report

Manuscript: sensors 1093914

Title:
Miniaturised low-cost gamma scanning platform for contamination identification, localisation and characterisation: a new instrument in the decommissioning toolkit 

Authors:
Yannick Verbelen, Peter G. Martin, Kamrad Ahmad, Thomas B. Scott

The manuscript describes a system capable of scanning a scene with a gamma ray spectrometer while at the same time collecting the 3D coordinate data of each scanned position. 
The overall idea is attractive, but I have several doubts ad found a few weak points that in my opinion the authors should clarify. 

The authors make use of a commercial gamma ray spectrometer plus a commercial LIDAR system, and assembled them on a two-degrees-of-freedom platform with stepper motors. 
As for the 3D scene reconstruction, the authors rely on an existing software package. 
Then they show two sets of gamma spectra, that represent the obvious output of the commercial spectrometer which can be easily collected with any radioactive sample or NORM material. 
So far no innovation in all this. 
Coming to the relevant part, i.e. the inspections, there is no indication of the effective/typical performance of the device, either fixed or installed on a robot. 
For instance: how long does it take to scan a given area at a given distance? What statistical significance and energy resolution can one expect by inspecting such an area where some specified emitting source is located?
Is the gamma analysis performed online or offline? Does it need an operator or is the peak identification done automatically?
Is it realistic to scan in 4096 bins, as it needs a longer time, or is it better to make a trade-off between energy resolution and acquisition duration, thus rebinning at 2048?
Is the overall performance competitive with existing solutions? How? Is it faster? 

I have the feeling that in the mobile version a non-imaging but more efficient (bigger, and replacing the lead collimator volume with an active detecting volume) gamma detector would be much more effective. It could be moved around, possibly installed on a movable arm, and quickly inspect in close-up mode and find the hot spots: a closer inspection reduces the needed time quadratically with the distance.
If this is not the case the authors should explain why and support it with numbers. 
I do not exclude, however, that the system could be quite useful as part of a waste characterization equipment where, for instance, radwaste drums could be placed for inspection in front and then rotated for the back inspection. 

Finally, I have a technical question. The authors did not mention the geometrical overlap of the inspected area between adjacent points as a function of the distance (it depends on the collimator geometry), which limits the maximum distance where an image still makes sense.

Even though my comments can sound rather hard, I encourage the authors to rearrange their manuscript highlighting the qualifying points, adding more quantitative data in order to prove the realistic usefulness and competitiveness of their device. 

Below I list a few specific comments to the manuscript.

There are a few mismatches in references and figures, they need to be double checked.

line 64, 66: reference to Figure 1? should be Figure 2, I guess. And the real Figure 1 is not recalled in the text.

line 68: Figure 2(a)? should be Figure 3(a). And so on, mismatch of figure numbers.

line 98: I am not sure if this is correct. Perhaps the authors mean -40° to 90° from the horizontal position.

Eq.1: Not all the variables are explained, Ki, mtheta, mphi. Moreover, I fail to see why integrals are used for discrete variables, instead of summations.

lines 119-130: I find this part rather confused. I suggest to simply state what the motors do in the chosen configuration (and eventually what they could do). In the end, no idea is given of the possible duration of a scan, not even of typical values.

line 138: check the formula. I think the reference axes of theta and phi should be explicitly specified. If I understand correctly, theta is referred to the vertical upward axis, thus x3=rcos(theta). If this is the case then x2 is wrong and should be rsin(theta)sin(phi). 

Eq.4: I did not even try to check the correctness of the transformation matrix, I hope the authors will double-check it (it is rather boring for me having to, and I already found x1 and x2 had been wrongly written as equal to each other in line 138). 

line 143: at line 81 it was stated that the collimator aperture is 5mmX5mm, therefore I assume it is a square. But Eq.5 implicitly assumes it is circular. Moreover, the area of such a circle should be pi times the square of the projected radius, not diameter. 
Indeed 2rtan(alpha/2) is the diameter.
In addition, the authors are assuming such a projection as stemming from a single point at the center face of the CZT, thus neglecting penumbra effects. 

line 148: the reference [7] to CloudCompare is wrong, it should be [5].

Figure 4: the figure is strongly penalized by the lack of the visible counterpart image. The intercomparison would be instructive, as it would give the reader the feeling about what kind of contaminated object was involved (like in figure 6 c,d). 

Figure 5: Is the system capable of automatic gamma lines identification? Or does it need a trained operator to do it? How long does it take in the first or in the second case?

line 179-180: it would be useful to know the number of points inspected for figure 4, along with the (typical, if not constant) duration for each point. 

Figure 6a,b: Same as for my comment for figure 4.
Figure 6: no info is provided about the distance where the data were collected from, and about the time duration of the scans. 

 

Author Response

Please see attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

My questions has been addressed by authors. I recommend this paper to be published.

Author Response

The authors thank the reviewer for their contribution.

Reviewer 2 Report

I think the manuscript can be accepted for publication in the present format.

Author Response

The authors thank the reviewer for their contribution.

Reviewer 3 Report

Manuscript: sensors 1093914v2

Title:
Miniaturised low-cost gamma scanning platform for contamination identification, localisation and characterisation: a new instrument in the decommissioning toolkit 

Authors:
Yannick Verbelen, Peter G. Martin, Kamrad Ahmad, Thomas B. Scott

My feeling is that the authors did not get to the indications I suggested in my previous review. 
While they put down a complex graphical/mathematical derivation of the effective angular aperture (with mistakes, in my opinion, as I pointed out), which could have been done quickly with a simplified simulation, they failed to provide the real strength of their device. I did not see a quantitative description of the performance. If the CC-RIAS has to be used in a real decommissioning environment, apart from a couple of images and a qualitative set of spectra one would like to know the activity or dose rate, in order to know whether operators can approach the radioactive scene and for how long. 
The authors have collected a lot of data, why not making them count? 
Information on total and partial dose rates form each identified isotope could be extracted from the data. Why not doing it and reporting the results on the manuscript?
I think that a full chapter about this kind of performance is needed, as it would be quite useful and provide a relevant added value to the manuscript. 
The way it is now, the manuscript sounds like a nice engineering work moonlighting into radiation detection technology. 
I am pretty sure that the CC-RIAS is much more powerful than shown in the manuscript.
Therefore I invite the authors to make a deeper reconfiguration, in order to give it the relevance it deserves. 

----------------------------------------------

line 106
I would remove "from the detector vertically upwards position"

line 108
" foal length" --> "focal length"

line 148
"When choosing the integration time T before starting the scan, care must be taken to not saturate the spectrometer, which happens beyond the threshold of 30 kcps."
This is improperly stated: if the detector is in saturation (because of too close distance and/or high activity), any choice of T will be useless, as the detector will suffer pile-up. The choice of T only depends on the counting rate, in order to collect enough statistics. 
By the way, for a better optimization one could even decide to run at variable time and constant statistics. 

line 151-152
Meaningless if one does not say which is the distance: 10MBq at 100m will be invisible. 

eq.4
alpha is not defined in the text, nor is it indicated in figure 4.
However, I feel that eq.4 is wrong: 
   1) it would be tan(alpha), not sin(alpha); 
   2) it does not take into account the skewness, the limiting ray goes form a corner 
      of the detector to the OPPOSITE corner of the collimator. 
   3) sin(a)=t/sqrt(h^2+t^2+t^2)
It is fine in plane geometry.

eq.5
It inherits the problem of eq.4. 

eq.7
units? m2? cm2? please indicate.

line 198-199
"at an arbitrary angle θ where a scattered gamma photon travels through
the collimator over a distance z and loses energy in the process."
The process is more complex then described here (Compton scattering...). Moreover, by crossing the collimator the gamma rays lose intensity (i.e. a few of them are absorbed or scattered away), not energy. Unless one considers also Compton-scattered gammas. 

Line 205 and eq.9
v is not indicated in the figure. I assume eq.9 is wrong (sin instead of tan)

eq.12
sin(theta) should be tan(theta)

eq.13
alpha and beta are wrong, twofold: (i) sin should be tan; (ii) the numerators should be y+g and x+g. I do not even try to go on with checking the equations up to eq.1, I am not supposed to do the authors' job. 
But I think that instead of these mathematical acobatics the authors could play a simple (approximate) simulation, that can also be done with a worksheet calculation without using sophisticated tools. That should be enough to evaluate the projected cell size and the rough penumbra effect.

line 231 
satistical --> statistical

line 231-234
The low energy data are significant, but generally they mainly come from environmental background (and perhaps electronic noise). Moreover, as their distribution is typically a negative exponential, even tiny fluctuations in the instantaneous behavior of the detector could heavily influence the counting rate. This is why one employs a threshold (E0), which I would not call "high-pass filter" that suggests some sort of transform while it is simply "count only above E0". 

line 236
The reader might wonder how good the pixel separation in figure 6 is, in light of the claimed 35° limit for non-overlapping pixels. Perhaps some explanation is needed, along with the missing info about the distance from the scene and the real size of the framed objects. 

Figure7 and 9
As I said also in my previous review, these figures are not surprising or extraordinary: they are the obvious output of a CZT spectrometer. If the authors wish to put them in the manuscript, the plots should be supported by an ample description of the meaning for the curves. For instance, type of inspection (material), distance, duration, findings versus expectation. 
Moreover, the Y axis has neither ticks nor numbers, nor is it clear whether they are in linear or log scale, which is unlikely for a scientific publication. One gets no idea of the statistics and of the significance of each "peak". 

line 271-274
Additional information is needed: nothing is said about the typical counting rate (from the few numbers shown I presume it should be of the order of 1kcps). 
Moreover, the way I see it 500uSv/h is high and not low radiation level. 

Author Response

All comments have been addressed in the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 3 Report

143. line 155-156The readout electronics of the spectrometer are internally limited to 10 Hz, and the minimum integration time is therefore 0.1 s.
144.  
145. This sentence is unclear: it appears in contrast with the stated 30kcps of the CZT sensor. I suggest to remove it or to rephrase it as
146. the minimum integration time is 0.1 s.

Author Response

The authors wish to thank the reviewer for their comment. The suggested changes have been applied, and the confusing sentence has been removed from the manuscript.