Image Analysis

Analysis of biomedical imaging data is no less a task than the image acquisition itself. Due to the complex nature of such data and variability across different subjects, information extraction has largely remained a time-consuming process. Vendors of imaging instruments and specialty software offer a variety of tools; however, a user still needs to work through a series of complicated and time sapping procedures for image analysis involving extensive manual operations. It is not just tedious, it is prone to error. The evolvement of image analysis has seriously lagged behind the rapid advancement of imaging technology.

As an oncology CRO with years of experience across a full spectrum of preclinical imaging services, knows this all too well. This is why we have invested in the development of in-house tools to increase automation of image analysis.

Automation of image analysis does not just boost our throughput; it substantially increases the accuracy of analysis since human bias is eliminated. Both are showcased in our fully automated BLI data analysis using our BLIZZARD^TM tool. Users only need to specify a folder containing raw data to be analyzed, and BLIZZARD^TM automatically identifies and analyzes it, performs whole-body signal and/or localized signal based on the tumor model, then writes results with subject information into an Excel data sheet. Analysis of hundreds of subjects only takes a few minutes. No manual creation of region of interest (ROI) over signal is needed, greatly reducing analysis time. The human side of BLIZZARD^TM offers a result inspection view for quality assurance, and a representative image feature enabling a user to quickly export results across different time-points in a study on a global scale for review and comparison purposes. Since its release, BLIZZARD^TM has made our already successful BLI imaging service even more compelling.

Fully Automated BLI Flux Analysis of Whole-body and Primary Tumor Using Blizzard^TM

Representative Image Exporting View in BLIZZARD^TM

We do not just deliver images and numbers—we offer scientific solutions using imaging to address a customer’s business needs. Development of an image analysis protocol is thus equally important as development of an imaging method; as it dictates the data deliverables for customers. With smart design, an image analysis protocol allows us to fully exploit the capability of the imaging system and in turn, our customers receive more meaningful information from their study to help them make informed decisions.

On the implementation level, we strictly follow a disciplined scientific approach, adhere to methodologies widely cited in academic journals and commonly accepted in the industry. A fully quantitative dynamically contrast enhanced (DCE) MRI analysis procedure, for example, requires fitting a pre-contrast T1 map, converting signal intensity into actual tissue concentration of contrast agent, and finally, solving a pharmacokinetic model. It generates a series of results including T1, contrast concentration, initial area under curve (iAUC), volume transfer constant (Ktrans), fractional extracellular extravascular space volume (Ve) and fractional plasma volume (Vp). These procedures and outputs have been long established in clinical studies. Our in-house developed REDCAT^TM tool is capable of performing such an analysis with just a few mouse clicks, and offers the results in both ROI-based numerical values and pixel-based parametric maps. REDCAT^TM does not just simplify our analysis; it extracts more accurate and clinically-relevant information from a classical MRI method used extensively in oncology imaging.

Tumor Permeability Analysis using REDCAT^TM for DCE MRI Analysis

Another handy tool we’ve developed, called OASIS^TM, increases efficiency of our imaging development process and enables us to focus on imaging sequence design and agent application. For example, when we apply our new tissue pH measurement method using Chemical Exchange Saturation Transfer (CEST) MRI, our OASIS^TM tool runs in parallel with the scanner, so that the effect of a parameter change can be evaluated in real-time. In vivo measurement of tissue pH using CEST MRI is very sensitive to tissue uptake of contrast agent and response to radio frequency pulses. OASIS^TM does not just report pH values; it also visualizes more perspectives of the data collected, thus offering us more insights into the complex interrelation among imaging sequence, contrast agent, and tissue physiology. Use of this tool is instrumental in our effort to maximize the capability of a novel imaging approach.

Tumor pH Measurement using OASIS^TM for CEST MRI Analysis

Development of in-house analysis solutions for biomedical imaging data is no easy task. In-depth knowledge of imaging and biology are prerequisites. Comprehensive scientific computing techniques ranging from curve fitting and equation solving, to pattern recognition and machine learning are essential. Such development can be financially unsound for many companies, not to mention the additional cost of training, maintenance, and upgrade. We have a team with years of hands-on experience in image analysis, and our tools have been utilized on a daily basis in numerous commercial studies. If you think we can help lower your image analysis costs and add value to your data sets, please do not hesitate to get in touch to discuss your image analysis needs.