Reviewer's report: The work under review on “shared data for IMRT optimization research the CORT dataset” by Craft et al deals with providing a set of dataset to the researchers working in radiation oncology optimization. This type of shared data is very much needed by optimization community in order to calculate and compare the various algorithms and evaluate their performance.

General purpose of this paper is incredibly useful and practical. A lot of the research in radiotherapy optimization is being done by individuals who do not have easy access to the type of information that this study provides. There are four sets of contoured CT dataset provided along with the dose influence matrices which can be used to perform optimization calculation. Overall the paper is well written, where the authors have provided an example linear programming optimization for the four cases. The work requires improvement in order for other researchers to make use of this valuable tool.

Detailed Comments:

Major Compulsory Revisions:

1. Dose influence matrices are provided, however no mention of what beam energy was used to generate these matrices. Do all matrices correspond to 6MV beam or various sites have other clinical energies such as 10MV, 6FFF, 10FFF, 15 MV and 18MV. It would be beneficial to have various energy based fluence matrices for different sites. More detail need to be included with regards to dose calculation.

2. More detailed information into the parameter setting used in the dose deposition matrix calculation (in the IMRT GUI) such as the “dose term” and “scatter method” selected. This need not be stated this in body of the paper but should be either added as an extra text file with the code to disclose this information if expectations are that other researchers would use their pre-calculated dose deposition matrix.

3. Why was the dose deposition matrix rescaled to Gy/MU?

4. Your code provides info on x and y coordinates of the beamlets. Can you use graph the fluence map for each beam from the resulting fluence vector solutions obtained from optimization. Does this beamlet position information also allow you to create a fluence vector from the optimization solution to be read in by a commercial treatment planning system so one could re calculate dose to get more accurate result.

5. A dataflow diagram providing the context of these images sets in the overallpicture of a radiotherapy treatment plan development would be very useful and is strongly recommended.

6. The amount of data is huge, it would be useful to mention that the total data amounts to 70-100 GB. I was not able to download the data during for review from “http://aspera.gigadb.org/” however ftp site worked but was very slow. How would a user access the data? What is the permanence of data on these data servers – probably related to the GiGScience policies? Do you have an alternative way of providing access of the datasets? Current method has a few bugs (one has to sign in multiple times; code doesn’t start downloading when you click on links).

7. The authors spent quite a bit of time describing the indexing voxel coordinate system e.g. Fig 2-3 on page 11. How much the indexing differs from that generated by CERR, if indexing is same these figures would be unnecessary and a reference to CERR would suffice.

8. CERR is an open source environment used for development to optimization community. CERR also provides sets of Dicom datasets, how unique are the datasets provided here. How does this tool complement the CERR’s features? It would be beneficial to see a comparative table in discussion (/a paragraph in background) for salient features of the CORT vs CERR.

Minor Essential Revisions:

• Pg 5 : Add ‘beamlets’ after 3678 in SBRT liver section

• Pg 7 line 13 - correct “Prlem”

• Pg 8 discussion – correct “reasearchers “

evel of interest:

An article of importance in its field

Quality of written English:

Acceptable

Statistical review:

No, the manuscript does not need to be seen by a statistician.

Declaration of competing interests:

'I declare that I have no competing interests'

Reviewer's report:

Discretionary Revisions:

Comments 1) and 2) - The authors still did not explicitly state what energy was used for calculating the dose deposition matrix for the four different cases. However, it was stated that the default settings were used in the IMRTP GUI and specific the primary and scatter options were used when calculating the dose.

Comment 3) - It was not addressed why they rescaled the dose matrix to Gy/MU instead of the usual units of Gy/fluence.

Comment4) - The authors only specified that for the CT data scans and DICOM files that they can be imported into a commercial TPS but there is no mention that this can be done for the beamlet solution to import it into a TPS and recalculate.

Comment 8) - They do not provide a comparative table as previously suggested. They do mention throughout how this CORT system can be used.

Other Corrections/Suggestions:

• Page 10, line ~ 13, under DAO and VMAT applications - should fix "To than end" so it reads "To that end"

• There are no units provided in any of the tables provided. The authors should at least specify in the table heading that all doses are in Gy etc. - table 2 to table 6.

• They refer to the equation d = Dijxi quite often throughout the text. It should be given an equation number when they first present it (bottom of page 4) and then refer to this equation in the remainder of the article.

• Midway through page 5 it is mentioned OAR but never stated what this acronym refers to. Should at least say Organs at risk (OAR) here and then can use OAR term like they do for the remainder of the article.

• Placement of figure 5 is awkward

• Should explicitly state where the data can be obtained from.

Level of interest:

An article of importance in its field

Quality of written English:

NA

AcceptableStatistical review:

No, the manuscript does not need to be seen by a statistician.

Declaration of competing interests:

'I declare that I have no competing interests'