Thanks! You've successfully subscribed to the BONEZONE®/OMTEC® Monthly eNewsletter!

Please take a moment to tell us more about yourself and help us keep unwanted emails out of your inbox.

Choose one or more mailing lists:
BONEZONE/OMTEC Monthly eNewsletter
OMTEC Conference Updates
Advertising/Sponsorship Opportunities
Exhibiting Opportunities
* Indicates a required field.

Imaging Analytics and the Changing World of Orthopaedics

Imaging analytics is the extraction of meaningful quantitative information from digital images using software-based image processing and analysis techniques (i.e., algorithms). These techniques can be part of standard software suite, or custom-tailored to a specific application using a combination of advanced image processing filters optimized for speed and precision. Independent of application, approaches to imaging analytics should be validated and automated as much as possible to eliminate observer-bias and to minimize data variability. Examples of image analysis include 2D, 3D and 4D (temporal) object recognition, anatomical feature segmentation, density measurement, morphological characterizations (including volume and shape parameters) and rate of change for these parameters over multiple time points. In essence, imaging analytics enables us to quantitatively measure and track what the human eye can only qualitatively assess.

Drawbacks of Gold Standard, Traditional Image Scoring
In traditional image scoring, images are manually reviewed one after another by a trained technician and then given a score relative to image features that are being measured. For example, a radiologist might confirm (yes or no) or score (1 through 5) for the presence or grade of cartilage damage in a MR image. In some cases, image objects (e.g, cells or lesions) are counted within a defined region-of-interest. These scores or counts, in addition to comments or dictated notes, are compiled into a final data report. Unfortunately, manually scored image data is almost always qualitative, rather than quantitative. Furthermore, this approach to image analysis is time- and labor-intensive, inconsistent, subjective and often fraught with intra-/inter-observer variability. Observer-based image analysis will, by default, introduce subjectivity into measured outcomes that can vary significantly based on the rater's experience level and training. Thus the reproducibility/precision of reported results is generally low among observers, and impaired even for a single rater depending upon his emotional/physical disposition at the time of image evaluation. Historically, a common way to combat these drawbacks is simply to increase the number of patients or specimens in a study to meet statistical significance. This, of course, significantly increases associated time and cost. 

Benefits of Image Analytics
Software utilizing custom-tailored image processing and analysis algorithms to extract a wide range of application-/study-specific measures (e.g., tissue type, implant size/shape/material, anatomical region-of-interest, tissue repair rates, object resorption rates, etc.) can offer significant improvements in terms of performance, precision, quality and quantity of output data. The goal of most customized image analysis algorithms is to mimic visual perception to delineate features in an image in a quantitative manner. In some cases, these algorithms can also detect details or changes within images that even a trained and experienced observer may not see or reliably score. As a result, these algorithms yield highly quantitative and consistent data.

Image analysis algorithms can extract data in either automated or semi-automated routines. In either case, the automation enables batch processing and analysis of imaging data with little or no user interaction or input. This significantly reduces the time and labor traditionally associated with image review and scoring. Automation also minimizes the introduction of user-bias, significantly reduces inter-/intra-observer variability and increases throughput. As a result, analyzed data will be efficiently and objectively evaluated for speedy and sound decision-making. The organization of the quantitative data output by a particular set of analysis algorithms can also be custom-tailored to facilitate statistical analysis by a wide range of statistics programs. The data can also be visualized (2D, 3D or 4D renderings/movies) to enable efficient communication of results to stakeholders (e.g., management, regulatory agencies, physicians, patients, colleagues, etc.).