Enhanced Diagnostic Accuracy of Fast-Field Echo Resembling a Computed Tomography Using Restricted Echo-Spacing for Hand Tendon Injuries: A Comparative Study

Abstract

Purpose

Fast-field echo resembling a computed tomography using restricted echo-spacing (FRACTURE), a magnetic resonance imaging (MRI) sequence (Philips Healthcare) providing computed tomography-like bone contrast, enables excellent visualization of bones and tendons. Volume rendering (VR) images created from this sequence show anatomical details of a patient’s actual structures. This study evaluated the diagnostic performance of FRACTURE and its VR images for visualizing wrist tendons on MRI images from patients and healthy individuals.

Methods

Twenty subjects were included, 10 patients with tendon injuries and 10 healthy individuals who underwent MRI scanning. The imaging protocols were FRACTURE (with VR images), T2-weighted imaging (T2WI), and proton density-weighted volumetric isotropic turbo-spin-echo acquisition (PD-VISTA). An orthopedic surgeon, a hand-surgery specialist, and a radiologist not involved in the surgeries used three methods to interpret the images: FRACTURE multiplanar reconstruction, FRACTURE VR, and PD-VISTA + T2WI. The assessment criteria included tendon injury diagnostic accuracy, interpretation time, and ease-of-interpretation scoring. Diagnostic accuracy was calculated as the correct diagnosis rate. Interpretation time and scoring were analyzed by paired t tests and Bonferroni correction for multiple comparisons.

Results

Fast-field echo resembling a computed tomography using restricted echo-spacing VR showed the highest tendon injury diagnostic accuracy, with considerably better interpretation time performance and ease-of-interpretation scoring versus PD-VISTA + T2WI. In the surgical cases, the tendon rupture sites on FRACTURE images corresponded to the intraoperative findings.

Conclusions

Fast-field echo resembling a computed tomography using restricted echo-spacing was excellent across all assessment criteria, and FRACTURE VR had particularly high diagnostic performance and clinical utility. Although obliquely oriented tendons, such as of the thumb and little finger, require consideration of the magic angle effect, clear visualization of these tendons was possible with appropriate positioning. Additionally, FRACTURE’s high tendon visualization capability enabled the identification of anatomical variations, enhancing its usefulness in clinical hand-surgery practice.

Clinical relevance

Tendon injury diagnosis was faster and more accurate with FRACTURE VR than with conventional sequences.

Finger-tendon injuries, including traumatic ruptures and subcutaneous ruptures associated with degenerative joint disease, are clinically common conditions. Treatment typically involves surgical intervention, primarily tendon suturing, tendon grafting, or tendon transfer procedures.

For these surgeries, accurate preoperative assessment of the location of tendon injuries and the condition of donor tendons are critically important. Conventional diagnostic methods for tendon injuries include 3-dimensional computed tomography (CT), ultrasonography, and magnetic resonance imaging (MRI). Although 3-dimensional CT can effectively visualize and enable the diagnosis of flexor–tendon injuries, no reports have indicated its superiority in depicting the finger-extensor tendons, which are thinner and flatter than flexor tendons. Ultrasonography has been reported to excel in evaluating injured tendon sites.^1,2 Nevertheless, although ultrasound does not inherently struggle with anatomical comprehension, it is operator-dependent and is considered to lack comprehensive 3-dimensional spatial resolution. Magnetic resonance imaging is one of the most effective modalities for assessing both the anatomical configuration and structural integrity of wrist tendons. T2-weighted imaging (T2WI) is generally considered to have high visualization capability, and previous studies have suggested that proton density-weighted volumetric isotropic turbo-spin-echo acquisition (PD-VISTA) also excels in tendon visualization.^3,4 However, these methods remain challenging for comprehensively understanding the 3-dimemnsional configuration of finger tendons; hence, it is a more convenient diagnostic method that would be beneficial. Furthermore, finger tendons, especially extensor tendons, frequently present with anatomical variations.5, 6, 7, 8, 9, 10 For smoother surgical procedures on finger tendons, it is ideal to identify before surgery such variations, but conventional imaging techniques have limitations in accurately diagnosing these tendon variations.

Fast-field echo resembling a CT using restricted echo-spacing (FRACTURE), a CT-like bone contrast sequence from Philips Healthcare, was developed as an MRI technique for bone that excels in visualizing cortical bone and trabecular structure, which traditionally are difficult to evaluate by MRI.¹¹ We observed that when performing FRACTURE imaging of finger tendons, they appeared distinctly white with excellent contrast against surrounding soft tissues (Fig. 1A). We then considered that FRACTURE might excel not only in bone visualization but also in tendon depiction. We also discovered that creating volume rendering (VR) images from FRACTURE sequences produced atlas-like anatomical images that clearly depicted tendon morphology, course, and insertion sites (Fig. 1B; Video S1, available online on the Journal’s website at https://www.jhsgo.org).

Figure 1.

A Fast-field echo resembling a CT using restricted echo-spacing (FRACTURE) multiplanar reconstruction (three directions). The tendon exhibits a clear high signal similar to the signal for bone cortex, with distinct contrast against the surrounding muscles. B FRACTURE VR: The course of the tendon is depicted three-dimensionally along with the skeletal structure, making it easier to assess the condition of individual patient’s tendons.

There are no previous reports of tendon visualization using FRACTURE; hence, the capability of FRACTURE for tendon depiction remains unclear. We hypothesized that FRACTURE images would provide superior visualization compared with conventional tendon imaging sequences, such as T2WI and PD-VISTA, and would be valuable for diagnosing tendon injuries, whereas VR images created from FRACTURE would further enhance physicians’ diagnostic capabilities. The study’s aim was to evaluate the diagnostic performance of FRACTURE and its VR images for visualizing wrist tendons by MRI of patients and healthy individuals.

Materials and Methods

Participants

The study included 20 participants, 10 patients with suspected tendon ruptures based on physical examination who underwent MRI at our hospital from September 2021 to April 2023 and 10 healthy volunteers who underwent MRI from October 2022 to March 2023. This study was approved by the ethics committee of our institution. For patients with tendon injuries, we used clinical information from patients who had not opted out. Written explanations and consent forms were obtained from healthy volunteers. For reference, in a paired t test with a 0.0167 two-sided significance level, a sample size of 20 was calculated to ensure 80% power to detect an effect size of 0.8 (Cohen’s large effect size).

MRI acquisition protocol

A 3.0T MRI unit (Intera Achieva Nova Dual Release 5, Philips Medical Systems), with an eight-element ds small-extremity coil, was used to perform imaging. In addition to the FRACTURE sequence (isotropic 3-dimensional), we also obtained 2-dimensional T2WI and 3-dimensional isotropic fat-suppressed PD-VISTA. The imaging protocols for each sequence are shown in Table 1. When imaging healthy volunteers, scans were performed in two positions: the neutral position (0° wrist extension, 0° ulnar deviation, fingers extended) and the ulnar-deviated position (0° wrist extension, maximum ulnar deviation, fingers extended); however, only images in the neutral position were used for interpretation. For patients with tendon injuries, radial-ulnar deviation was adjusted during imaging to ensure that the suspected ruptured tendon was as parallel as possible to the static magnetic field.

Table 1.

Sequence Protocol

Sequence Parameter	FRACTURE	PD-VISTA	T2WI
Image orientation	coronal	coronal	axial
Repetition time (ms)	18	1,300	5,000
Echo time (ms)	2.3/4.6/6.9/9.2	34	80
Flip angle (°)	15	90	90
Echo train length	n/a	45	15
Number of signals acquired	1	1	2
Turbo spine-echo factor	n/a	45	15
Section thickness (mm)	0.6	0.6	3.0
Resolution (reconstruction) (mm)	0.6/0.6/0.6 (0.3/0.3/0.3)	0.6/0.6/0.6 (0.3/0.3/0.3)	0.3/0.3/3.0 (0.2/0.2/3.0)
Field of view (mm2)	200	200	160
Acquisition matrix	332 × 264	332 × 250	532 × 415
Band width (Hz/pixel)	558	339	287
Acquisition time (min)	3.32	3.46	3.40

VR image creation

For FRACTURE sequences, VR images displaying bones, flexor tendons, and extensor tendons were created on a workstation (Ziostation2 plus, Ziosoft Inc).

Evaluation methods

Three physicians who were not involved in the surgeries, an orthopedic surgeon (6 years of orthopedic experience), a hand-surgery specialist (14 years as a hand-surgery specialist, 21 years in orthopedics), and a radiologist (32 years in radiology), used the following three methods to perform image interpretation:

• 1.FRACTURE multiplanar reconstruction (MPR) images in three planes (axial, sagittal, and coronal) (FRACTURE MPR) (Fig. 1A).
• 2.FRACTURE VR images (Fig. 1B).
• 3.Non-FRACTURE sequences: PD-VISTA (MPR in three planes; Fig. 2A) and T2WI axial images (Fig. 2B; PD-VISTA + T2WI).

Figure 2.

A Proton density-weighted volumetric isotropic turbo-spin-echo acquisition (multiplanar reconstruction in three directions); B T2-weighted imaging (axial). In both images, the tendon exhibits a low-signal intensity and is depicted relatively clearly.

In this study, we aimed to evaluate the utility of FRACTURE MPR images in three orientations (axial, sagittal, and coronal) and FRACTURE VR images separately, which is why we conducted independent readings for each.

Also, as a control for these FRACTURE images, we selected PD-VISTA, which has been conventionally used at our institution for tendon injury assessment with non-FRACTURE sequences. Additionally, considering the potential influence of the magic angle effect (which will be discussed later), we decided to include T2WI as part of the control reading protocol.

The diagnostic capability was evaluated according to the following three criteria:

• 1Accuracy of tendon injury diagnosis: Correct answers were counted, including both the presence and location of the injury when applicable.
• 2Time required for interpretation: The time from the beginning to the end of the interpretation process was measured.
• 3Ease-of-interpretation scoring: Interpreters used a visual analog scale-like method to score the ease-of-interpretation from 0 (difficult to interpret) to 10 (easy to interpret; Fig. 3).

Figure 3.

Scoring scale used for readability assessment: For ease-of-interpretation, the participants were asked to mark (orange line in the figure) on a visual analog scale, where 0 represents “difficult to interpret” and 10 represents “easy to interpret,” and the results were measured.

For individuals who underwent surgery, the accuracy of the FRACTURE VR images was confirmed by comparison with surgical findings.

Statistical analysis

To evaluate how accurately tendon injuries were diagnosed, we calculated both the correct diagnosis percentages and their 95% confidence intervals (CI). To determine if there were meaningful differences between these accuracy rates, we used Cochran’s Q statistical test.

For the time required for interpretation and ease-of-interpretation scoring, paired t tests and Bonferroni correction for multiple comparisons (three corrections) were used, with a significance level of P = .0167. Additionally, for these measurements, we calculated the intraclass correlation coefficient and weighted kappa coefficients, respectively, to evaluate interobserver errors.

Results

Of the 10 patients diagnosed with tendon injuries, three were men and seven were women. The mean age was 66.5 years (range: 27–87 years). The causes of tendon injuries were rheumatoid arthritis in five patients, traumatic extensor tendon rupture in four, and traumatic flexor–tendon rupture in one. Surgery was performed in eight patients, with tendon suturing in one and tendon transfer in seven. Among the 10 healthy volunteers, there were five men and five women, aged 27–55 years (mean: 37.4 years, including two in their 20s, four in their 30s, two in their 40s, and two in their 50s). Incidentally, hypoplasia of the extensor indicis proprius (EIP) was unexpectedly detected in one healthy volunteer (Fig. 4A; a normal case is shown in Fig. 4B).

• 1Diagnostic accuracy: The mean correct diagnosis rates were 83.3% (95% CI, 71.3% to 91.0%) for FRACTURE MPR, 92.6% (95% CI, 82.5% to 97.1%) for FRACTURE VR, and 81.5% (95% CI: 69.2% to 89.6%) for PD-VISTA + T2WI (Fig. 5). As a result of Cochran's Q test, P = .57, and no significant difference was found among the three reading methods. When comparing the correct diagnosis rates for patients without injuries, both the hand surgeon and radiologist achieved 100% accuracy in all methods, whereas the orthopedic surgeon incorrectly diagnosed injuries in one patient (10%) when using FRACTURE MPR and in two patients (20%) when using FRACTURE VR. The radiologist was able to make detailed diagnoses of lesions and anatomical variations when using all sequences.
• 2Interpretation time: The mean values and standard deviations for each sequence were 69 ± 42 seconds for FRACTURE MPR, 36 s ± 31 s for FRACTURE VR, and 87 s ± 49 s for PD-VISTA + T2WI, (Fig. 6). Comparisons of interpretation times showed significant differences between all groups: FRACTURE MPR versus FRACTURE VR, P < .0167 (P = 2.99 × 10⁻⁷), FRACTURE VR versus PD-VISTA + T2WI, P < .0167 (P = 1.68 × 10⁻¹¹), and FRACTURE MPR versus PD-VISTA + T2WI, P < .0167 (P = 8.04 × 10⁻³) (Fig. 6).
• 3Ease-of-interpretation: The mean values and standard deviations for each sequence were 8.0 ± 1.3 for FRACTURE MPR, 8.4 ± 2.1 for FRACTURE VR, and 6.6 ± 2.1 for PD-VISTA + T2WI. Although there was no significant difference between FRACTURE MPR and FRACTURE VR (P = .137), both FRACTURE MPR and FRACTURE VR showed significant differences compared with PD-VISTA + T2WI (Fig. 7).

Figure 4.

A Shown is a case of extensor indicis proprius hypoplasia (EIP); B Normal case. The EIP running along the ulnar side of the extensor digitorum of the index finger is noticeably thinner and hypoplastic compared with the normal case.

Figure 5.

Accuracy of tendon injury diagnosis. The correct diagnosis rate was 83.3% for fast-field echo resembling a CT using restricted echo-spacing (FRACTURE) multiplanar reconstruction (MPR), 92.6% for FRACTURE VR, and 81.5% for proton density-weighted volumetric isotropic turbo-spin-echo acquisition (PD-VISTA) + T2-weighted imaging (T2WI), with FRACTURE VR showing the highest accuracy.

Figure 6.

Time required for interpretation. Fast-field echo resembling a computed tomography using restricted echo-spacing (FRACTURE) multiplanar reconstruction (MPR) took 69 ± 42 seconds, FRACTURE VR took 36 ± 31 seconds, and proton density-weighted volumetric isotropic turbo-spin-echo acquisition (PD-VISTA) + T2-weighted imaging (T2WI) took 87 ± 49 seconds, with FRACTURE VR taking significantly less time than the others.

Figure 7.

Readability scoring. The scores were 8.0 ± 1.3 for FRACTURE multiplanar reconstruction (MPR), 8.4 ± 2.1 for FRACTURE VR, and 6.6 ± 2.1 for proton density-weighted volumetric isotropic turbo-spin-echo acquisition (PD-VISTA) + T2-weighted imaging (T2WI), with FRACTURE VR receiving the highest score.

In evaluating interobserver variability, the intraclass correlation coefficient for reading time was intraclass correlation coefficient (2,3) = 0.475, and the weighted kappa coefficients for scoring are shown in Table 2.

Table 2.

Inter-Rater Reliability for Ease-of-Interpretation Scoring (Weighted Kappa Coefficients)

Rater Pair	A・B	B・C	C・A
Kappa coefficients	0.086	0.027	0.199

A, hand-surgery specialist; B, orthopedic surgeon; C, radiologist.

In the surgical cases, the tendon rupture sites shown on FRACTURE images generally corresponded to the intraoperative findings. A representative case is presented below.

Case presentation

A 59-year-old woman with rheumatoid arthritis presented with difficulty extending her little finger and discomfort when extending her index, middle, and ring fingers. She was diagnosed with little finger-extensor tendon rupture and dislocation of the index, middle, and ring finger-extensor tendons. FRACTURE VR images showed rupture of the little finger-extensor tendon (at the wrist level) and dislocation of the extensor tendons of the index, middle, and ring fingers at the metacarpophalangeal joints (Fig. 8A). For the little finger-extensor tendon, the distance of the distal end of the tendon rupture from the radiocarpal joint relative to the total distance between the radiocarpal and little finger metacarpophalangeal joints was 13.2% on FRACTURE images and 13.5% when measured during surgery, demonstrating close agreement (Fig. 8B).

Figure 8.

A 59-year-old woman rheumatoid arthritis patient. A Fast-field echo resembling a computed tomography using restricted echo-spacing (FRACTURE) VR image of the patient. Rupture of the extensor digiti minimi tendon at the wrist level and dislocation of the extensor tendons at the metacarpophalangeal joints of the index, middle, and ring fingers are observed. B Actual intraoperative photograph. Rupture was confirmed at the same location as shown in the FRACTURE VR image.

Discussion

Conventionally, the visualization capability of T2WI is considered to be high, and previous studies have suggested that PDWI also excels in visualizing tendons.^3,4 In the present study, for all evaluation criteria, the results for FRACTURE were superior to those of high visualization sequences, such as T2WI and PDWI. For finger tendons, the results suggest that FRACTURE can provide diagnostic capabilities at least equivalent to or better than those of T2WI and PDWI, indicating its usefulness. Furthermore, FRACTURE VR exhibited the highest diagnostic accuracy rates. The interpretation time was significantly shorter for FRACTURE VR than for the other methods. Regarding ease-of-interpretation, both hand surgeons and radiologists tended to give the highest scores to VR. These results show that FRACTURE VR, which enables quick interpretation with high accuracy, should be very useful in clinical practice.

However, in this study, some patients had tendon ruptures of the thumb or little finger that were erroneously identified as intact tendons; 10% when using the MPR method and 20% with the VR method. These errors are probably attributable to the magic angle effect, which occurs when imaging structures with collagen fibers align in a specific direction, as seen with tendons when using short TE sequences, such as FRACTURE.^12,13 This phenomenon causes normally low-signal tendons to show increased signal intensity, making them appear partially deficient, and occurs when artifacts are generated in a conical range at approximately 55° to the direction of the static magnetic field (imaging axis; Fig. 9). This phenomenon can make tendon injury diagnosis difficult, necessitating careful interpretation. To confirm the presence of tendons in areas affected by the magic angle effect, verification by T2WI, which has a relatively long echo time and is less susceptible to the magic angle effect, is necessary.¹² All of the erroneous diagnoses of healthy patients as injured were made by orthopedic surgeons with limited interpretation experience (6 years). Regarding the ease-of-interpretation scoring by the orthopedic surgeon alone, the average for all patients was 6.88 for FRACTURE MPR and 6.13 for FRACTURE VR, which were easier to interpret than PD-VISTA + T2WI (4.89). However, for patients with healthy tendons misdiagnosed as injured by FRACTURE MPR and VR, the average ease-of-interpretation score was 4.87, markedly lower than the overall average. To reduce misdiagnoses of healthy cases and improve the accuracy of FRACTURE, the magic angle effect must be well understood and recognized as a factor that affects tendon imaging.

Figure 9.

(Left) Fast-field echo resembling a computed tomography using restricted echo-spacing (FRACTURE) VR image of the palmar side. (Right) FRACTURE VR image of the dorsal side. Because of the magic angle effect, on the palmar side, partial defects in the abductor pollicis longus tendon (blue arrow) and flexor pollicis longus tendon (green arrowhead) are observed, whereas on the dorsal side, an interruption in the extensor pollicis longus (orange arrowhead) is observed.

Optimizing the imaging position is crucial to minimize the influence of the magic angle effect. During imaging, the positioning of the target area should be parallel to the imaging axis so that it does not fall within the conical range of approximately 55° from the imaging axis. For example, even when a rupture of the flexor pollicis longus tendon is suspected in the neutral position (Fig. 10A), adjusting the imaging position by ulnar deviation of the wrist can show the tendon with continuity, as shown in Figure 10B, which reveals that the apparent rupture in the neutral position is actually a pseudo lesion. In clinical practice, since the target tendon or lesion can usually be specified, positioning the patient to avoid the magic angle effect on the area of interest should allow accurate tendon visualization.

Figure 10.

A When imaged in the neutral position (wrist at 0° of radioulnar deviation), the flexor pollicis longus appears partially defective because of an artifact caused by the magic angle effect. B When imaged during ulnar deviation (wrist in maximum ulnar deviation). The course of the flexor pollicis longus is clearly depicted as it avoids the magic angle.

Fast-field echo resembling a computed tomography using restricted echo-spacing can clearly depict finger tendons, enabling preoperative assessment of injury sites and the presence of tendons needed for transfer and anatomical variations. Extensor tendons frequently exhibit variations, with numerous pattern variations documented in the extensor digitorum communis, and anatomical diversity observed in the abductor pollicis longus and extensor pollicis brevis.⁶ Anomalous muscles that are not normally present, such as the extensor digitorum brevis manus and flexor carpi radialis brevis (vel profundus), may also be found. Taking the EIP as an example, which is commonly used in tendon transfer procedures for extensor tendon injuries, it typically consists of a single tendon that inserts on the ulnar side of the index finger metacarpophalangeal joint. However, this pattern represents only approximately 80% of these cases, meaning approximately 20% of the patients exhibit variations in insertion site, course, or number of tendons.5, 6, 7, 8, 9, 10 Patients with EIP absence or hypoplasia also have been observed, with absence in approximately 1% of the population previously reported.^14,15 Indeed, one of our healthy volunteers had EIP hypoplasia, which was incidentally discovered by FRACTURE VR and had not been initially noticed (Fig. 4).

A limitation of this study is that there may have been bias in the ease-of-interpretation because of the somewhat skewed sample of tendon injury patient data. We designed the study to include 10 patient cases, but the number of interpretations may have been insufficient. Additionally, we cannot rule out the possibility of potential confounding factors because of differences in imaging positions between healthy subjects and patient data. Future studies should increase the number of cases to determine if high diagnostic capabilities can be demonstrated for various injury sites, standardize imaging positions, and may also consider verifying results with sample sizes appropriate for detecting moderate effect sizes. The number of interpreters, two orthopedic surgeons and one radiologist, may have been somewhat limited for comparative analysis. Furthermore, we used T2WI and PD-VISTA as controls for comparison, which have been suggested in the previous literature as excelling in tendon visualization, but this represents a limited comparison of sequences. Future studies may need to compare these sequences with other sequences to further evaluate their usefulness. When performing tendon transfer or grafting procedures, knowledge of tendon thickness would be beneficial. Although we verified the high diagnostic capability of FRACTURE in this study, we could not address the size reproducibility of FRACTURE, particularly in VR images. Further research, such as to compare measurements of tendons from cadaveric specimens with FRACTURE images, is necessary.

In conclusion, FRACTURE, especially FRACTURE VR, provided excellent visualization of both flexor and extensor tendons, enabling 3-dimensional comprehension of all hand and finger tendons and instantaneous recognition of previously unnoticed anatomical variations or anomalous muscles. If FRACTURE enables accurate preoperative assessment of these diverse variations in finger tendons in a wide variety of patients, it could make notable contributions to hand-surgery practice, including for surgical indications and technique selection.

Conflicts of Interest

No benefits in any form have been received or will be received related directly to this article.

Supplementary Data

Movie S1

Demonstrates FRACTURE VR images of the entire hand (clockwise rotation).