Case Studies

Explore some of the ways in which QIPCM can help facilitate various types of clinical trials: logistically, easily, accurately, and securely. We are able to give a number of solutions to wide range of users. From Education, to Securely Linking Sites, Artificial Intelligence, Theranostics or Central Images and Radiotherapy Plans.

Case Study #1

Central Image and Radiotherapy Plan – PATRON

PI: Dr. Cynthia Menard (CHUM)

Study Description & Challenges

Images and Radiotherapy (RT) plans from 776 patients
1388 exams (600 RT plans/ 388 PSMA-PET/400 diagnostic MR)
Data de-identified and transferred from 15 clinical sites across Canada
PET-CT scans reviewed by an expert central radiology reviewer
RT plans to be reviewed by an expert central radiation oncology reviewer

QIPCM Solution

QIPCM sets up the CTP DICOM data anonymizer and transfer pipeline at 15 sites
Automated e-mails sent out to the sites confirm receipt of data
Images and RT Plans are received and undergo auto-QA. 10% of plans are manually QA’d from each site
Notification to Radiologist (PET-CT scans)and Radiation Oncologist (RT plans) that a new case is ready for review
Central Reviewers log into their QIPCM Virtual Desktop and perform the review

Case Study #2

Theranostics – OZM-067 NET-PRRT Trial & Lutathera

PI: Dr. Rebecca Wong (UofT)

Study Description & Challenges

190 Patients received Lu-DOTATATE therapy at 4 centres across Ontario
These patients received Ga-68 PET-CT at UHN in order to determine eligibility
The eligible patients then received 4 cycles of Lu-DOTATATE Therapy based on individualized dosimetry determined by SPECT-CT imaging.

QIPCM Solution

Installed trial specific QIPCM CTP Client at each institution
Performed SPECT-CT Sensitivity QA at each site to ensure accurate dosimetry
Received SPECT-CT images from all sites, performed QA and archived to QIPCM PACS
Worked with Radiation Medicine Program (RMP) Physics program to design an individualized dosimetry extension for MIM
QIPCM created workflow provides automated image QA
Tracking of dose to Organs at risk and tumours over the course of treatment
Extended for use in Standard of Care Lutathera Dosimetry at UHN

Case Study #3

Linking to Multiple Sites & Systems – OCTANE

PI: Dr. Phillipe Bedard (OICR)

Study Description & Challenges

The Octane trial led by Dr Phil Bedard shows how QIPCM has worked with other teams to link our imaging database to other available clinical data. Both the Octane 1 and Octane 2 trials were collecting imaging data from multiple sites and centralizing it at QIPCM.

However, the primary difference with the Octane trials is that they were also collecting a large amount of genomic and tissue sample data, which would be made available via another web based solution called cBioportal.

The hope was that we could work with the cBioportal team to also include the imaging data via this platform.

Octane 1.0

# of patients: 1000-2000/yr annually, up to 10000 pts by year 10
# of sites: 7 across Ontario
Currently archived (UHN Only) : ~200 pts / 4843 studies (CT/MR)

Octane 2.0

# of patients: up to 200 pts/yr, up to 460 pts by year 5
# of sites: 7 across Ontario
Currently archived: UHN = 38 pts, LHSC = 6 pts, 219 studies altogether (CT/MR/PT)

QIPCM Solution

The CTP anonymizer was used to anonymize and centralize imaging data which underwent automated and manual QA using the QIPCM imaging toolbox
We then worked with the cBioportal team to create a link between their platform and our viewer to allow study researchers to view imaging data alongside other clinical data
Additionally, as part of this trial there was a plan to extract radiomic features for these imaging sets, and that data would also be made available in cBioportal

Case Study #4

Using QIPCM for Education – PSMA

PI: Dr. Glenn Bauman (LHSC) & Dr. Katherine Zukotynski (McMaster)

Study Description & Challenges

Dr. Zukotynski came to us looking for a way to create and host a database of Prostate-Specific Membrane Antigen (PSMA) PET images complete with annotations and case reports
She wanted this information to be made easily available via a web viewer to other radiologists and radiation oncologists to train them on how to better read and understand novel PSMA PET images
The other requirement was that this curated and annotated database could be utilized to train machine learning algorithms to better detect PSMA avid lesions

QIPCM Solution

QIPCM worked with London Health Sciences Centre (LHSC), Dr. Zukotynski, and UHN to ensure proper data sharing agreements were in place
The imaging data was collected and curated using both automated and manual quality controls. As we provided a virtual machine for an expert nuclear medicine radiologist to use to annotate and segment the imaging as well as create case reports.
Expert reviewers worked to annotate image sets from a QIPCM Virtual Desktop. You can see an example of one of these cases with the segmentation as performed by the radiologist on the left as well as the case report on the right.
QIPCM worked to generate reports in DICOM SR format which can be visualized alongside the annotations and images in the QIPCM Web Viewer which would make training easier for the radiologists
Weare currently in the process of working with Dr. Zukotynski to create a set of YouTube videos to help train learners to use the QIPCM viewer to navigate the cases and more accurately review PSMA PET cases
Discussions also continue with other collaborators looking at ways to expand this annotated database. Conversations have been initiated with AI scientists on the best ways to use this data to generate and train machine learning algorithms

Case Study #5

Providing AI Sandbox for Industry Partners

Study Description & Challenges

Industry partners would like to build and train Artificial Intelligence & Machine Learning algorithms on UHN’s valuable imaging and clinical data
However, Clinical and Imaging Data cannot leave UHN without being fully de-identified and questions of data ownership make this very challenging

QIPCM Solution

QIPCM has worked to anonymize and collect the data from UHN’s Clinical Picture Archiving and Communication System (PACS)
We work to provide it to the researchers on the QIPCM framework so that industry partners can leverage the anonymized data without the data ever leaving UHN utilizing the QIPCM AI Sandbox