- Participants should understand the mechanism and effects of digitalization and also critically analyze the advantages and problems of this global process - from artificial intelligence to fake news on the net.
7.0.1 - Digitalisation - A Pandemic Process (Pre-Recorded)
7.0.2 - Big Data in Imaging
Abstract
Introduction
Big data in imaging
Content
Big data in imaging
Ali Guermazi, MD, PhD
Boston University School of Medicine, Boston, USA
Osteoarthritis (OA) is nowadays considered as a disease that has a spectrum, from triggering event that initiates the disease process, ‘preclinical’ OA (molecular changes in the composition of bone, cartilage, and other soft tissues), clinically detectable OA (pre-radiographic and radiographic OA), and end-stage OA that requires joint replacement. Moreover, different phenotypes of OA have been recognized to account for seemingly heterogeneous pattern of disease progression in different patients. While there are more than one phenotypic classifications proposed in the literature, one such example includes 5 phenotypes: post-traumatic, metabolic, ageing, genetic, and pain. Especially for the purpose of OA clinical trials, it seems increasingly important to recognize different phenotypes of OA so that researcher can target certain subgroups of OA patients that are most suitable for the drug being tested. The most widely used imaging modality for OA assessment if radiography. Semiquantitative evaluation of OA severity can be done using Kellgren and Lawrence grading or OARSI atlas grading. Radiograph-based outcomes are still the only currently FDA-recommended imaging-based outcomes. OARSI also recommends radiographic joint space width as an option for trials of structural modifications. However, we do need to be aware of limitations of radiography. More advanced imaging of OA can be done with MRI, including conventional MRI for assessment of morphologic changes and compositional MRI for detection of ‘pre-morphologic’ physiologic/biochemical changes. Morphologic changes can be evaluated using semiquantitative or quantitative approach, and articular and periarticular structures can be assessed, such as cartilage damage, meniscal tear, synovitis and effusion, bone marrow lesions, and ligamentous damage. Examples of compositional imaging include dGEMRIC, T2 mapping, T1rho mapping, sodium imaging, and diffusion imaging. These techniques are particularly useful for early and pre-clinical stage of OA. In OA research, population-based studies help researchers analyze and understand causative factors of OA, disease mechanisms of OA, and prevalence of OA in different and specific populations. There are multitude of population-based OA studies, such as: Framingham OA study (USA), Rotterdam study (Netherlands), Model of Early Diagnosis of Knee OA (MoDEKO) and Knee OA Progression (KOAP) studies (Canada), Hallym Ageing study (Korea), Tasmania study (Australia), Beijing study (China), and Research on Osteoarthritis-Osteoporosis Against Disability (ROAD) study (Japan), to name but a few. In addition, there are non-population based epidemiological studies. Examples are Osteoarthritis Initiative, Pivotal Osteoarthritis Initiative Magnetic Resonance Imaging Analyses (POMA), Foundation for the NIH Osteoarthritis Biomarkers Consortium, Multicenter Osteoarthritis Study (MOST), Boston Osteoarthritis of the Knee Study (BOKS), and Mechanical Factors in Arthritis of the Knee (MAK). These studies focus on specific subcohorts of a given population usually those at risk to develop disease or with established OA. This allows enriching populations and decrease the required number of subjects, which are typically large for a population-based study because many subjects are ‘normal’ or not at risk of OA. With the current popularity and increasing importance of artificial intelligence (machine learning), researchers are now applying AI algorithms to existing big data of OA. Recently published papers focus on key issues such as detection of early OA, prediction of OA disease progression and total knee arthroplasty, and identification of risk factors for OA progression.