CT-like images based on T1-weighted gradient echo MRI sequences for the assessment of fractures of the hand and wrist compared to CT

Wrist and hand fractures are among the most common bone injuries [1, 2]. Due to the outstanding importance of the upper extremity as a functional and expressive organ, early diagnosis and therapy of wrist and hand fractures are essential for a positive outcome in affected patients [3, 4]. If a fracture of the wrist or hand is suspected, an X-ray examination is usually the first step in radiological diagnostics. Computed tomography (CT) is performed if the clinical picture is typical but there is no evidence of a fracture on the X-ray, or if the therapeutic procedure is to be planned. Magnetic resonance imaging (MRI) is routinely used to detect occult fractures, and to diagnose non-bony concomitant injuries, especially of ligaments and cartilage [5, 6].

Since X-ray and CT examinations are always associated with radiation exposure of the patient, fracture diagnosis using MRI alone would be desirable, especially in younger patients and those who require regular follow-up [7].

Due to its low proportion of free water and its rapid T2* relaxation, direct imaging of cortical bone using MRI has long been difficult. In recent years, however, great progress has been made in the development of new MRI sequences, which have improved the representation of bony structures to such an extent that it is now possible to acquire so-called CT-like images [7, 8]. Their potential for detecting bony fractures, degeneration and osseous pathologies has already been proven in various studies, in particular, in the area of the spine, skull, mandibula, shoulder, knee and tibia [9‐16].

The aim of our study was to evaluate the performance of CT-like MRI images derived from a T1-weighted gradient echo (T1GRE) sequence for the detection and classification of fractures of the wrist and hand compared to CT as standard of reference and to compare image quality and diagnostic confidence between the modalities.

The frequencies of different fractures according to the AO classification are listed in Table 1.

There was no significant difference regarding the overall image quality (p = 0.65) and diagnostic confidence (p = 1.00) between CT-like MRI and CT images (Table 2).

There was no significant difference in the diagnostic confidence for CT-like MRI compared to CT for both raters (rater 1: CT-like MRI, mean ± SD, 1.27 ± 0.55 vs. CT, mean ± SD, 127 ± 0.52; p = 1.00; rater 2: CT-like MRI, mean ± SD, 1.10 ± 0.36 vs. CT, mean ± SD, 1.17 ± 0.42; p = 0.36). The same is applied for the assessment of image quality between CT-like MRI and CT for rater 1 (CT-like MRI, mean ± SD, 1.29 ± 0.59 vs. CT, mean ± SD, 1.14 ± 0.39; p = 0.07) and rater 2 (CT-like MRI, mean ± SD, 1.27 ± 0.64 vs. CT, mean ± SD, 1.42 ± 0.79; p = 0.23), respectively.

Agreement between CT-like MRI and conventional CT images regarding the AO fracture classification was excellent for both readers (reader 1: κ, 0.89; 95% confidence interval (CI), 0.80–0.97 and reader 2: κ, 0.89; CI, 0.80–0.97). Interrater agreement for AO fracture classification was excellent for CT-like MRI (κ, 0.82; 95% CI, 0.70–0.93) and CT images (κ, 0.85; 95% CI, 0.75–0.94).

All 43 fractures were correctly detected by both readers on CT.

The results of the four-square panel for both readers for the fracture detection of CT-like MRI are listed in Tables 3 and 4.

In the case of multiple fractures, all fractures were correctly localized by both readers. According to the consensus reading of the fracture AO classification, reader 1 accurately classified 36 out of 43 fractures in CT-like MRI, compared to 36 out of 43 in CT. Reader 2 accurately classified 40 out of 43 fractures in CT-like MRI vs. 42 out of 43 in CT.

Our analyses revealed that CT-like MRI based on a 3D T1GRE sequence is suitable for the detection and classification of wrist and hand fractures.

Reading CT-like MRI provided equal image quality and diagnostic confidence compared to conventional CT scans. The sensitivity, specificity and accuracy of CT-like MRI for detecting fractures were excellent (reader 1: 1.00, 0.92, 0.96; reader 2: 0.98, 0.94, 0.96). Furthermore, there was an excellent agreement for AO fracture classification between both modalities and between the readers.

The CT-like MRI sequence used in this study has previously been reported to provide excellent image quality and diagnostic confidence for fracture detection in the spine, skull, mandibula and shoulder [10‐13, 15]. While radiography remains the primary means of fracture diagnosis due to its cost-effectiveness, low radiation dose and ubiquitous availability, CT offers advantages over radiography in detecting subtle fractures, visualizing fracture displacement and aiding preoperative planning [18].

CT-like MRI has been proven to reliably detect fractures and even discrete bony avulsions such as bony Bankart lesions of the glenoid after a traumatic shoulder dislocation, subtle fractures of the skull and the tibial eminence in children [11, 13, 16]. Osteoligamentous avulsions in the skeleton of the hand are even more common than in these regions, where apparently small bone avulsions can have major functional complications [19]. Ulas et al. reported an improved specificity of erosion detection in hand arthritis with bone-like MRI based on susceptibility-weighted imaging [20, 21]

However, to the best of the authors’ knowledge, there have been no studies on fracture detection with bone-like MRI of the hand and wrist so far. Our study cohort included a representative cross-section of wrist fractures seen in routine clinical practice, including scaphoid fractures, discrete avulsion fractures of the triquetrum, as well as intra- and extra-articular fractures of the distal radius. Hand and wrist fractures were reliably detected with CT-like MRI (sensitivity > 97%). CT-like MRI’s specificity has been negatively influenced by the false positive rate due to the fact that CT occult/subtle fractures, such as non-displaced scaphoid fractures, become more visible after a few days due to resorption processes at the fracture site (Fig. 4). To minimize this influence, we made efforts to keep the interval between the MRI and CT scans as short as possible (mean time interval between CT and MRI in the fracture group 5.37 ± 7.52 days). In contrast to CT, where bone marrow oedema is not always visible, pronounced oedema may appear bright on inverted images on CT-like MRI, confirming the suspicion of a fracture (Fig. 1) and increasing the diagnostic confidence.

The sensitivity of MRI for detecting wrist and hand fractures was already very high in our study, although only the CT-like sequence was used for diagnosis. In clinical use, the CT-like sequence would typically be integrated into a more comprehensive MRI trauma protocol, including an oedema-sensitive sequence, which has the potential to further improve the sensitivity of the examination [22].

In addition to fracture detection, assessment of fracture morphology and displacement is essential for treatment planning. Our results showed excellent agreement for AO fracture classification between CT-like MRI and CT. This is consistent with the results of Feuerriegel et al., which show accurate agreement between CT and CT-like MRI for the classification of traumatic mandibular fractures and measurement of dislocation [15].

The isotropical, CT-like MRI sequence offers the advantage of multiplanar reconstruction, like a CT scan. Parasagittal and paracoronal multiplanar reconstructions along the longitudinal axis of the scaphoid are particularly helpful in the assessment and treatment planning of scaphoid fractures [23, 24].

Usually, standard sequences are acquired together with CT-like 3D T1GRE sequences. These sequences allow the detection of further injury patterns, including traumatic bone marrow oedema (“bone bruises”), as well as ruptured ligaments (e.g. scapholunate ligament rupture and TFCC injuries) and ruptured tendons.

This study has several limitations. First, the acquisition of the 3D T1GRE CT-like sequence requires a 3 T MRI scanner with a state-of-the-art gradient system. On older scanners, examination duration might be prolonged due to increased minimum TR. Furthermore, the 3D T1GRE sequence is susceptible to metal artefacts. CT should therefore continue to be the preferred examination method for patients with metal implants. Patients with metal implants were not included in our study.

Particularly in the context of emergency care, MRI is usually less accessible than CT and associated with higher costs. Data on the imaging cost-effectiveness of hand and wrist fractures is very limited. According to a cost-effectiveness study on radiographically occult scaphoid fractures by Karl et al. from 2015, the mean price readjusted to 2014 values based on the Medical Consumer Price Index of a radiograph, CT and MRI was $54 ± $14, $374 ± $166 and $521 ± $120, respectively [25]. In summary, the study concludes that utilizing advanced imaging for suspected scaphoid fractures despite negative radiographs is cost-effective due to its affordability, reliable diagnostic precision and better projected health outcomes compared to empiric cast immobilization. Choosing between CT and MRI depends on institutional expenses, the willingness of the patient to undergo a radiation-dependent modality and the specific diagnostic capabilities available locally. Although radiography remains the first-line imaging method for the detection of hand and wrist fractures, CT-like MRI combined with a single coronal IM fs could be used instead of CT for fracture diagnosis in case of inconclusive radiographs and assessment of fracture morphology in the clinical setting. Reducing the number of sequences would indeed lower the MRI cost by saving time and might reduce the waiting period for MRI appointments.

In addition, the longer examination time compared to CT requires good patient compliance, and the MRI is contraindicated for individuals with certain implants, such as pacemakers. Therefore, the use of the proposed method may be limited to centers with the necessary resources and/or to patients who meet the above requirements.

3D T1GRE cannot distinguish between bone and ligaments, both appearing bright on grey inverted reformatted images. This may lead to false positive findings of avulsion fractures in the carpus due to the close proximity of tendons and bones. Crystal deposition associated with hydroxyapatite crystal deposition disease or calcium pyrophosphate dihydrate crystal deposition disease may also appear bright on intensity-inverted reformatted images, leading to false positive findings, but we had none of these cases in our cohort [26, 27].

This study shows the effective diagnostic performance of CT-like MRI in the assessment of fractures in the wrist and hand compared to CT, the standard of reference, with comparable image quality and diagnostic confidence. The routine use of CT-like MRI in emergency diagnostics for suspected fractures of the wrist and hand has the potential to reduce the number of CT scans and the associated radiation exposure. Further research is needed to evaluate the potential impact on healthcare costs and MRI waiting times.