Warning! OUTDATED BROWSER DETECTED!   Please update your browser immediately for a better experience on this website. Learn More
  • Look Ahead
    The Future of Musculoskeletal Imaging

    Experts discuss the demands, challenges and opportunities of the subspecialty. By thomas M. link, md, phd, And Daniel I. Rosenthal, MD

    June 1, 2018

    While radiology has become progressively specialized over many decades, some areas of special expertise have developed earlier than others. Chest disease became an early focus of many radiologists because of the inherent ability of radiographs to demonstrate the structures of the lungs and heart and because of the importance of chest diseases such as tuberculosis. Pioneers including Felix Fleishner, MD, Benjamin Felson, MD, and Aubrey Hampton, MD, quickly developed a specialized body of knowledge. Gastrointestinal imaging as a field of study can be dated to the introduction of barium in 1908. Nuclear medicine came later (after World War II), but has always been a distinct subspecialty because of the specialized technology involved. Pediatric radiology was pioneered by innovators including John Caffey, MD, and was the first subspecialty to be recognized with its own certification program in 1994.

    In this context, musculoskeletal (MSK) radiology is a young subspecialty. Although it might be said that the field of MSK radiology began with the first radiograph ever produced (the hand of Wilhelm Roentgen’s wife, Bertha, taken on Dec. 22, 1895), the potential corpus of knowledge was limited. Bone tumors and metabolic bone disease could be evaluated with some sophistication, but they were rare. Soft tissue structures were all but invisible. For decades, imaging of bones remained the purview of general radiologists. Everything changed with the explosion of technology in the 1970s and 1980s that brought us CT, MRI, quantitative osteoporosis imaging and, eventually, high-resolution ultrasound.

    Changing Patient Populations Create Challenges

    History also played a part in spurring the growth of the specialty. During the “baby boom" years in the U.S. (generally defined as from 1945 to 1960-1964) more than 65 million babies came into the world. Baby boomers are now between 56 and 73 years of age, and either already have, or will shortly develop problems of the locomotor system. Every organ system is subject to its characteristic afflictions, and patients may experience one or more of them. However, virtually everyone who lives long enough will have to contend with problems that fall within the domain of orthopedics: osteoporosis, fractures, arthritis of one or more joints, back pain and spinal stenosis. Ironically, public health efforts to combat cardiovascular disease have encouraged participation of adults and even older adults in physical activities that can exacerbate these conditions. Therefore the recent past has seen a huge growth in the medical specialties that diagnose and treat these problems. The immediate future will probably generate even higher demands.

    Most of the patients requiring these services will be elderly, many will be retired, some will be on Medicare. It is generally recognized that expenditures on healthcare have risen disproportionately and that this trend cannot be permitted to continue. Orthopedic diagnosis and treatment is particularly vulnerable to cost-cutting methods as almost all can be viewed as elective. What form will it take?

    Future Requirements for MSK Imaging

    Recently, a great deal of attention has been given to eliminating examinations that do not meet some criteria for “appropriateness.” “Decision support” for this purpose is now federally mandated. This approach has yielded some benefits, but it is unlikely that they will be sufficient to satisfy the demands of payers. As imaging has improved, it has moved ever earlier in the care cycle, often becoming a precursor (or even substitute) for the history and the physical exam. Therefore, the concept of appropriateness, which depends upon prior knowledge of the patient, becomes less meaningful. The easiest method to save money is to pay less, and therefore further payment reductions are likely.

    In order to survive in this environment, radiology departments will need to devise faster, less expensive methods to deliver their services. Every aspect of the imaging encounter will have to be re-imagined for greater efficiency from ordering to examination selection (protocoling), performance and interpretation.

    Simplification and standardization will be key goals in cost reduction. Instead of envisioning a world where each examination is tailored to the specific patient’s needs, imaging will increasingly be viewed as either screening (similar to screening mammography), or focused problem solving. As in the example of mammography, the concept of an appropriate number of “call backs” will develop for much of imaging. Too many callbacks will call into question the expertise or confidence of the interpreting radiologist; but too few will indicate the possibility of over-imaging. These trends will probably work well for the common MSK concerns (knee, shoulder and spine), but will be problematic for the less commonly imaged body parts (hand, foot and elbow).

    Fig 1 Knee

    Fig 2 Knee

    Figure 1: Two knee radiographs (1a and 1b) showing degenerative changes (left) with saliency heat maps (right) identifying pixels that the Artificial Intelligence uses more in assigning the diagnosis of osteoarthritis. Highlighted pixels correspond to osteophytes and joint space narrowing. Images courtesy of Valentina Pedoia, PhD, Department of Radiology and Biomedical Imaging, UCSF.

    For example, lumbar spine MRI is known to have little clinical value in most patients with back pain. Is the time that the radiologist spends to protocol it justified? We know that sometimes the use of contrast is helpful. However, if contrast were never used as part of an initial study, but reserved for selected “call backs” would it be better or worse for most patients? What is the smallest number of images and the fastest scan that can be used to identify the important findings? Careful cost-benefit analysis can answer these questions and will be necessary for the majority of our most common high-tech imaging.

    New Developments in MSK Imaging Research

    Another global trend is related to the economic and scientific advances in Asia. The economic success of this region has permitted many more people to participate in medical research. As a result, there has been a rapid rise in contributions to the medical literature from Asia.

    The world of radiology is currently abuzz with discussions of what machine learning (ML) and artificial intelligence (AI) will do to our specialty. Musculoskeletal imaging appears to be a prime candidate for the introduction of computer-assisted image interpretation because of its very large number of measurements and grading systems. Radiologists know that their colleagues in orthopedics, podiatry and hand surgery make use of these measurements but frequently omit them from their reports. There are two reasons for this: they are tedious and time-consuming, and individual variation in measurements could lead to a conflict between the radiologist and the clinician.


    Figure 2: Automatic segmentation of cartilage using statistical parametric mapping of cartilage T2 with voxel-based relaxometry. Images courtesy of Valentina Pedoia, PhD, Department of Radiology and Biomedical Imaging, UCSF.

    Computer-assisted interpretation has the potential to address both. In the near future, imaging of every body part could come with a form or table indicating which measurements are normal and abnormal. For each diagnostic entity, there will be assistance to assign grades or stages when they exist. We do not think that full computer interpretation will occur in the near or intermediate term. However, it is very likely that the rapid progress in neural networks will enable computers to effectively recognize (and even grade) certain specific conditions (probably including degenerative arthritis and “normal”).

    For example, Figure 1 demonstrates how AI (convolutional neural network) — without any a priori knowledge — is able to look at the training data and identify the correct features, in this case joint space narrowing and the presence of osteophytes, qaz `1zas distinctive osteoarthritis features. Figure 2 shows an example of cartilage automated segmentation with a novel voxel-based relaxometry approach to analyze local patterns in cartilage composition using T2 and T1 rho imaging biomarkers and statistical parametric mapping. As a result, the MSK radiologist of the future will have many more tools at his or her disposal and the job of image interpretation will become more sophisticated and demanding.

    Interventional MSK Radiology is Also Evolving

    Interventional procedures are also in a state of rapid change. Many MSK radiologists perform an interesting mix of advanced diagnostic imaging and interventions including arthrography, biopsy, tumor ablation, vertebroplasty and pain management. There has been enormous progress recently, with the introduction of many percutaneous therapeutics including tumor ablation, percutaneous fracture fixation and percutaneous soft tissue release. The introduction of injectable active biological agents such as osteoinductive agents promises to greatly enhance the effectiveness of such procedures. Steady innovation in materials research also promises to enhance the effectiveness of percutaneous treatment.

    There are, however, several major threats, both from outside forces and from within radiology. Clinicians are not unaware of these advances, and there is a strong tendency for each clinical specialty to develop minimally invasive procedures and proceduralists who will compete with interventional radiologists. There is also a distinct increase in the amount and specificity of documentation that is required if radiologists are to play the role of direct caregiver. This will undoubtedly take the form of procedure-specific credentials and documentation of quality and safety. In short, these changes blur the boundary lines between departments: clinicians are becoming more like radiologists and radiologists are becoming more like clinicians.

    Finally, there is also a threat from within radiology. MSK radiology is often a mix of diagnostic and interventional procedures. The recent bifurcation of our specialty into diagnostic and interventional training tracks will create a new class of “interventionalists” who are specifically sanctioned to do these procedures. The current concept of what intervention means does not always extend to include MSK procedures, and individuals selecting this path will probably be more focused upon higher volume chest and abdominal procedures. MSK radiology practices will be faced with choosing individuals who have not expected to do any interventions (diagnostic trainees) or those who expect to exclusively do interventions of a different sort. It will be very difficult to maintain the attractive current blend of sophisticated diagnostic imaging and specialized interventions in MSK radiology.

    Expanding Opportunities in MSK Imaging

    We strongly believe that with new developments in MSK imaging we will see faster and more precise imaging with an expanding number of technologies and interventions ultimately improving patient care. However, this requires strategic planning and a combined effort of MSK radiologists to remain leaders in their field, educating their trainees and promoting research.

    Thomas M. Link, MD, PhD, is chief of musculoskeletal imaging and clinical director of the Musculoskeletal and Quantitative Imaging Research Group in the Department of Radiology and Biomedical Imaging at the University of California, San Francisco. Dr. Link’s research interests include imaging of osteoporosis, osteoarthritis and cartilage. He serves on the editorial boards of scientific journals including Radiology and is a member of the RSNA Committee on Scientific Affairs and Public Information Advisors Network.

    Daniel I. Rosenthal, MD, professor of radiology at Harvard Medical School and vice-chairman of radiology at Massachusetts General Hospital, Boston. His research interests include diseases of the skeleton and locomotor systems and he has aided in the development of new therapeutic treatments for osteoid osteomas as well as Gaucher’s disease.