Computational Pathology Portal

You can access the CP Portal (AstraZeneca) using the following link: CP Portal.

You can access the CP Portal (Definiens) using the following link: CP Portal.

If you are an AstraZeneca user you can simply log in using your AZ single sign-on (SSO) credentials and start using CP Portal right away.

For users external to AZ (CROs, academic collaborators, partners, vendors, etc.) you need someone from AstraZeneca to request an account to be created for you, before you can log in and use the CP Portal.

To request an account to be created for an external user, please follow the steps in the FAQ below "How to I request to add an external user to the CP Portal?".

You can write a message in the Send a message in the Teams channel: Dictionaries and answer the following question:

Alternatively, you can send an email to DEMU_ProductManagement@astrazeneca.com and answer the following question:

• What is the new term?
• Explanation of the requested term.

Process description:

1. In case of the Team channel, Data governance team will manage the discussion
2. In case of the Team channel, SData governance will reach out with information to the PO
3. PO will contact the responsible team for approval, if not already provided
4. PO creates an item in backlog
5. Team will implement the requested change
6. PO informs the user
7. PO informs the IODA team to make new markers available in IODA

Email the CP Portal Support team and answer the following question:

• What is the Email address of the external user?
• What is the End date of access for the external user to the CP Portal?

The Support team will take care of account creation and inform the external user about the login information.

Yes, there is! Slides should be converted. Email the CP Portal Support team and answer the following question:

• What is the project in which slides should be converted?

The Support team will inform the CP Portal team who will trigger the slide conversion. Once the process is done, the user will be notified

Email to DEMU_ProductManagement@astrazeneca.com and answer the following question:

• Which indication or indication subtype should be added?
• Why is it needed?
• When is the new class required?

Process description:

1. Responsible team will review and approve or reject the request
2. PO creates an item in backlog
3. Team will implement the requested change
4. PO informs the user
5. PO will update the indication and indication subtype overview pages: Indication, Indication Subtype

Email to DEMU_ProductManagement@astrazeneca.com and answer the following question:

• Which type of annotation should be added?
• Why is it needed?
• Is the new class aligned with the IODA and drcs dictionary? See the IODA dictionary overview for details

Process description:

1. PO will contact the responsible team for approval
2. PO creates an item in backlog
3. Team will implement the requested change
4. PO informs the user
5. Responsible team will update the annotation overview page: What are the supported region annotation classes?

Send a message in the Teams channel: Dictionaries and answer the following question:

• Which class, feature/object variable be added and why is it needed?
• When is the new entry required?
• Does this entry have to be added also to other dictionaries (e.g. annotation classes, readouts, ...)?

Process description:

1. Responsible team will review and approve or reject the request
2. PO creates an item in IODA backlog
3. Team will implement the requested change
4. PO informs the user
5. Responsible team will update the IODA dictionary page, if needed and ensure that changes are propagated to other dictionaries

For a detailed description on the Data Ownership process, roles, and responsibilities, please see here.

To request a Data ownership role or to have a project marked as a Home project:

1. Please write to DEMU_ProductManagement@astrazeneca.com

2. and indicate:

   • The user who should be assigned the role, or
   • The project which should be assigned the Home project status, and
   • The reason (project) why you need these privileges
3. PO will inform Portal data admins
4. Portal data admins will then assign the right permissions in the CP Portal
5. Portal data admins will inform the user

To delete a project:

1. The Project admin or the Project data owner adds "TODELETE_" in front of the project name

2. Write a ticket to the Support team, and indicate:

   • the URL of the project
   • if all project related data should also be removed (annotations, DRCs etc.)
3. Support will remove you from the project and dev teams schedule a deletion at a later time point

The complete list of supported Region annotation classes can be found here: Region annotation classes.

The complete list of supported Region annotation classes can be found here: CMA classes.

You can find the controlled vocabularies for the Portal here: Metadata Table: Data dictionary.

Click on the blue entry in the list to lear more about the allowed entries.

Note: Slides deleted in the last 60 days can be restored. If the slides have been deleted more than 60 days ago, they are permanently deleted and cannot be restored.

If you have accidentally deleted slides from a project and you would like them restored, please contact the Support team.

Please provide the ID of the project from which slides were deleted, to the Support team. They will be able to help and guide you from there.

You can find an extensive list of computational pathology tools and applications here.

Protocols from Cambridge TransPath Lab

Point of Contact: DEMU_ProductManagement@astrazeneca.com.

• If you want to know more information about a protocol from an assay performed in Cambridge Trans Path lab, Covance, or Akoya please follow the link to Assay protocol.
• You will land in the Translational Pathology SharePoint Site for IHC Protocol Library. There:

• You can find a specific protocol searching by the unique ID reference number (ID Ref #)
• You can find Covance protocols in the "Covance Protocols for TICA" folder
• You can find Akoya protocols in the "Akoya Protocols for mIF" folder

• If you want to understand how those assays were validated, please follow the link to Validation reports.
• You will land in the Translational Pathology SharePoint Site (see snapshot below) for Validation Reports. There:

• You can find a specific validation report for an assay searching by the assay name

Protocols from CPSS

Point of Contact: DEMU_ProductManagement@astrazeneca.com.

• If you want to know more information about a protocol from an assay performed in CPSS:

  1. Check the excel list below to find the reference ID for your assay of interest
  2. Use the contact from above and they will email you the requested protocol

• List of protocols

  ID Prefix	ID# (XXXXX)	Name	Instrument
  PM_LS	00029	CD28(DAB).pdf	Leica Bond
  PM_DM	00001	CD3(Pu)+CD68(DAB)+CK(Yel).pdf	Ventana Discovery
  PM_DM	00013	CD3(Pu)+CD68(Yel).docx	Ventana Discovery
  PM_DM	00008	CD3(Pu)+Ki67(Yel).pdf	Ventana Discovery
  PM_DS	00018	CD34 (DAB).docx	Ventana Discovery
  PM_LS	00012	CD39 (br).pdf	Leica Bond
  PM_DS	00040	CD39.pdf	Ventana Discovery
  PM_DS	00019	CD4 (DAB).docx	Ventana Discovery
  PM_DS	00020	CD40 (DAB).docx	Ventana Discovery
  PM_DS	00021	CD56 (DAB).docx	Leica Bond
  PM_DS	00022	CD56 (Pu).docx	Ventana Discovery
  PM_DS	00023	CD73 (DAB).docx	Ventana Discovery
  PM_DS	00014	CD8 (SP239).docx	Ventana Discovery
  PM_DM	00002	CD8(DAB)+FoxP3(Pu).pdf	Ventana Discovery
  PM_DM	00003	CD8(Pu)+FoxP3(DAB)+CK(Yel).pdf	Ventana Discovery
  PM_DM	00028	CTLA4(Pu)+FoxP3(Yel).pdf	Ventana Discovery
  PM_DS	00010	CXCL9 (Pu).pdf	Ventana Discovery
  PM_DM	00006	CXCL9(Pu)+CD8(DAB).pdf	Ventana Discovery
  PM_LS	00036	EGFR (DAB).pdf	Leica Bond
  PM_DM	00015	GITR(Pu)+FoxP3(DAB)+CK(Yel).docx	Ventana Discovery
  PM_DS	00016	Granzyme B (DAB).docx	Ventana Discovery
  PM_DS	00017	Granzyme B (Purple).pdf	Ventana Discovery
  PM_LS	00030	IL-33.Nessy-1(DAB).pdf	Leica Bond
  PM_DM	00009	Lag3(Pu)+CXCL9(DAB).pdf	Ventana Discovery
  PM_LM	00007	Lag3(Red)+PDL1(DAB).pdf	Leica Bond
  PM_DS	00033	MBP Assay Transfer Form signed.pdf	Ventana Discovery
  PM_LS	00027	NKp46 Final Assay.pdf	Leica Bond
  PM_LM	00038	NKp46(br)+CD20(red).pdf	Leica Bond
  PM_DM	00025	OX40 (DAB)+FoxP3 (Pu).pdf	Ventana Discovery
  PM_LS	00026	p16 DAB Protocol Form.pdf	Leica Bond
  A_Procedure	1	Pathology Method numbering convention_work instruction.docx	Combo
  PM_LM	00004	PD1(DAB)+PDL1(Pu).docx	Combo
  PM_DS	00011	Perforin (Pu).pdf	Ventana Discovery
  PM_DS	00005	pHH3 (DAB).pdf	Ventana Discovery
  PM_VS	00041	TROP2(DAB).docx	Ventana Discovery
  PM_BS	00024	V-PD-L1(SP263) Benchmark Protocol #15.pdf	Ventana Benchmark

You can download the List of assays by method ID, from 04.23.21 (Excel file).

Slides have different status depending on where they are in the upload pipeline process.

Image: Slide status after upload

If the slide status is available for longer than 10 minutes and you cannot view the slide in the viewer, please email the Support team and submit the URL from the project page.

Please use Google Chrome on Windows 10 or later. While different browser and operating systems may work, we cannot guarantee that all feature will work as intended in any other browser or operating system.

Background information

• Images of large size (e.g. 10'000 x 10'000 pixels and larger) can be very slow to visualize in any viewer if they don't incorporate optimized representation (tiled storage of pixel data and downscaled overlays or pyramids). The CP Portal is not an exception - if you upload such image it will be not possible to visualize it completely on the screen (with e.g. 5% scale from origin) in reasonable time (e.g. faster than 10 seconds). It is good to make sure that TIFF files you are uploading are fitting these requirements (tiles + pyramids), especially if these files are created by you or be someone who can alter the process of image creation. In this article there are 2 sections:
• how can you check that a TIFF file has necessary data (and in a right form)
• what can you do to make sure created images are having right structure
• what can you do to improve existing TIFF files

How can you check that a TIFF file has necessary data (and in a right form)

• On high level - you can use various software utilities that will check format and metadata for your review. There are plenty of tools that might help with this:

• identify (from imagemagick tool)
• gdalinfo (from GDAL package)
• some other tools could be used (e.g. QuPath tool allows to open TIF files and shows in the metadata were pyramids taken from a file or are generated on-the-fly by slowing down the viewing process)

Example of use

• Image Magick identify tool, to get information about the file, make sure you have the tool installed and then run from the CLI:

  $ magick identify a-file-name.tif
  ...

     Click here to expand "identify" example...

   

  Example use of identify for an image file with 4 overlays (pyramids).

  A GOOD image: pay attention that dimensions of every next level is continuously decreasing (it could be by 2 or by 4 or by other factor).

  $ magick identify image-file-xyz.tiff
  image-file-xyz.tiff[0] TIFF 1024x1024 1024x1024+0+0 16-bit Grayscale DirectClass 42.7KB 0.000u 0:00.000
  image-file-xyz.tiff[1] TIFF 512x512 512x512+0+0 16-bit Grayscale DirectClass 42.7KB 0.000u 0:00.000
  image-file-xyz.tiff[2] TIFF 256x256 256x256+0+0 16-bit Grayscale DirectClass 42.7KB 0.000u 0:00.000
  image-file-xyz.tiff[3] TIFF 128x128 128x128+0+0 16-bit Grayscale DirectClass 42.7KB 0.000u 0:00.000
  image-file-xyz.tiff[4] TIFF 64x64 64x64+0+0 16-bit Grayscale DirectClass 42.7KB 0.000u 0:00.000

  A SLOW image: notice that it is shown only 1 image layer of large dimensions 31042 𐄂 22080 pixels:

  $ magick identify image-file-no-pyramids.tiff
  image-file-no-pyramids.tiff TIFF 31042x22080 31042x22080+0+0 8-bit sRGB 311.80MiB 0.000u 0:00.000

• GDAL gdalinfo tool, to get information about the file, make sure you have the tool installed and then run from the CLI:

  $ gdalinfo a-file-name.tif

     Click here to expand "gdalinfo" example...

   

  Example use of gdalinfo for an image file with 4 overlays (pyramids).

  A GOOD image: pay attention that there are many sub-datasets of various dimensions:

  $ gdalinfo image-file-good.tiff
  ...
  ...
  Subdatasets:
  	SUBDATASET_1_NAME=GTIFF_DIR:1:image-file-good.tiff
  	SUBDATASET_1_DESC=Page 1 (24960P x 34560L x 1B)
  	SUBDATASET_2_NAME=GTIFF_DIR:2:image-file-good.tiff
  	SUBDATASET_2_DESC=Page 2 (12480P x 17280L x 1B)
  	SUBDATASET_3_NAME=GTIFF_DIR:3:image-file-good.tiff
  	SUBDATASET_3_DESC=Page 3 (6240P x 8640L x 1B)
  	SUBDATASET_4_NAME=GTIFF_DIR:4:image-file-good.tiff
  	SUBDATASET_4_DESC=Page 4 (3120P x 4320L x 1B)
  	SUBDATASET_5_NAME=GTIFF_DIR:5:image-file-good.tiff
  	SUBDATASET_5_DESC=Page 5 (1560P x 2160L x 1B)
  	SUBDATASET_6_NAME=GTIFF_DIR:6:image-file-good.tiff
  	SUBDATASET_6_DESC=Page 6 (780P x 1080L x 1B)
  	SUBDATASET_7_NAME=GTIFF_DIR:7:image-file-good.tiff
  	SUBDATASET_7_DESC=Page 7 (390P x 540L x 1B)
  	SUBDATASET_8_NAME=GTIFF_DIR:8:image-file-good.tiff
  	SUBDATASET_8_DESC=Page 8 (195P x 270L x 1B)
  	SUBDATASET_9_NAME=GTIFF_DIR:9:image-file-good.tiff
  	SUBDATASET_9_DESC=Page 9 (97P x 135L x 1B)
  	SUBDATASET_10_NAME=GTIFF_DIR:10:image-file-good.tiff
  	...
  	...

  A SLOW to read image: there are no sub-datasets, so entire image is stored as an image without overlays on higher levels:

  $ gdalinfo image-file-no-pyramids.tiff
  Driver: GTiff/GeoTIFF
  Files: image-file-no-pyramids.tiff
  Size is 31042, 22080
  Metadata:
  	TIFFTAG_RESOLUTIONUNIT=2 (pixels/inch)
  	TIFFTAG_XRESOLUTION=72
  	TIFFTAG_YRESOLUTION=72
  Image Structure Metadata:
  	COMPRESSION=LZW
  	INTERLEAVE=PIXEL
  	PREDICTOR=2
  Corner Coordinates:
  Upper Left  (     0.0,     0.0)
  Lower Left  (     0.0, 22080.0)
  Upper Right ( 31042.0,     0.0)
  Lower Right ( 31042.0, 22080.0)
  Center      ( 15521.0, 11040.0)
  Band 1 Block=2048x1 Type=Byte, ColorInterp=Red
  Band 2 Block=2048x1 Type=Byte, ColorInterp=Green
  Band 3 Block=2048x1 Type=Byte, ColorInterp=Blue

What can you do to make sure created images are having right structure

• On a high level - make sure that written TIFF image has overviews (pyramids) inside. Concrete solution depends on a library, tool or software you are using to generate TIFF images. Some approaches will require adding special set of steps to generate each level separately per channel, some other will allow to use a simplified approach (when lib will come with most reasonable strategy for pyramids). For example if you are using GDAL library to write TIFF images

• For Python make (at the end of writing of all data to the image) an additional call to BuildOverviews method:

  # Python
   
  image = gdal.Open(....)
   
  # at this step write the data into image
  gdal.SetConfigOption('COMPRESS_OVERVIEW', 'DEFLATE') # other compression options are also available
  image.BuildOverviews('NEAREST', [4, 8, 16, 32, 64, 128], gdal.TermProgress_nocb) # will take quite some time, will add all pyramids of specified scale
   
  del image  # close the image (flush)

• For C++ you can use the BuildOverviews function from GDALDataset (it is same to what Python is doing above).

  // create a dataset (image)
  GDALDataset* poDataset = ...
   
  // at this step write data into the dataset (via bands,...)
   
  // generate pyramids
  int anOverviewList[3] = { 2, 4, 8 };
  poDataset->BuildOverviews( "NEAREST", 3, anOverviewList, 0, nullptr, GDALDummyProgress, nullptr);
   
  // close the dataset
  ...

What can you do to improve existing TIF files

• What if you have 1 or 100 TIF files that were already generated, you don't want to regenerate them and be able to use them efficiently?

• Solution will be to add overlays (pyramids) to the file. This you can do in various ways, one most easiest one is to use the gdaladdo tool.

  $ gdaladdo -r average a-file-to-be-improved.tiff
  0...10...20...30...40...50...60...70...80...90...100 - done.
   
  It will update the file (recreate it) and extend with required pyramids (with zoom levels automatically calculated, see also parameter to the gdaladdo tool for more precise control).