OA diagnosis is usually based on patient symptoms and physical examination. Common cartilage and meniscus related symptoms include pain, stiffness, effusion, catching or locking sensation, and clicking on moving the knee. To ensure the diagnosis, native X-ray image is usually taken. The severity of OA is usually assessed from the a native X-ray images by Kellgren-Lawrence grading sys-tem [118]. However, as the synovial fluid and soft tissue structures within the knee joint attenuate the X-rays almost equally, cartilage and meniscus can not be differentiated on plain radiographs. Thus, the Kellgren-Lawrence grading system is based on assessing joint space narrowing and the structural changes in bone, features that are related to moderate- to late-stage OA changes. Furthermore, cartilage and meniscus injuries can also be asymptomatic, which again delays the diagnosis of these injuries [108, 119, 120].
MRI enables detection of cartilage and meniscal lesions and evaluation of cartilage thickness [105, 121, 122]. It provides a good soft tissue contrast, and thus is the preferred imaging modality for evaluating cartilage and meniscus [17, 106]. There are several grading systems for the semi-quantitative assessment of lesions and overall knee joint health, including such as Internal Cartilage Repair Society (ICRS) grading system, Whole Organ MRI Score (WORMS) method, and Boston-Leeds Osteoarthritis Knee Score (BLOKS) meth-od. In addition, contrast enhanced MRI techniques, such as dGEM-RIC, enable quantitative assessment of cartilage health. In this
technique, mobile anionic gadopentate molecules are assumed to diffuse and distribute in cartilage in inverse proportion to GAGs [23, 26]. Although it is assumed that only FCD in the tissue and the charge of the contrast agent molecule affect the diffusion, it has been demonstrated that dGEMRIC is also sensitive to other factors, such as, collagen content [123–125]. This technique has also been used for imaging of the meniscus and assessment of its health [33, 36, 126]. Even though MRI provides a good soft tissue contrast, it lacks rapid 3D image acquisition with sufficient reso-lution. In addition, relatively high imaging costs, long acquisition and queue times limit its use in the detection of early degeneration or acute injuries.
CT arthrography (CTa) employs iodinated contrast agents to en-hance the contrast between the synovial fluid and the articular car-tilage and meniscus. In CTa, the image acquisition is performed rapidly after intra-articular injection of the contrast agent. This en-ables morphological evaluation of cartilage and meniscus but not assessment of internal structures. CTa may be used for the detec-tion of articular cartilage and meniscal lesions [127].
Arthroscopy is the gold standard in the evaluation of cartilage and meniscal lesions [106, 120, 128]. The modality enables direct visual inspection and probing of cartilage and meniscal surfaces, and relies on subjective assessment by the surgeon. Arthroscopic ultrasound imaging and arthroscopical optical coherence tomog-raphy may also be performed during a conventional arhtroscopic investigation. Even though these modalities allow high resolution imaging [129–132], they are not yet in widespread clinical use.
3 Contrast enhanced com-puted tomography
The German physicist, Wilhelm Röntgen, discovered X-rays in 1895, and the use of X-rays in medical imaging was widespread already before the beginning of the 20th century. In 1917, Johann Radon introduced the mathematical theory (i.e. Radon transform) behind computed tomography reconstruction. However, it was not until 1971 when Godfrey Hounsfield introduced X-ray computed tomog-raphy into medical practice. In CT, X-ray projections are acquired at multiple projection angles around the imaged object. Based on these projections, a 3D image of the object can be reconstructed.
3.1 X-RAY COMPUTED TOMOGRAPHY
X-rays are electromagnetic radiation with the usual wavelength of 0.01 to 100 nanometers. The X-ray imaging is based on attenuation differences between tissues. Due to various interactions with the medium, X-rays attenuate according to the Beer-Lambert law:
I(x) = I0e−µx, (3.1) where Iis the intensity of the radiation at distancex, I0is the initial intensity, andµis the linear attenuation coefficient, which depends on the energy of the photon, atomic number (Z), and the electron density of the material. When X-ray energies are in the diagnostic range (< 200 keV), the attenuation is due to three interactions: pho-toelectric effect, Compton effect, and elastic scattering. The photo-electric effect is the predominant form of the interaction with diag-nostic X-ray energies. In general, the photoelectric effect contributes to the image contrast whereas Compton effect contributes to image noise. [133, 134]
In a CT scan, the X-ray source and the detector rotate around the imaged object while acquiring multiple images at multiple pro-jection angles. These propro-jection images cover 180 to 360 degrees of the imaged object. Subsequently, the acquired projection im-ages are mathematically reconstructed into two-dimensional (2D) axial images using either analytical (e.g. filtered back projection) or more computationally demanding iterative reconstruction algo-rithms. Subsequently, these 2D axial images are used to form a 3D image of the imaged object. [133, 135]
Modern clinical CT scanners employ either a fan-shaped or a cone-shaped X-ray beam. The fan beam shape is achieved by ex-tensive collimation of the X-ray beam. This reduces the amount of scattered photons reaching the detector, and thus increases the signal-to-noise ratio (SNR). Fan-beam shape is commonly applied in full-body CT scanners, with multiple detector rows (up to 320 rows). In cone beam scanners, the collimation of the X-ray beam is wider, which increases the scattering. This, with other CBCT char-acteristics (e.g., more complex reconstruction) results in lower SNR and contrast resolution compared with that of a fan beam scanner.
Cone beam CT (CBCT) scanners, however, typically have smaller radiation doses than fan beam CT scanners [136, 137]. CBCT scan-ners are commonly used in dental and extremity imaging, as well as in X-ray microtomography (microCT). MicroCT scanners are not in clinical use, but in the laboratory use they can be employed to achieve isotropic voxel size down to one cubic micrometer. To ob-tain maximal resolution with microCT, the rotating object is placed closer to the X-ray source. This magnifies the image on the detector as follows
M= O
0
O = dso
dsd, (3.2)
whereO0 is the magnified object size, Ois the original object size, dso is the source-to-object distance, anddsdis the source-to-detector distance.
Contrast enhanced computed tomography