Each clinical study that relies on imaging is going to have many measurements to sift through. The larger the study – the more study sites involved – the more you risk acquiring a potentially overwhelming number of measurements, particularly if the analysis team is not prepared. Key to that preparation is proper and comprehensive validation of the image analysis software being used in your study. While the software you’re using may have been validated by the vendor or developer, that doesn’t necessarily mean it’s the most effective tool for your study. Validating the software yourself is a necessary step for two key reasons – 1) to ensure that the software is right for your study and will produce accurate data, and 2) to avoid regulatory and compliance hang-ups.
Here are some steps you can take to validate your image analysis software:
Step 1: For standard image measurements, such as distances and angles, it is important to ensure all expert image readers (e.g., radiologists, dermatologists, etc.) on the team are using the same version of the same software
Step 2: Make sure all of the image readers are performing their measurements in the same way. A well-written image evaluation protocol (IEP) is a great start.
Step 3: When using a non-standard technique, such as a new piece of software developed specifically for your study, request that your radiology team – or the team doing the imaging analysis – provide you with any and all peer-reviewed references that support your non-standard technique. Noting these references in your trial report will help to defend their use to the FDA and other regulatory bodies.
Step 4: Keep a log documenting all imaging equipment involved in your study. This includes the make, model and software version of each scanner at each study site.
Step 5: Collect the training records and credentials of all image readers in your study. Any and all information that reflects and represents their expertise with image evaluation protocol (IEP), the software and other tools used in the trial, as well as their expertise with the anatomy and disease states relevant to your study, will be useful additions to your trial report. This documentation can be particularly important if using a non-standard image analysis technique such as in Step 3 above.
Step 6: Take the time to validate empirically the image analysis software being used in your specific trial. The best way to accomplish this is to have an imaging technician scan an imaging phantom. Be sure to choose a phantom that closely matches the needs of your study’s indication and anatomical region of interest in terms of size, shape, and density. Have the image analysis team perform image measurements on the phantom-generated images with the image analysis software, so that you can prove that the measurement data generated by the software-based technique match the actual physical characteristics of the imaging phantom.
By following these steps, you can have additional confidence that your image analysis software will deliver the results you’re looking for and will satisfy the regulatory requirements of your clinical trial management efforts. For more on validating your image analysis software, check out our Imaging Bytes video series.
OUR NEXT BLOG: IMAGE ANALYSIS SOFTWARE, PROFESSIONAL READER, OR BOTH?