Axial spondyloarthritis (axSpa) is a chronic inflammatory disease mainly targeting the axial skeleton, i.e. the spine and the sacroiliac joints (SIJ) [1,2]. The sacrum and the SIJ are the most challenging for adequate imaging because of their complex anatomy, their composition of multiple tissues and their posterior location in the pelvis [3]. Sacroiliitis is characterised in early stages by active inflammation, of which bone marrow edema is the hallmark feature. It is the most important imaging finding signalling active inflammation and a key element for diagnosis and therapy. Concomitantly or secondary to this inflammation, structural changes occur, including erosion, sclerosis and ankylosis [1,2,[4], [5], [6]].

The introduction of MRI and the subsequent shift from structural lesions as detected by radiography towards active inflammation has been an important step towards an earlier diagnosis. This became necessary with the advent of modern biologicals that allow for effective suppression of inflammation, can prevent the development of structural damage and stimulate healing processes when inflammation subsides [7]. They can help to preserve the full physical functionality of axSpA patients but warrant their early identification before structural changes, especially ankylosis, appear.

Magnetic resonance imaging (MRI) is the imaging modality of choice in evaluating bone marrow edema [1,3,4,8]. Fat-suppressed fluid-sensitive images such as fat-suppressed T2-weighted images, short tau inversion recovery, chemical shift Dixon method or spectral attenuated inversion recovery are preferable because they are fluid sensitive [1,[9], [10], [11], [12], [13], [14]]. Contrast media is usually unnecessary for assessing the SIJ. However, bone marrow edema seen in MRI is a very sensitive but less specific imaging sign. It is present in several groups, including healthy volunteers, patients with mechanical load or post-partum women [15]. For this reason, the last update of the ASAS classification criteria shifts the focus from the mere presence of active inflammation towards a combination with structural lesions [16].

Structural lesions of the sacroiliac joint include erosion of the articular surface, fat lesions and sclerosis of the bone marrow and forms of new bone formation such as bone buds or backfill. While the latter is a lesion that was solely described for MR imaging, not all the above-mentioned findings are best detected by standard MR sequences. Conventional MRI gains signal from protons, i.e. hydrogen atoms, within tissues. Hydrogen is abundant in the fat and water of bone marrow but sparse within the calcified bone matrix of trabeculae and cortex. Therefore, standard MR sequences such as T1 and short-tau inversion recovery (STIR) that are usually applied display the cortical bone and bone structure only indirectly. They also depict the subchondral cortex, which is the target for erosion, with comparably low contrast. Therefore, other modalities or more sophisticated MR techniques are necessary to detect those lesions [17,18]. Fig. 1 compares CT and MRI images in three different patients, illustrating the detection of structural lesions and/or bone marrow edema.

In the following, the article will discuss several approaches for direct bone imaging at the sacroiliac joint. A complete overview of the pros and cons of these imaging modalities is presented in Table 1 [1,3,5,19].

Computed tomography (CT) is known for its excellent soft tissue-bone contrast and high spatial resolution. Therefore, it is the imaging modality of choice concerning the evaluation of structural changes of the calcified bone structure in the sacroiliac joints. Of note, fat lesions of the bone marrow are less amenable to CT detection. CT is widely available and has a short examination time. Unfortunately, CT is associated with undeniable radiation exposure [1,3,4,19]. An example of structural lesions detected on CT is given in Fig. 2.

Low-dose CT largely tackles the issue of radiation exposure, with a reduction of radiation to the level of radiography. Despite the comparable radiation dose, it has a significantly higher sensitivity for erosions compared to radiography [1,[19], [20], [21]]. While a reduction of the radiation is linked to increased noise, high contrast imaging features, such as erosions, tolerate this fact and, therefore, the accompanying dose reduction. The application of low-dose CT techniques is further facilitated by modern reconstruction techniques that allow a sufficient reduction of image noise and the diagnostic interpretation of acquisitions that would be non-diagnostic with previous postprocessing techniques. Coming from filtered back projection 30 years ago, the introduction of image-based and raw-data-based iterative reconstructions and subsequently the up-to-date reconstructions with artificial intelligence increasingly reduced the image noise while improving the image quality or allowing further reduction of the radiation dose [22,23]. Fig. 3 illustrates these different postprocessing techniques.

In addition, there are several new hardware developments to improve the image quality and reduce the radiation exposure of CT scans. For example, tin-filtration of the X-ray beam reduces the amount of harmful radiation by filtering low-energy photons that are usually absorbed by the patient's body and never reach the CT detector [24]. While this can reduce the image contrast, it will also markedly reduce the radiation exposure and ensure sufficient image quality. On the other hand, new detector technologies, namely photon-counting CT, promise an increase in sensitivity to X-rays and better spatial resolution, allowing further dose reduction and improvement of image quality [25].

Regardless of being impeccable at detecting erosions, sclerosis and ankylosis in the sacroiliac joints, standard CT and low-dose CT fail to adequately visualise bone marrow edema [[19], [20], [21], [22], [23], [24], [25], [26]].

Dual-energy computed tomography (DECT) provides additional information. As the name assumes, the CT is performed at two different energies. These different tube voltages and the different attenuation values between bone marrow edema and normal bone marrow allow for the detection of osteitis [5,19,[27], [28], [29], [30]]. In postprocessing, calcium components in the bone are subtracted, which results in a virtual image that contains only the bone marrow. These so-called virtual non-calcium images can be presented as grey-scaled or colour-coded maps [1,5,31]. DECT is specific, meaning that areas suspected of bone marrow edema, actually do represent bone marrow edema [5]. However, diagnostic accuracy of DECT in bone marrow edema is reduced if underlying red bone marrow or sclerosis is present, which is mainly in the subcortical region [1,5,19,28,32]. This contributes to a moderate sensitivity [5].

Like standard CT, DECT has a wide availability and a short examination time. However, despite the many strengths of DECT, experienced readers are needed for a reliable evaluation of the images. Also, DECT involves a considerable amount of radiation and low-dose variants are currently not possible due to noise susceptibility in postprocessing [7,28,33]. Nonetheless, DECT is a reasonable alternative in the evaluation of bone marrow edema for patients contraindicated to undergo MRI [1]. Fig. 4 illustrates the detection of bone marrow edema with DECT, whereas Fig. 5 correlates the DECT findings with MRI.

Most structural changes (e.g. backfill, fat metaplasia and ankylosis) can be assessed in standard T1-weighted images. However, as described above, MRI is less sensitive to small changes of cortical and trabecular bone. Three-dimensional gradient echo sequences (3D GRE) are better in the detection of small erosions compared to classical T1-weighted images because they increase not only the spatial resolution but also the contrast between cartilage and bone surface. Susceptibility-weighted imaging (SWI) is a 3D GRE sequence that detects small magnetic field inhomogeneities that are caused by calcification or bone. Therefore, it allows direct depiction of the cortex and trabeculae by measuring the signal loss that they cause. Inverting SWI images creates a CT-like image impression, which can depict erosion and sclerosis better than T1-weighted images [1,3,4,34,35]. Similar sequence techniques use ultra-short or zero-echo-time imaging for a direct depiction of the cortex [36,37].

A different and even more elegant approach involves artificial intelligence that makes it possible to create radiography-like and CT-like images of MRI sequences, called synthetic CT or Bone-MRI. These images excel in the detection of structural changes compared to T1-weighted images and are as reliable as CT for the detection of structural changes [1,3,38]. In this way, a combination of MRI and Bone-MRI images combine the evaluation of bone marrow edema and structural changes, as shown in Fig. 6. They require synthetic CT postprocessing software, which is completely automised [1,39]. Fig. 7 compares MRI findings and low-dose CT instead of Bone-MRI.

Another advantage of MRI, is the complete absence of radiation. Multiple disadvantages are present, such as the high cost, long examination time and high sensitivity to movement. Claustrophobia, foreign bodies and various metallic implants can be contra-indicated to MRI [3].