ECR 2014 - SF7b - Imaging biomarkers in cancer drug development

An area where medical imaging informatics could play a major role is imaging biomarkers. The involvement of imaging informatics in this area is not only in the image processing, but also in validation, evaluation and collection of imaging data and related quantitative data using imaging biobanks.

Imaging biomarkers are currently a hot topic, in imaging this relates to:

1. size measurement based imaging biomarkers (RECIST, mRECIST and EASL)

2. functional and molecular biomarkers (perfusion, vascular function, cellularity, cell-death, metabolism, proliferation, and target specific tracers)

there are three biomarker modalities:

1. biospecimen removed from a patient

2. biosignal measured using in vivo device

3. patient reported outcomes

Important is that biomarkers require extensive validation and standardization of acquisition, evaluation, and measurement.

The validation of imaging biomarkers however, is a very hard job. 

Imaging biomarkers should be used to better identify the patients most likely to benefit and to obtain better hypothosis testing.

problems in using imaging biomarkers include the variety in accquisition devices and the fact they are not designed for biomarker measurements.

RSNA 2013 - RC725 - quantitative imaging: informatics

Radiology currently is primarily concerned with pattern recognition with some linear measurments. Therefore, the demand for quantitative tools is little in radiology. Residents should be trained in using 3d software and planning during their regular rotations. To determine the correct measurements validation datasets are needed. What is required:

1. Better tools to create measurements are required with integration into the PACS to integrate into the regular workflow.

2. Better tools for representing the results (DICOM SR and AIM are available but adoption by vendors is slow).

3. Education.

4. Validation datasets. How to use existing tools and how not to mis-use them.

The main immediate informatics challenge is that the workflow for 3D postprocessing is poor.

QIBA is trying to streamline the process of obtaining quantitative imaging biomarkers. Using standards is a prerequisite in this effort.

RSNA 2013 - RC326 - Quantitative Imaging: A Revolution in Evolution (In Association with the Society for Imaging Informatics in Medicine)

Vendors provide easy tools to perform quantitative imaging but the question is how reliable and repeatable this quantification is. 

Many clinical examples can already be listed where quantitative imaging is used in clinical practice. Such as carotid stenosis, coronary artery stenosis, calcium scoring, pulmonary nodules, renal donor evaluation, liver and tumor volumetry, brain perfusion, emphysema quantification. 

Each of these show good results in literature and could be applied in clinical practice.

However, the question is whether the numbers we get out of the software are usefull and what is reality. Different vendors provide different results in the same patient and even within one software system measurements are influenced by postprocessing choices but also by decisions during the acquisiiton. 

One of the things that can be done to get proper quantification we should provide reference datasets.

QIBA is ran by a group of stakeholders to improve quantitative imaging. They define profiles to get precise, repeatable measurements. QIBA has setup a imaging data warehouse (QIDW) including standard datasets that can be used to validate quantitative imaging algorithm.

The QIDW is free, open source, modular software based on MIDAS.

In conclusion tagging of the image data in radiology is essential to allow computers to work with the information. The quantification is part of this tagging. Developments like AIM are trying to cover this and allow export in XML or DICOM SR, however current PACSs and EMRs do not yet support these kind of measurement. When the storage and data mining of all the information available in the images becomes possible it will provide the key information to get to personalized diagnosis and treatment.