Charting the development of optical CT and allied methods for 3-D radiation dosimetry: 2022 update - mapping the global research community

To recognise the many and varied contributions from groups around the world over more than 25 years of research into 3-D radiation dosimetry using optical imaging (primarily optical computed tomography), a systematic review has been conducted with the aim of charting the development of this field. Some 493 publications matched the criteria of the review, and these have been coded with information relating to the category of research, the type of dosimeter material used and the geographical location where the research was conducted. The results of the analysis demonstrate a steady growth in the number of academic groups involved, from 4 at the first DOSGEL meeting in 1999 to over 30 in 2016. The important role played by the DOSGEL and IC3Ddose conference series is demonstrated by the marked biennial trend in publications. The division of publications between categories was 36% materials development, 34% scanner technology, 24% applications, 5% review and 1% teaching. The most common application areas were brachytherapy, proton/particle therapy and IMRT/VMAT. Prior to the introduction of PRESAGE® in 2003, polymer gels were the most popular material, whilst radiochromic dosimeters have dominated in recent years. These and other trends are discussed on the basis of the data presented. A brief discussion of the future of the field, including some personal opinions, ends this brief review.


Introduction
With the series of conferences inaugurated by the First International Workshop on Gel Dosimetry (DOSGEL '99 held in Lexington, Kentucky) now well into its third decade, it seems appropriate to attempt to provide an overview of what has become a significant body of work in the area of optical computed tomography (CT).A previous review article [1], prepared for the DOSGEL conference in 2008 (Hersonissos, Crete) aimed to capture the state of a rapidly-developing field; in 2010, optical CT was described briefly as part of a broader review of polymer gel dosimetry [2]; and in 2017, Khezerloo et al. reviewed the use of the radiochromic dosimeter PRESAGE® [3,4].However, none of these publications were formal, systematic reviews of the literature.
The work presented here adopts the systematic approach in order to chart the growth of the field from an originally small number of pioneering laboratories to its current state, to acknowledge the many groups around the world who have played a role in its development, and to provide quantitative bibliometric indicators of the distribution of effort in the various sub-fields.
I closed my previous DOSGEL review in 2008 with the words "Optical CT has now arrived" and the work conducted here confirms that, in the 12 years since that article, this statement has been borne out.

Review protocol
Following the overall outline set down in Figure 1 of Xiao and Watson [5] -itself the product of a systematic review of best practice -the procedure below was adopted.
• Formulate the problem and define the research questions.• Assess quality: no formal assessment was performed, because the research aim was to discover the full extent of the research community, not to answer questions regarding the accuracy or efficacy of this type of dosimetry.
• Analyse and synthesise: data were first coded in Excel according to the headings described below; the dataset was then analysed using RStudio v1.4.1106 (https://rstudio.com) to produce quantitative statistics, which were mapped spatially.

Definition of research question
The aim of the research was to map the development of the technique of optical CT and allied methods for radiation dosimetry, highlighting the unique role that the DOSGEL and IC3DDose series of biennial conferences has played in advancing the field.From the very start, the meeting organisers aimed to be inclusive, welcoming a diversity of contributions, and this has been reflected in broad inclusion criteria here.To prevent the review from becoming unmanageable in size, the subject matter was restricted to 3-D optical imaging methods applied to "integrating" dosimeters, and the development of the materials used for these.This meant that a number of other exciting and no less important optical methodologies, in adjacent areas such as scintillation imaging and Cerenkov imaging, were excluded from the review.Similarly, the large body of work in gel dosimetry based on Magnetic Resonance Imaging was excluded, even though many of its goals are the same as for optical CT.

Literature search
The five "seed" datasets were: • a PubMed query conducted in March 2022 with the following search expression: (("three-dimensional dosimetry") OR ("3-D dosimetry") OR ("3D dosimetry") OR (gel dosimetry) OR ("radiation dosimetry") OR (radiochromic) OR (Fricke) OR (PRESAGE)) AND (("optical computed tomography") OR ("optical CT") OR (spectrophotometer)) 161 entries • the reference list of the 2008 DOSGEL review by Doran [6] 123 entries • the reference list of the optical CT section of the 2010 review by Baldock et al. [2] 29 entries • the reference list of the Khezerloo et al. review of 2017 [3] 92 entries • references matched manually from the combined DOSGEL and IC3Ddose abstract books 1999 -2021 149 entries The five seed datasets were combined in EndNote 20 (Clarivate Analytics), which removed obvious duplicates.The resultant datasets were screened for inclusion, rejecting articles that did not relate to the research question.In most cases, this was possible simply by inspecting the title, but sometimes reading the abstract or scanning the entire text was necessary to determine relevance.The above steps are easily replicable and would yield close to identical results for all users.
From this point, the study proceeded iteratively, with the aim of discovering the most comprehensive dataset possible within the time available for the project.Extensive use was made of Google Scholar in the following ways: • keyword search using similar phrases to the PubMed search above -Google Scholar returns an extensive list of hits, of gradually diminishing relevance.For example, the search string "gel dosimetry optical CT" returns 11,000 "hits" (some duplicated) making it, to all intents and purposes, impossible to conduct an exhaustive search.Items were displayed in "screens" each containing 10 references and for each search, the stopping criterion chosen was (n = 50 screens) OR (m = 5 consecutive screens with no "hit" of interest found).
• "backward" searching based on scanning the reference lists of selected articles already discovered; • "forward" searching using Google Scholar to list all articles citing selected articles already discovered; • author searches -based on a knowledge of the field, exhaustive searches were performed for key players involved in optical CT research.
This iterative part of the process was more time-consuming but yielded approximately 40% of the final total of 493 references, demonstrating the importance of the interactive part of the search.By definition, this step is not replicable, not least because the results returned by the Google search engine are tailored to the individual performing the search.

Data coding and processing
Each publication in the Excel spreadsheet was "coded" with the following additional information: • the city name associated with the academic group from which the publication came, taken from the affiliation of the lead or senior author as appropriate; • the longitude and latitude of the city, found using the geocode function of R package ggmap, which in turn makes use of the Google Cloud Platform Geocoding API; • flags marking the reference as follows: o conference abstract (e.g., IC3Ddose), research article, review or teaching article; o category: materials development, scanner technology or applications (some publications fell into more than one category); o type of gel; o type of application An R script was written to return the bibliometric results presented below.The get_googlemap() function from the ggmap library (in turn calling the Google Cloud Platform Static Map API) was used to obtain a world map and the locations of research groups were plotted.

Results and Discussion
Figure 1(a) shows the variation with publication year of the number of publications meeting the review criteria, starting in the mid-1970's with early work by Gupta and co-workers (e.g., [7].Despite the existence of an earlier conference abstract from Tarte et al. [8], the first experimental demonstrations of optical CT scanners are generally credited to Gore et al. [9] (scanned laser) and Winfree et al. (CCDbased tomography system) [10].From these beginnings, the number of publications rose rapidly, peaking at around 45 per year in 2010.There was a subsequent decline but still significant interest throughout 2010's, followed by a sharp fall in 2020 largely related to the COVID-19 pandemic, although it may also reflect a trend towards other methods of dose verification.2022 has seen a significant rebound and, once the IC3Ddose contributions (not yet analysed) for 2022 are included, a healthy resurgence of the field is likely to be found.The biennial trend in abstract publications is strongly influenced by the DOSGEL and IC3Ddose conference series, which have long been an important forum for presenting this type of research.A weakness in the research methodology of this review is that, whilst the Journal of Physics Conference Series (where the DOSGEL and IC3Ddose proceedings are published) and Medical Physics (where the American Association of Physicists in Medicine conference proceedings are published) are both listed on Google Scholar, other sources of abstract information may not be, and so some relevant material meeting the inclusion criteria may not have been captured, thus biasing the results.Nevertheless, the extended abstract format of DOSGEL/IC3Ddose often provides important information that is not found in full journal articles and so a focus on these sources is valuable.Notice that the peaks shown in green in Figure 1(a) are irregularly spaced, which reflects occasions where proceedings were published in the year following the conference.Year ranges for aggregation of results are not evenly spaced but reflect the timings of the DOSGEL/IC3Ddose conferences, as these have been major drivers of the global research effort.(A known methodological weakness here is that different academic groups in the same city are amalgamated and this will be addressed in future bibliographic work.) Figure 1(c) presents an overview of the findings related to application area.Three applications dominate the uses described to date: brachytherapy, IMRT/VMAT and proton/particle therapy.Although the numbers involved overall are small (data not shown), in each case, the majority of publications occurred in the decade between 2005 and 2015 with only sporadic work performed since.Recent applications-related publications have primarily been in the areas of small-field treatments and MR linacs.These trends suggest that the radiotherapy community sees 3-D dosimetry as primarily of use in the early phases of introducing new technologies to the clinic and that, once this initial confidencebuilding phase is complete, other less unwieldy tools are used on a day-to-day basis.
Figure 1(d) shows results concerning the dosimeter type.Following Gupta's work [7] there was strong initial interest in Fricke dosimetry combining ferrous sulphate with xylenol orange and gelatin.For most of the 1990's and early 2000's, the materials most commonly employed were polymer gels, which were first introduced in the context of MRI-based dosimetry [11] and later characterized for optical measurements in [12].However, since the invention of the PRESAGE® dosimeter in 2003 [13], radiochromic dosimeters have once more predominated.Contrary to the situation described above for applications of 3-D dosimetry, research into new dosimeter materials remained strong until 2019, immediately before the pandemic.This suggests, on the one hand, a wealth of ideas and new chemical systems to investigate, but on the other, an acknowledgement that, despite more than 25 years of active research, no existing material fulfils all the criteria necessary for a tool that is easily usable in the standard clinical workflow.Arguably, the difficulty of obtaining either absolute dose quantification or results that are reproducible from batch to batch is a fundamental limitation of chemical methods of dosimetry.
Figure 2 maps the global distribution of research groups discovered by the systematic review for the entire period from 1974 to spring 2022, whilst Figure 3 presents the same information as a series of snapshots leading up to each of the DOSGEL and IC3Ddose conferences.The size of each marker symbol relates to the fraction of total world research output in the relevant time-window as conducted at each of the laboratories, and thus the symbol sizes do not represent the same absolute numbers of publications at different time points.The marker colours reflect the mix of categories (as defined on the inset Maxwell colour triangle) published by individual research groups.
This mode of presentation makes it easy to identify trends in research.For example, North America as a whole (with notable exceptions) has focused largely on scanner technology and applications, while Asia (again with exceptions) has concentrated on dosimeter development.The time series shows the interplay between categories of research.For the first two DOSGEL meetings, global research focused on the development of materials and early generations of optical CT scanners.During the period 2004 -2008, scanning technology was the dominant theme of the literature, whilst from 2008 -2014, the overall colour of plotted points changes from red to blue, indicating a readiness of the technology to tackle real-world radiotherapy applications.In more recent years, as limitations of the existing systems have become apparent, there has been a return to further rounds of materials development -with the emergence, for example of flexible dosimeters [14,15] -novel scanner designs [16] and improved reconstruction techniques [17].Predicting the future is, as ever, not easy.However, massive current interest in FLASH radiotherapy [18,19] seems likely to translate into a need for dosimetry tools to monitor treatments that could be revolutionary but also have the potential to be dangerous if delivery errors occur.Previous research in synchrotron microbeam therapy [20] suggests that optical CT imaging of radiochromic plastics will have an important role to play in these high dose-rate scenarios, and I see this as a key application area.
In terms of equipment development, optical CT is still an inconvenient technique, relying as it does on refractive index matching via a liquid-filled tank.Over the years, a variety of studies have investigated the extent to which the index matching medium can be minimised [21] or dispensed with entirely [22,23] but further progress is still needed towards a rapid-scanning tool usable in the clinic by regular (i.e., non research) medical physicists.
As discussed above, it is this author's belief that the non-reproducibility of dose-response calibrations between dosimeter samples (even from the same batch) will continue to be a major stumbling block in any bid to gain more widespread adoption of optical CT technology.To date, 3-D chemical dosimeters have been single-use devices.If each one continues to require lengthy calibration procedures, or is capable only of relative dosimetry, it will be difficult to incorporate this method into the clinical workflow.Attempts to create reusable chemical dosimeters (e.g., [24,25]) have so far met with limited success, relying as they do solely on an unstable material response -itself a major problem, limiting flexibility in the timing of the device readout -that decays back to an equilibrium value over time.In this context, one exciting development on the horizon is the introduction of new imaging materials and methods based on optical stimulated luminescence (OSL) [26].Distributing OSL particles within an elastomeric matrix holds out the tantalising prospect of a flexible and reusable 3-D dosimeter that is reliant on physical properties, rather than chemical reactions, and is erasable via an annealing process.

Conclusion
The last 25 years of research into optical CT imaging and related methods have seen a large and diverse literature develop with contributions from more than 30 different groups all around the world.It has become traditional to predict the demise of 3-D dosimetry, only to find that, at each successive IC3Ddose conference, new insights and discoveries revitalise the field, and it is my hope that this year will be no exception.

Figure 1 :
Figure 1: (a) Full journal publications and conference abstracts as a function of year of publication, with labels corresponding to the IC3Ddose conference series added.(b) Number of active research groups in the areas covered by the review vs year range.The left-hand bin represents relevant research prior to the first DOSGEL conference apart from the 1999 bin, all subsequent points include both the year of the conference itself and the year(s) between successive DOSGEL/IC3Ddose conferences.(c) Number of publications in each of the different application areas coded.(d) Number of publications for each of the types of dosimeter material coded.

Figure 1 (
Figure 1(b) shows the number of groups worldwide engaged in research in scope for the systematic review.This exhibited steady growth during the years from 1999 to 2017 and fell slightly in the 2018 -2019 period.The subsequent larger fall during the 2020 -2022 period again likely reflects the global impact on research due to COVID-19 rather than a genuine reduction in number of academic groups.

12th 6 Figure 2 :
Figure 2: Location of research groups involved in the study of 3-D radiation dosimetry using the optical methods in-scope for this review (primarily optical CT).Each coloured dot has a size reflecting the publication output of the group.Dot colours reflect the mix of publication categories produced by the laboratory, as indicated by the colour triangle key.The inset pie chart shows the overall balance of research across the community between publication categories.

Figure 3 :
Figure 3: Tine-resolved snapshots of research activity in the field covered in this systematic review.Colour-coding of research categories is the same as for Figure2.Note, however, that symbol sizes are relative to the total research activity across all centres for the particular time period and so a given symbol size reflects different absolute numbers of publications in each plot.

•
Search the literature: o Start with five "seed" datasets.o Combine and de-duplicate the "seed" datasets.o Screen the combined dataset according to the inclusion criteria.o Use Google Scholar iteratively (see below) to discover further data, employing a stopping criterion.