Review of literature

In Neuro Oncol (2015) 17 (6): 784-800. doi:10.1093/neuonc/nou322

Apart from the described biological ambiguities of ADC changes, methodical concerns such as standardization of measurement parameters, artifact minimization, and improvement of spatial resolution still remain unresolved.

Algorithm in oncology

Quantitative Imaging Biomarkers Alliance published a set of profiles intended to serve as guidelines in protocol definition. They provide general imaging guidelines UPICT and derived profiles for FDG-PET/CT as an Imaging Biomarker Measuring Response to Cancer Therapy (Publicly Reviewed Version ,QIBAPET) and CT Tumor Volume Change v2.2 (Publicly Reviewed Version, QIBACT).

The format allows for assigning relative importance of individual steps by segmenting characteristic parameters of each step to ideal, target and acceptable regions. I like such segmentation because it provides inclusion and selection criteria for samples in a study, and directs the compliance efforts in patients and study personel.

Standardization of PET/CT harmonization

Groups around the world divested signifcant amount of effort into search for reconstruction/imaging parameter set pairs that would match imaging performance of different scanners. Different harmonization procedures exist, for example efforts by Ron Boellaard and co-workers in the Neatherlands and simmilar efforts by Lalitha Shankar in the USA. Both procedures share a number of common characteristics: - usage of phantoms - evaluation of numerous reconstruction algorithms and settings of those algorithms - centralized qualification authority

Impact of harmonization

While the harmonization procedure is relativelly straightforward to implement, impact of harmonization is somewhat hard to measure. In clinical setting, harmonization of scanners is mixed by inter-patient and intra-patient variation, often obscuring source of variation in performed studies. Again Ron Boellaard gives a nice overview of pitfalls and successes of harmonization in oncology.

Potential tasks related to standardization of FDG PET CT imaging in neurology

The specific aim is to define a work plan for the following topics:

Harmonisation of brain FDG imaging procedures and PET/CT performance.

Tasks:

• Inventory of current FDG brain PET/CT imaging procedures to determine variability in execution of brain pet studies

• Sensitivity analysis – use selected dataset and generate/re-reconstruct data with different image characteristics (resolution, noise, PSF) and evaluate

  • impact on quantitative accuracy and precision of tracer uptake and
  • impact on multivariate data analysis (of which the GLIMPS procedure is only one).

• Define level of required harmonisation – how to assess impact of image quality variability on disease probability/classification performance?

• Harmonisation strategies – imaging procedure
• Harmonisation strategies – PET/CT system performance/image quality
• Setting up a multicentre quality control programme- feasibility testing-recommendations-QC programme implementation at EU or global level.

• Guideline writing group and assigning tasks:

  • review of existing guidelines for FDG brain imaging
  • gap analysis – what is missing or suboptimal in current guidelines?
  • sensitivity analysis – see above, this determines how strict the quality of data needs and can be.
  • update/new guideline?.

• White paper - workshop summary?

• Exchange (healthy) control images (or from oncology scans, no brain mets).

Clinical perspective

Tasks:

• How to collect clinical imaging and non-imaging data, patient demographics
• Testing the influence of the harmonization efforts on the various multivariate analysis methods.
• Validation of disease pattern etc/comparability of disease patterns when derived from data from different institutes? Developing new/other disease patterns?
• Test-retest performance

IT requirements/software developments/database enhancement

Tasks:

• Aim is to define requirements for a fully automated IT infrastructure to support multicentre studies in a sustainable manner.
• What do we have now?
• Can and should it be modified
• Suggestions for workflow enhancement
• Automated QC on incoming data/images