Imaging Modality Review for Orthopaedic Clinical Research

“Garbage in, garbage out,” goes the saying. An upfront understanding of how imaging modality options best support crucial research data will avoid unnecessary “data garbage” as research is underway. Starting a new clinical study with the end in mind will allow you to select the right imaging modality relative to the data needed to support and defend desired efficacy claims for a given implant.

By matching imaging modality with an image acquisition protocol (IAP) tailored to both the imaging modality and the desired image measurements, clinical researchers may ensure that their study will produce the highest quality images, and thus have the best possible chances for success. If your study is producing sub-par images because you haven’t selected the right imaging modality or properly “tuned” your IAP, chances are the qualitative or quantitative data you plan to extract from those images will be equally lackluster. Any visual data generated can yield valuable marketing collateral, so by defining an appropriate IAP ahead of time, one can reduce expense and time involved in elaborate post-marketing studies, and support communication and defense of product performance claims to key stakeholders (e.g., regulatory, management, customers, patients, etc.).

In orthopaedics, appropriate imaging modality selection is particularly important given the widespread use of imaging in orthopaedic device clinical trials, and the significant impact good imaging data can have on a company’s ability to get regulatory approval and market adoption. With this in mind, we present an overview of the more common (and uncommon) imaging modalities used in orthopaedic clinical research.

Computed Tomography

Computed tomography (CT) uses many 2-dimensional (2D) x-ray projections to generate a 3-dimensional (3D) volumetric image of a given anatomical region. Collectively, these projections are called a sinogram. Fourier filtered back projection algorithms are applied to the sinogram projection data, translating it into 2D slice data which is then rendered into a 3D volume. Typical CT scanners offer out-of-plane (axial) resolution of about 600 um (thin-slice acquisition) and in-plane resolution around 300 um (sometimes as low as 100 um), depending on the limitations of radiation dosage relative to the proximity of the scan to certain highly sensitive organs (e.g., reproductive).

CT imaging offers a 3D view of the anatomical region-of-interest (ROI) without the large amount of distortion present with Magnetic Resonance imaging (MRI) acquisitions and without 2D areal averaging characteristic of x-ray imaging. Spatial relationships between features are not subject to interpretation, as in x-ray images. As such, quantitative measures are multi-dimensional and enable morphological assessment, providing a clinician with more information from which to make a diagnosis. An additional advantage of 3D CT is the reduction in variability associated with inconsistent patient positioning across multiple time points—a major drawback with x-ray imaging.

The time it takes to acquire a 3D CT volume is measured in seconds rather than minutes, with axial resolutions in the sub-millimeter range rather than 3-4 mm for a standard MRI sequence. Thus, patient motion artifact is rarely an issue since acquisitions can be gained within a single breath-hold. With speed and consistency in acquisition, comparison of multiple time points for a single patient (longitudinal assessment) can be made using spatial volumetric registration and image-based segmentation techniques (e.g., bone growth, implant resorption, etc.). This enables accurate quantitative evaluation of treatment/device response over time. Of course, CT is not without its limitations. X-ray metal artifact (stainless steel, cobalt chromium in particular) can make it very difficult to visualize anatomical features near a component (e.g., bone-implant interface) due to reconstruction artifacts, beam hardening, etc. Additionally, high ionizing radiation dosages can preclude frequent imaging time points near or around sensitive organs. It’s also important to note that softer tissues, due to their inability to attenuate x-rays, exhibit less contrast and more noise in a CT scan.

Since CT measurements are most often derived from differences in tissue density, this imaging modality is best for detecting and visualizing denser tissue such as cortical and trabecular bone. This makes CT ideal for orthopaedic studies seeking to quantify new bone growth on or around an implant or fracture site. However, with the help of an injected contrast agent, CT can also be used to visualize soft tissues (e.g., vasculature). Furthermore, due to the density based attenuation of x-rays, CT systems are generally “calibrated” to a known scale (Hounsfield Units) for comparison of images and resultant segmentation across time-points, patients, scanner vendors and study sites.

Radiography (X-ray)

X-ray imaging is a 2D single-planar imaging modality most frequently used for visualizing skeletal pathologies and gross soft tissue pathologies. Like CT, its use is accompanied by ionizing radiation, however at a much lower dosage. Unlike CT, data is “integrated” through the patient for a given view (usually A-P or anterior-posterior or lateral). Rather than visualizing a plane or slice of a given thickness, x-ray provides an intensity projection without depth discrimination.

Each x-ray technique has its pros and cons. For example, computed radiography (CR) is an improvement upon the old manual chemistry-based approach to acquiring and developing an x-ray image via film. By combining digital acquisition and a more automated approach to chemical-based film development (e.g., CR read), the end-user is no longer directly exposed to any chemical and the final image product is in digital format. Digital radiography (DR), like CR, is a form of digital x-ray, operating much like a hand held digital camera directly detecting x-rays without the need for chemical conversion. Dual-energy X-ray absorptiometry (DEXA) uses two x-ray beams to measure bone mineral density, and is most often used in diagnosing osteoporosis. Fluoroscopy, a low-dose x-ray, is used commonly for real-time imaging of various internal anatomical structures (e.g., during surgery), implant insertion, etc.

The biggest advantages of x-ray are time and cost. The scans are relatively cheap and take only seconds to acquire. This modality also offers higher spatial resolution than CT or MR. There are, however, several negatives in addition to the presence of ionizing radiation. First, since you are looking through a patient when viewing an x-ray, it’s difficult to distinguish spatial relationships. Second, the inherent variability in patient positioning in a clinical study utilizing multiple x-ray time points can make it difficult for a human image reader to extract reliable quantitative measurements over time. Third, due to the lack spatial orientation/depth, magnification markers must be used to standardize image measurements since patient proximity to the detector will affect the resolution of the image. As a result of the complexity associated with patient positioning, the importance of a well-written and validated x-ray IAP cannot be understated. This is particularly true in a study involving multiple study sites where “routine” techniques employed by imaging staff may vary significantly.

X-ray is best used when detailed 3D information is unnecessary and cost and time are big factors. X-ray, even more so than CT, is widely used in orthopaedic clinical research. In particular, x-ray is most commonly used in evaluating component/bone interfaces for osseointegration or loosening, fracture healing or bone loss/remodeling due to osteoarthritis, injury or cancer metastasis. When patient positioning is standardized and magnification markers are appropriately used, quantitative measurements for component evaluations or various pathologies can be made fairly reliably.

Magnetic Resonance Imaging

Relying on the variation in relaxation times of protons in water within different tissue types, MRI is particularly adept at detecting and differentiating between soft tissue structures. Similar to CT, MRI acquires imaging information in 2D slices which can be used to render a crude 3D image volume. Specific 3D sequences can be used generate more accurate volumetric renderings. Unlike CT, MRI does not require ionizing x-ray radiation to detect and visualize differences in density to resolve tissue structures. MR uses a powerful magnet to align atomic nuclei in the body and subsequently measures their relaxation following various pulse sequences via specialized RF (radio frequency) coils. As a result, there are numerous image acquisition modes called “sequences” which allow an MR scanner operator to custom-tailor the IAP for distinguishing one tissue type over another. Acquiring a good MR image is closely tied to the scanner’s magnet field strength. The relationship between image quality and magnet field strength is simple: the higher the field strength (e.g., 1, 1.5 or 3 Tesla), the lower the scan time and better the signal-to-noise ratio (SNR) will be. (Certain MRI artifacts, however, can accompany higher field strengths.)

MRI also offers increased contrast resolution for distinguishing different adjacent soft tissue types with similar densities. By fine-tuning and combing various MR image acquisition sequences, a skilled operator has a vast array of options from which to build a highly customized IAP to imaging various implant, tissue types and tissue structures. Thanks to the lack of radiation exposure, MRI lends itself to faster institutional review board approvals and works well in studies requiring patients to undergo multiple imaging time points within a short period of time. In contrast, MRI scans take a long time relative to CT (20-45 minutes), the scanners and building infrastructure are relatively expensive to purchase and maintain, offer less spatial resolution than CT and have non-standardized scanner inputs and outputs. Unlike CT, which uses the Hounsfield scale as a way to calibrate/standardize acquisitions, MR sequences and resulting image outputs can vary across scanner brands and often within a single scanner. This can have disastrous implications for a clinical study that is not using an MR IAP that has been tested, validated and incorporates the use of an anatomical ROI-specific phantom which can be used to normalize data across scanners, patients and study sites. Finally, successful operation of an MR scanner requires access to both a highly trained operator and an MR physicist. Without these individuals, resulting image quality variability can make quantitative image analysis difficult, if not impossible.

In orthopaedic clinical research, MRI is most often used to diagnose sports-related injuries and track pathology progression relative to an intervention (e.g., device or drug). Multi-echo MR sequences can detect and visualize cartilage collagen organization and stratification (T2 mapping). Since cartilage layering is an indicator of tissue health, evaluating T2 values in cartilage zones can be critical for assessing the efficacy of various repair strategies. This approach can also be used to detect cartilage degeneration over time, or to evaluate implant-mediated rotator cuff or ACL tissue repair.

A drawback to one method of cartilage repair, the injection of hyaluronic acid (HA), is confirming how well the material fills the entire cartilage defect volume. In Exhibit 1, two MR images form a test suite of different types of HA formulations that were applied to tissue defects. The goal of this analysis was to evaluate and confirm HA formulation effectiveness for void volume filling. A known quantity of each formulation was used for each defect, and post-operative imaging analysis was applied to segment the HA “plug” from the surrounding tissue and quantitatively measure each HA plug volume. The red, green and yellow colors in the right side of Exhibit 2 were applied manually to indicate different HA formulations. In this case, the combination of MRI and custom-tailored imaging analytics enabled researchers to confirm which formulation of HA most effectively filled the cartilage defect.

Positron Emission Tomography

Positron Emission Tomography (PET) is a nuclear medicine technique that yields a 3D image of a functional physiological progress within the body. This is accomplished through the detection of gamma ray pairs emitted by a radionucleotide (tracer) which is attached to a bioactive molecule and then introduced into the body. Localized concentration of the tracer is visualized in 3D relative at a single time point or longitudinally at predefined intervals. Interpretation of PET imaging data can be difficult because, unlike the other imaging modalities mentioned above, it is comprises information relating to a physiological process, rather than well-defined tissue structures or densities.

PET is designed to detect and visualize metabolic processes. Since the quantity of tracer administered is known, kinetic calculations can be made using a PET scan to quantitatively define compartment clearance and metabolic activity throughout the body. PET’s serious shortcoming is the lack of tissue density and structural information in a scan. Fortunately, PET is often combined or fused with CT so that both functional and structural imaging data can be acquired. PET is also ideal for tracking slowly-moving substances through the body in 3 dimensions. To accomplish this with a CT would require massive amounts of radiation exposure. Low image resolution is another downside to PET. That said, the information provided by the activity of the tracer often outweighs the lack of image resolution or anatomic localization. Finally, PET tracers have very short half-lives, requiring that they be made on site (often using a cyclotron). As a result, a hospital can incur enormous costs to build and maintain the supporting infrastructure, making it difficult for smaller hospital systems to offer this imaging modality.

Traditionally, PET is used to evaluate systemic or localized drug metabolism, to assess brain activity or to detect the locations of various tumors by detecting and analyzing local metabolic “hot spots” of an ingested tracer-linked glucose analog. In orthopaedic clinical research, PET is most often used in a technique called bone scintigraphy. This is used to detect regions of fast bone remodeling which occurs due to metastasized cancers spreading to the musculoskeletal system. Essentially, localized hot spots of high NaF presence can be correlated with active bone remodeling.

Ultrasonography

Ultrasound (US) imaging uses piezoelectric transducers to send sound waves into an anatomical region. Relying on the fact that various tissue types and fluid flow reflect these waves differently, an image can be generated depicting tissue anatomy and blood flow. Visually resolving tissue structures involves a balance between frequency, image resolution and imaging depth. A high frequency transducer will yield better image resolution (down to 30um), but will sacrifice penetration depth and is best suited for eye anatomy. Alternatively, lowering the probe frequency will enable the user to see deeper into the tissue, but with less resolution.

Types of ultrasound include traditional 2D, 3D, ultrasonic microscopy (material science), trans-ultrasound and Doppler (measuring flow without a contrast agent). This 2D imaging technique is easy to use, involves zero ionizing radiation, is non-invasive and relatively cost-effective. Most ultrasound scanners are portable, so a clinic can cover many floors with a single scanner. However, ultrasound images tend to have low SNR due to the larger number of tissue interfaces that the transducer must penetrate. As a result, an ultrasound image can be relatively difficult to use for quantitative measurements and interpretation of pathologies. Another drawback is that acquiring an ultrasound image or video is often more art than science. At lot of the final image quality relies on both patient positioning and the technique used by the operator. This means inter- and even intra-operator image acquisition variability can be high, and it’s important that the operator be familiar with the specific anatomy of interest. Thus, having a detailed well-written IAP and standardized image acquisition training across all study sites can ensure reliability of outcomes of a clinical study using this modality.

In orthopaedic clinical research, ultrasound is not widely used. That said, there have been applications combining traditional and Doppler ultrasound to assess musculoskeletal wound healing for sports medicine injuries (e.g., rotator cuff repair) relative to implants and devices which facilitate the healing process.

Conclusion

The various factors that must be considered when planning an orthopaedic clinical study can be overwhelming. Given the diverse nature of the imaging modalities at one’s disposal, it’s easy to see how careful imaging modality consideration can ensure reliability and quality of outcome parameters in a clinical study. Additionally, mixing and matching certain modalities can increase the number of output parameters available by improving ability to discriminate structures of interest, enable implant performance measurements not originally thought possible, and reduce a study’s timeline by leveraging the additional resolution and/or data provided by the combined imaging approach. With all of this in mind, it’s important to know that as a clinical research you’re not alone. If you find yourself unsure as to which imaging modality is right for your study, or how to best optimize an IAP to ensure you acquire the best possible image data, be sure to either work through these critical aspects of your study with your internal imaging expert or an external imaging contract research organization that you know and trust.

In his role as a Vice President at ImageIQ, Brett Hoover relies on ten years of experience as a biotechnology and life science researcher. product manager and commercialization specialist. Contact Brett at

Amit Vasanji, Ph.D., in his role as Chief Technology Officer for ImageIQ, relies on more than ten years of experience with basic and clinical research image acquisition, processing, analysis visualization and software programming. Save Supplier