Circumferential Meniscal Extrusion Mapping Using Radial Multiplanar Reconstructions

Abstract

Meniscal extrusion is a marker of meniscal dysfunction and has been associated with osteoarthritis progression. Assessment is typically performed on a single coronal magnetic resonance imaging (MRI) slice, but this method may miss the location of maximal circumferential displacement. We developed an extrusion mapping technique using radially oriented multiplanar reconstruction (rMPR) images from high-resolution 3D MRI to enable circumferential meniscal position measurements perpendicular to the tibial margin. Extrusion was measured by three independent readers on rMPR images generated from non-fat-saturated 3D fast spin-echo MRI in 20 patients stratified by meniscal tear and extrusion status. Intrareader and interreader measurement variabilities were assessed using intraclass correlation coefficient (ICC), 95% repeatability coefficient (RC; intrareader), and 95% reproducibility coefficient (RDC; interreader). Corresponding rMPR and standard 2D coronal midbody extrusion measurements were compared. Intrareader variability was low, with a mean within-meniscus within-angle standard deviation (SD) of 0.2 mm and mean RC of 0.6 mm. Interreader variability was also low, with a mean within-meniscus within-angle SD of 0.5 mm and mean RDC of 1.4 mm. At the midbody, there was 98% concordance for extrusion classification between rMPR and 2D coronal measurements within the interreader RDC (1.4 mm). Exploratory regional analysis suggested different extrusion patterns between medial and lateral menisci and between torn and intact menisci. These findings suggest that circumferential meniscal extrusion mapping using rMPR images from 3D knee MRI is feasible, with low intrareader and interreader measurement variabilities. This framework enables regional characterization of meniscal displacement and should support future investigations of meniscus–osteoarthritis relationships.

1. Introduction

The menisci play a central role in maintaining knee health and function, contributing to load distribution, shock absorption, and joint stability [1, 2]. When the circumferential meniscal fibers are disrupted through macroscopic or microscopic tearing or degeneration, the meniscus may become deficient or “defunctionalized,” meaning that it is no longer able to distribute axial loads into hoop stress [2, 3]. The damaged meniscus may become abnormally displaced, with its peripheral margin “extruded” beyond the tibial margin. This displacement can be seen on knee magnetic resonance imaging (MRI). Abnormal extrusion on MRI has been associated with the progression of cartilage loss, development of symptomatic osteoarthritis (OA), and subchondral insufficiency fractures [4, 5]. Meniscal extrusion has become a target for surgical correction, i.e., meniscal centralization, as a concomitant procedure for patients having meniscal repair, meniscal root repair, or meniscal transplant or for those with isolated extrusion [6, 7]. However, clinical practice operates without a consensus on the pathological thresholds for meniscal extrusion, which undermines informed decision-making for treatments. MRI and ultrasound studies have shown that with knee flexion, there can be a variable degree of “physiological” extrusion with normal meniscal motion [8, 9]. Establishing a rigorous, systematic understanding of normal meniscal position and extrusion is therefore necessary to advance evidence-based interventions.

MRI is the best noninvasive method for the evaluation of meniscal extrusion [10]. On MRI, extension of the peripheral margin of the meniscus beyond the tibial edge by more than 3 mm is generally considered abnormal [11]. However, some research suggests that lower thresholds should be considered pathological, as these thresholds may predict OA progression in the absence of other radiographic features [12, 13]. Analyses of Osteoarthritis Initiative (OAI) data [14] demonstrated that medial meniscal extrusion of more than 2.5 mm and lateral meniscal extrusion of more than 1.3 to 1.4 mm reliably predicted symptomatic knee OA progression [12, 13].

Regardless of the threshold deemed appropriate, accurate representation of the meniscal position is necessary to ensure that meniscal extrusion is precisely measured. In clinical practice, meniscal extrusion on MRI is most often measured at a single point, such as the body of the meniscus on the mid-coronal image through the medial tibial spine, or the coronal slice showing the greatest degree of extrusion [10]. Measurement on the mid-coronal image is most accurate because the imaging plane is orthogonal to the tibial margin; coronal image measurements performed elsewhere may overestimate the degree of extrusion because of partial volume averaging from the obliquity of the image slice relative to the tibial and meniscal margins [10]. However, sampling the meniscal position at only one point may miss the maximal meniscal displacement and underestimate the degree of extrusion [10]. These issues can be resolved by measuring meniscal extrusion on radially oriented images orthogonal to the tibia throughout the entire meniscus circumference [10, 15, 16]. Directly obtained 2D radial images have been used for this purpose, but the additional scan time required with this technique is a clinical obstacle [17]. 3D models created from MRI have been used to assess circumferential meniscal position and extrusion measurements around the tibial margin; however, these methods require accurate segmentation of the tibial bone, cartilage, and menisci, which can be challenging to implement into clinical workflow [10, 15, 16].

The goal of this study, therefore, was to address the lack of a practical method for circumferential assessment of meniscal extrusion by developing a computer-assisted radial multiplanar reconstruction (rMPR)-based approach that would measure meniscal margin position around the full tibial circumference from clinically available 3D MRI. We then sought to validate the feasibility of this approach by assessing reader agreement on fixed rMPR image sets and providing preliminary comparisons of extrusion patterns across meniscal pathology groups.

2. Methods

2.1. Patient Cohort

This retrospective study of a prospective multicenter patient cohort (Level III evidence) was approved by each site’s Institutional Review Board, and all participants provided written informed consent.

We selected a convenience cohort of 20 patients from the MOON (Multicenter Orthopedic Outcomes Network) Onsite Nested Cohort [18, 19]. All study patients had undergone unilateral anterior cruciate ligament reconstruction at least 10 years earlier and were randomly selected from cases with available MRI Osteoarthritis Knee Scores (MOAKS) [20, 21], allowing for an equal representation of operative and nonoperative knees (10 of each). Bilateral knee MRI examinations were performed on 3T scanners (Siemens SkyraFit or Philips Achieva) and included a 3D fast spin-echo (SPACE or VISTA) sequence with near-isotropic resolution acquired in the sagittal plane (0.43 × 0.43 × 0.7 mm), among other sequences. The SPACE MRI studies were used for the generation of rMPR images and for all analyses; other imaging sequences were used for exploratory investigations.

2.2. rMPR Meniscal Measurement Tool

A custom MATLAB graphical user interface (GUI), the radial MPR-Meniscal Measurement Tool (rMMT), was developed to generate rMPR images (Figure 1) from a set of 3D SPACE MR images imported as DICOM files. With this GUI, standard-orientation MPR images are displayed in 3 labeled planes (axial, coronal, and sagittal) within viewing panes at the top of the interface. rMPR images, corrected using the pixel spacing for that specific radial slice, are also displayed.

FIGURE 1.

A screenshot of the radial multiplanar reconstruction (rMPR)-based meniscal measurement tool graphic user interface (GUI) showing (from left to right) the axial (window A), coronal (B), and sagittal (C) MPR images used for meniscal center selection (crosshairs in window A) and display of the rMPR slice orientation (white line with square marker, window A). The generated radial slice is displayed in window D, center bottom. The user selected measurement points at the tibial margin (red line/dot) and the peripheral inferior corner of the meniscus (green line/dot), and the distance is displayed. The measurement of 2.38 mm is displayed in the middle right of the GUI (window E). The white dots on the coronal (B), sagittal (C), and rMPR (D) images are the center point of radial image rotation.

To produce rMPR images, users set the center point of rotation on the tibial plateau for the meniscus they wish to evaluate by clicking on the axial MPR reference image (Figure 1A) or by entering desired coordinates above the axial, sagittal, and coronal MPR images (Figure 1A–C). The center point is shown on all 3 MPR images, and the positions of the sagittal and coronal slices are indicated on the axial MPR image on the GUI. The desired angular interval between rMPR images is chosen, and the rMPR images are viewed in the bottom center pane by scrolling through the generated slices. Each rMPR image is an orthogonal cross-sectional view through the meniscus at the corresponding angle. The slice position is denoted with a white line and marker relative to the axial plane in the upper left panel of the GUI (Figure 1A). Users measure meniscal position by clicking on the desired point at the margin of the tibia followed by the desired point of the periphery of the meniscus (Figure 1D). For this study, the tibial margin position was selected as the most peripheral point of the subchondral bone plate (ignoring osteophytes), and the meniscal position was selected as the peripheral, inferior corner of the meniscus. Tibial-margin-to-meniscus measurements are automatically calculated from the selected reference points as the linear distance (in mm) between these points and displayed to the right of the rMPR image on the GUI (Figure 1E). Meniscal positions peripheral to the tibia are positive, and those central to the tibial margin are negative. Once all measurements for the selected angle interval have been made across the range of interest, the results can be saved and exported as a CSV file containing the angle of the rMPR image and the linear distance recorded for each measurement and/or displayed polar coordinate plot image that is a visual representation of the tibial-meniscal distance at each measured angle viewed in the axial plane. Both the medial and lateral meniscus may be measured by choosing the appropriate center points.

2.3. Meniscal Position Measurements

For this analysis, 3 readers (a graduate student researcher without a medical background [Reader 1], a recent MD graduate [Reader 2], and a musculoskeletal radiologist with >35 years of experience [Reader 3]) who were blinded to the clinical information and the MOAKS assessment independently performed meniscal position measurements for each rMPR image using the rMMT. Before these measurements were performed, the readers were trained by measuring separate sets of cases and reviewing the image measurements from all readers to optimize uniform tibial and meniscal point selections. For all rMPR image sets, the readers used the same center points of rotation for production of rMPR images at 1° increments for each meniscus. A center point for each meniscus was chosen from the average of center point locations independently selected by readers. By using predetermined center points on a fixed set of MRI scans taken using a single imaging sequence, measurements across readers could be evaluated in a controlled setting without noise introduced by different scan protocols, variations in patient positioning, biological variability, and other forms of scan-rescan variation.

For both intrareader and interreader reliability assessments, meniscal positions measured independently by each reader at the same 3° increments over a 120° arc (from 30° to 150°, where 0° is defined as anterior and 180° is defined as posterior) were compared for 41 measurements from each meniscus. This range was selected to avoid regions near the central meniscal attachments (i.e., roots) where the less triangular meniscal shape may make the outer margin less clearly defined. For the assessment of intrareader reliability, two readers independently repeated measurements on 5 knees (10 menisci) after a 3-month washout period. For the assessment of interreader reliability, all three readers independently measured the 40 menisci from the 20 study participants.

To compare rMPR measurements with standard clinical extrusion measurements, meniscal position was measured from 2D proton density-weighted turbo spin-echo (PDW TSE) coronal sequence (0.44 × 0.44 × 3.0 mm) image sets from the same patients. Comparison measurements were performed on a standard PACS interface (AquariusNetViewer, TeraRecon) by a separate reader (musculoskeletal radiologist with 9 years of experience) using the same definitions for the tibial margin and meniscal edge. Measurements were recorded by determining the horizontal distance between the most peripheral aspect of the tibia and the most peripheral aspect of the meniscus using two standard methods for slice selection: 1) the “Bony Landmark” method (i.e., Cor-Mid), which uses a fixed point in the knee on coronal MRI slices, roughly the midbody (MB) of the meniscus; and 2) the “Coronal Slices” method (i.e., Cor-Max), in which all coronal MRI slices are inspected and measurements are performed on the slice that demonstrates the greatest extrusion [10]. Concordance between the standard clinical and rMPR measurements was evaluated by comparing the maximum absolute difference (in mm) between the two methods for the MB and maximum extrusion measurements.

2.4. Meniscal Extrusion Mapping and Subregion Analysis

Average meniscal–tibial margin distances of each meniscus were calculated by averaging reader measurements at each angle and then averaging across the angle range for five anatomic subregions: the anterior horn-body junction (AHBJ, 30°–57°), anterior body (AB, 60°–87°), MB (90°), posterior body (PB, 93°–120°), and posterior horn-body junction (PHBJ, 123°–150°). These values were then compared for each of the groups based on MOAKS meniscal subscore evaluations [21]. Agreement between rMPR measurements and coronal 2D measurements for extrusion was evaluated for all menisci. Slice indices were used to denote the subregion of maximum meniscal extrusion for 2D coronal measurements. Polar plots were produced to visualize global meniscal position relative to the tibia for the meniscal groups.

2.5. Statistical Analysis

The statistical methods used for this research are described in detail in the Supplemental Material. Intrareader and interreader measurement variabilities were evaluated using several metrics. Measurement variability (in mm) was quantified using the within-meniscus within-angle standard deviation (SD), $s_{\mathit{\text{WMWA}}}$, calculated from deviations of each reader’s measurement from the mean reader measurement for the same meniscus at the same rMPR angle. The 95% precision coefficient was calculated as $1.96\sqrt{2}{s}_{\mathit{\text{WMWA}}}$. This metric is the limit below which the absolute difference between two readers’ measurements of the same image is expected to fall 95% of the time (half the 95% limits of agreement); the same formula defines a repeatability coefficient (RC) for intrareader comparisons and a reproducibility coefficient (RDC) for interreader comparisons [22]. Intrareader and interreader agreement was also assessed with a two-way random-effects, absolute-agreement, single-measure intraclass correlation coefficient, ICC(2,1), calculated at each of the 41 rMPR angles across all 40 menisci. $s_{\mathit{\text{WMWA}}}$, the per-angle RDC, and ICC(2,1) were summarized as the mean and range across the 41 angles, with 95% confidence intervals (CIs) obtained by resampling menisci (2000 bootstrap replicates) [23]. Coverage probability plots were generated for all readers and reader pairs to compare the absolute values of intrareader and interreader differences. Bland-Altman plots were generated for all interreader pairs to determine the influence of meniscal extrusion on measurement variability. The rMPR measurements were compared to 2D coronal measurements for overall concordance by evaluating the number of cases for which measurements fell within the interreader RDC (1.4 mm) at the MB. Concordance for classification was evaluated using a 3.0-mm threshold for clinical grade meniscal extrusion, treating the 2D coronal classification as the comparator.

3. Results

3.1. Patient Cohort Characteristics

Of the 20 patients, 11 were women. The mean patient age was 36 years (range, 27–46 years), and the mean body mass index was 27 (range, 19.6–54.8). The MOAKS characteristics of the 40 menisci are shown in Table 1.

TABLE 1.

Distribution of MOAKS meniscal subscore features for the 40 menisci in the 20 study participants.

MOAKS and meniscal injury classification	Meniscus involved
	All (N = 40)	Medial (n = 20)	Lateral (n = 20)
Tear
No tear	25	13	12
Tear	15	7	8
Extrusion
No extrusion	29	13	16
Extrusion >2 mm	11	7	4
Tear + extrusion
No tear, no extrusion	20	9	11
No tear, extrusion >2 mm	5	4	1
Tear, no extrusion	9	4	5
Tear, extrusion >2 mm	6	3	3

Abbreviation: MOAKS, MRI Osteoarthritis Knee Score.

3.2. Generation of rMPR Images

The rMMT program produced rMPR images at all angles for the 40 menisci. On rMPR images, the menisci appeared triangular throughout the circumference of the tibial margin (Figures 2, 3), avoiding the oblique cross-sectional meniscal appearance commonly seen at the junctions of the body with the anterior and posterior horns on standard coronal and sagittal MR images (Figure 4).

FIGURE 2.

Medial meniscus radial multiplanar reconstruction images reconstructed every 20° from the anterior root to the posterior root. Because the image slice direction is orthogonal to the tibial rim, the meniscal contour remains triangular throughout its circumference. The arrow points to the posterior horn root attachment. Asterisk: subcortical cyst beneath root attachment; PCL: posterior cruciate ligament.

FIGURE 3.

Lateral meniscus radial multiplanar reconstruction images reconstructed every 20° from anterior (0°) to posterior (220°). The white arrows point to the popliteus tendon as it passes peripheral to the meniscus. Note the sloping of the margin of the tibial plateau in this region (black arrows).

FIGURE 4.

Comparison of medial meniscal appearance in radial multiplanar reconstruction (rMPR) and standard imaging planes. The solid white line in axial reference image (A) shows the 135° slice plane of the rMPR image shown in (B). The dotted white lines in (A) show the slice planes of the corresponding standard coronal (C) and sagittal (D) images for the same posteromedial location of the meniscus, which is indicated as the crossing point of the lines on the reference image in (A). The white circle in (A) is the center point of rotation for the rMPR images.

3.3. Meniscal Position Measurement Reliability

All readers made measurements at all 41 angles for the 40 menisci (1640 measurements per reader). The overall interreader reliability statistics are presented in Table 2. The mean 95% RDC across the 41 angles was 1.4 mm (SD 0.33) for all menisci, 1.4 mm (SD 0.47) for medial menisci, and 1.3 mm (SD 0.26) for lateral menisci; for comparison, the mean voxel dimension for the rMPR images was 0.64 mm. The 95% RDC varied with angle, even for nonextruded, nontorn menisci, with the greatest values in the posterolateral corner of the lateral meniscus at angles greater than 130°, which is the region of the popliteus hiatus. Across all readers, the mean per-angle ICC(2,1) was 0.88 (SD 0.05) for all menisci, with similar values for the medial (0.88; SD 0.06) and lateral (0.85; SD 0.07) menisci.

TABLE 2.

Summary statistics of interreader reliability for measurements of tibial-meniscal edge distance (extrusion) on rMPR images.

Meniscus group	95% RDC (mm) (41 angles, 3 readers)		Within-angle ICC (41 angles, 3 readers)
	Mean (SD) [95% CI]	Range	Mean (SD) [95% CI]	Range
All menisci (n = 40)	1.4 (0.33) [1.1–1.7]	1.0–2.0	0.88 (0.05) [0.78–0.91]	0.79–0.95
Medial menisci (n = 20)	1.4 (0.47) [1.0–1.9]	0.81–2.3	0.88 (0.06) [0.69–0.91]	0.78–0.96
Lateral menisci (n = 20)	1.3 (0.26) [1.0–1.5]	0.82–1.8	0.85 (0.07) [0.73–0.90]	0.69–0.93

For comparison, the mean voxel dimension of the radial images was 0.64 mm (range 0.54–0.70 mm). Abbreviations: rMPR, radial multiplanar reconstruction; RDC, reproducibility coefficient; ICC, intraclass correlation; SD, standard deviation; CI, confidence interval.

The overall intrareader repeatability was excellent, with a mean 95% RC of 0.56 mm. Intrareader 95% RC values were 0.46 to 0.48 mm for Reader 1 and 0.55 to 0.73 mm for Reader 2, with per-angle ICC(2,1) values of 0.90 to 0.97 and 0.82 to 0.92, respectively (see Supplemental Table 1 and Supplemental Figure 1).

Comparisons of rMPR measurements with traditional clinical measurements taken from 2D coronal MRI are shown in Table 3. The rMPR measurements demonstrated excellent concordance with standard measurements when compared at the same point on the mid-coronal image. Overall, 98% of cases agreed within the interreader 95% RDC value of 1.4 mm, with 95% agreement for medial and 100% agreement for lateral within this range. Because the extrusion measurements for the Cor-Max method and rMPR images were not necessarily made at the same point on the meniscal circumference, the absolute values are not expected to be identical. The Cor-Max and maximum rMPR measurements agreed within 1.4 mm for 82% of cases, with 85% agreement for medial and 80% agreement for lateral. Considering meniscal extrusion as a binary classification in which “clinically extruded” is classified as a maximum measurement greater than or equal to 3 mm, rMPR and both 2D coronal assessments demonstrated good concordance, with 95% agreement for medial and lateral menisci.

TABLE 3.

Comparison of 2D coronal measurements with rMPR measurements.

Comparison	n	Mean absolute difference (mm)	SD (mm)	Maximum absolute difference (mm)	Within 1.4 mm	Agreement for extruded (≥3 mm)/nonextruded
Cor-Mid vs rMPR-Mid
Overall	40	0.5	0.6	1.5	39/40 (98%)	39/40 (98%)
Medial	20	0.7	0.7	1.5	19/20 (95%)	19/20 (95%)
Lateral	20	0.3	0.4	1.0	20/20 (100%)	20/20 (100%)
Cor-Mid vs rMPR-Max
Overall	40	0.7	0.8	2.5	34/40 (85%)	38/40 (95%)
Medial	20	0.8	0.9	2.5	18/20 (90%)	19/20 (95%)
Lateral	20	0.6	0.7	1.7	16/20 (80%)	19/20 (95%)
Cor-Max vs rMPR-Max
Overall	40	0.7	0.9	2.4	33/40 (82%)	38/40 (95%)
Medial	20	0.8	1.0	2.4	17/20 (85%)	19/20 (95%)
Lateral	20	0.6	0.7	1.7	16/20 (80%)	19/20 (95%)

Cor-Mid and rMPR-Mid represent the measurements made at the midbody slice for coronal and 90° for rMPR, respectively. Cor-Max and rMPR-Max represent the maximum measurements made across the entire meniscal range. Abbreviations: rMPR, radial multiplanar reconstruction; SD, standard deviation.

3.4. Meniscal Extrusion Mapping and Subregion Analysis

The coverage probability plots for absolute measurement differences among all readers and between each reader pair are shown in Figure 5. For all reader pairs, there was greater interreader variation in meniscal position measurements when meniscal extrusion exceeded a mean measurement of 5 mm (Figure 6).

FIGURE 5.

Interreader variability. The coverage probability plot shows the cumulative proportion (y-axis) of absolute differences between readers’ measurements (x-axis) for all reader pairs and overall. RDC: reproducibility coefficient.

FIGURE 6.

Bland-Altman plot of measurement differences between reader pairs (y-axis) and mean meniscal position measurement (x-axis) for all measurements shows greater interreader variation for extrusion greater than 5 mm, indicating that interreader variability increases when extrusion is larger.

Table 4 shows the average meniscal position measurements for the medial and lateral meniscus across the defined subregions. Extrusion maps, i.e., polar coordinate plots, for all menisci and each MOAKS-based meniscal group show the variations in meniscal position for both the degree and location of meniscal extrusion (Figure 7). For nontorn, nonextruded lateral menisci, the mean positions of the AHBJ and PHBJ were central to the meniscal edge. By comparison, for the nontorn medial menisci, the mean position of the AHBJ was 1 mm beyond the tibial margin. The torn and extruded medial meniscus group showed the greatest extrusion at the junction of the body and posterior horn, whereas the torn and extruded lateral meniscus group had greater extrusion at the AB region. The degree of extrusion was shown to be comparable to traditional 2D measurement methods in the MB (Figure 8).

TABLE 4.

Meniscal-tibial margin distance (in mm) measured on rMPR images across different meniscal regions.

Meniscal region	MOAKS group	n	AHBJ 30°–57°	AB 60°–87°	MB 90°	PB 93°–120°	PHBJ 123°–150°	All 30°–150°
Medial	All	20	1.16 (0.98)	1.12 (1.10)	1.08 (1.11)	1.08 (1.76)	1.06 (1.95)	1.10 (1.44)
	No tear	13	0.99 (0.93)	0.85 (0.94)	0.76 (0.89)	0.68 (1.02)	0.57 (1.13)	0.77 (1.00)
	Tear	7	1.46 (1.04)	1.63 (1.35)	1.66 (1.43)	1.83 (2.61)	1.97 (2.91)	1.72 (1.97)
	No extrusion	13	0.48 (0.90)	0.43 (0.82)	0.40 (0.99)	0.41 (1.08)	0.33 (1.14)	0.41 (0.98)
	Extrusion	7	2.41 (1.10)	2.41 (1.49)	2.33 (1.29)	2.33 (2.57)	2.41 (2.91)	2.39 (2.00)
Lateral	All	20	−0.28 (1.27)	0.31 (1.03)	0.36 (1.15)	0.34 (1.22)	0.02 (1.61)	0.10 (1.28)
	No tear	12	−0.66 (1.28)	−0.06 (0.92)	−0.03 (0.97)	−0.03 (1.22)	−0.50 (1.73)	−0.30 (1.28)
	Tear	8	0.28 (1.26)	0.86 (1.17)	0.94 (1.39)	0.91 (1.21)	0.79 (1.41)	0.72 (1.26)
	No extrusion	16	−0.63 (1.26)	−0.09 (0.97)	−0.02 (0.96)	0.01 (0.97)	−0.23 (1.46)	−0.23 (1.16)
	Extrusion	4	1.09 (1.31)	1.91 (1.16)	1.89 (1.72)	1.69 (1.90)	0.99 (2.07)	1.43 (1.61)

Values are mean (95% reproducibility coefficient). Abbreviations: rMPR, radial multiplanar reconstruction; AHBJ, anterior horn-body junction; AB, anterior body; MB, midbody; PB, posterior body; PHBJ, posterior horn-body junction.

FIGURE 7.

Mean tibial margin to meniscal distances. Polar coordinate plot displaying the mean meniscal position measurement (in mm) for angles from 30° to 150° for medial (left) and lateral (right) menisci overall and for meniscal groups based on MRI Osteoarthritis Knee Scores (MOAKS). The black semicircles represent the position of the tibial margin.

FIGURE 8.

The output of the radial multiplanar reconstruction (rMPR) program and comparable coronal 2D slices displaying examples of high midbody (MB) concordance, supporting the current MB standard-of-care approach. Discordance increases away from the MB because of the geometric limitations of 2D coronal imaging on a curved meniscal surface, introducing both true circumferential variation in the rMPR image and geometric distortion in 2D coronal measurements. The coronal plane is denoted by yellow dotted lines and the radial plane is denoted with white arrows on the axial view (left column). Tibia and meniscus reference points selected for measurements are marked with red and green lines, respectively. Extruded (>3 mm) measurements are in red. The top row shows the MB (90°); the bottom row shows an anterior body (AB) location (63°), where the rMPR and 2D coronal measurements diverge. PDW TSE: proton density-weighted turbo spin-echo.

3.5. Discordant Case Investigation

One notable difficulty in measurement occurred with a lateral meniscus believed to have no extrusion based on the MOAKS reading but measured as extruded at the PHBJ region on rMPR images. This was determined to be due to poor discrimination of the popliteus tendon from the meniscus on the SPACE rMPR images because of low image contrast (Figure 9). There was similar poor image contrast between these structures on the PDW TSE images. Sagittally acquired 3D double-echo steady-state sequence images were available for this case; these images more clearly differentiated the structures because of higher fluid signal.

FIGURE 9.

For one patient, the SPACE radial multiplanar reconstruction (rMPR) images (left column) showed poor contrast between the meniscus (yellow arrows) and the popliteus tendon (blue arrows), which could be falsely measured as meniscal extrusion at the PHBJ. The brighter fluid contrast of the double-echo steady-state (DESS) rMPR images (middle column) allowed differentiation of the tendon (blue arrows) from the periphery of the meniscus (yellow arrows). For comparison, standard coronal proton density-weighted turbo spin-echo (PDW TSE) 2D images with fat suppression (right column) showed the popliteus tendon (blue arrows) as it passes through the popliteus hiatus. Of note, the PDW TSE images demonstrated poor contrast between these structures.

4. Discussion

This study establishes a framework for circumferential mapping of meniscal extrusion from clinically available 3D MRI. We have created a new method for circumferential evaluation of menisci that greatly reduces the partial volume averaging limitations of sagittal and coronal image planes. This method also allows mapping of meniscal position for the assessment of meniscal extrusion throughout the meniscal circumference. Beyond demonstrating feasibility and reader agreement, this work provides the analytic scaffolding needed to move meniscal extrusion assessment from a single-slice measurement to a regional, pattern-based characterization of meniscal dysfunction. As segmentation and meniscal analysis pipelines become increasingly automated, frameworks of this kind will be essential for translating the resulting data into clinically meaningful understanding, including identifying which patterns of extrusion predict OA development, which reflect benign anatomic variations, and which are amenable to targeted intervention. Given the central role of the meniscus in knee biomechanics and the increasing recognition that specific patterns of meniscal dysfunction, not just global extrusion magnitude, drive downstream joint degeneration, tools that can faithfully resolve these patterns are a prerequisite for evidence-based meniscal care.

4.1. Overall Significance and Usability of rMPR-Based Meniscal Measurement Tool

Previous studies have proposed various MRI postprocessing techniques for more comprehensive measurement of meniscal position, tibial coverage, and meniscal shape; however, these methods require segmentation of the menisci, tibia, and femur [15, 16]. Such methods are time consuming and challenging to automate, particularly given the very small distances being measured, the typical volume resolution of 3D MRI, and the current accuracy of automated segmentation techniques. The rMMT program we have developed rapidly produces rMPR images, which may be subsequently evaluated visually or with extrusion measurements at any desired incremental angles or meniscal position. As only limited user inputs are required to evaluate meniscal extrusion or displacement using this tool, there is no need for advanced modeling or segmentation. The rMPR method can be employed on-the-fly with minimal preprocessing or software requirements, allowing for easier clinical integration.

In addition, while we used a 3D SPACE acquisition for this study, the rMMT program can be applied to any 3D DICOM data set whether acquired by MRI or by another modality such as computed tomography. Data from this program can be exported in many formats for use in statistical analysis, as well as for viewing in PACS. Although the primary purpose of this program is the visual assessment of meniscal extrusion, rMPR images can be used to evaluate any anatomical imaging features of interest; there are therefore numerous potential clinical applications. Readers who performed the measurements in this study varied considerably in training levels and clinical familiarity and included a musculoskeletal radiologist with decades of clinical and research experience, a physician in training, and a graduate student researcher. Despite the substantial variability in experience, the interreader agreement for meniscal position and extrusion mapping was excellent, demonstrating strong concordance between measurements. These findings support the feasibility of rMPR meniscal extrusion mapping for clinical and research applications, although additional studies are needed to evaluate performance across repeat imaging and varied acquisition conditions.

4.2. Meniscal Position Measurement Reliability

In this study, we found that meniscus extrusion position and extrusion mapping using rMPR images was reliable and consistent across readers when using the same fixed rMPR image sets. There was excellent concordance between extrusion measurements for rMPR and traditional clinical measurements taken at the same point on mid-coronal 2D images. While traditional clinical methods measure meniscal extrusion on a single coronal MRI slice, research has shown that the meniscus may appear more extruded on different coronal slice positions [10]. Measuring extrusion using rMPR images provides a more comprehensive and geometrically accurate method by orthogonally sampling multiple planes around the meniscus. The precision of rMPR-based extrusion measurements averaged over all angles was less than 1.5 mm, roughly 2 voxel widths, for both medial and lateral menisci; most studies have reported extrusion measurements in 1- to 2-mm increments [20, 21, 24]. The precision is likely higher for nonextruded menisci, as reader agreement was lower in regions where meniscal extrusion exceeded 5 mm, an overtly pathological state. This method showed good concordance with traditional 2D coronal measurements taken at the MB, where the coronal plane best approximates the local meniscal geometry. Away from the MB, discordance increased, likely reflecting both true circumferential variation captured by rMPR and geometric distortion introduced when a fixed 2D coronal plane is used to measure extrusion along a curved meniscal surface.

4.3. Subregion Analysis

Although only a limited number of menisci were evaluated, the rMPR-based extrusion mapping showed differences in meniscus extrusion patterns across several groups of meniscal pathology and nontorn menisci. Nontorn, nonextruded lateral menisci typically presented as slightly “intruded” compared to the tibial plateau, with the greatest degree of extrusion occurring in the PB region. Nontorn, nonextruded medial menisci typically did not exceed 0.5 mm past the tibial plateau and showed the greatest displacement in the anterior subregions. For the medial meniscus, there was substantially more meniscal extrusion in the torn meniscus group than in the nontorn group. Interestingly, the most severe pathologic extrusion in the torn meniscus group was in the region of the PHBJ rather than the MB, where extrusion is most often measured. In contrast, the extruded but nontorn medial meniscus cohort demonstrated maximal extrusion at the AHBJ and demonstrated mean extrusion only slightly greater than 2 mm. This suggests that in the medial meniscus, extrusion may be more pathologic at the PHBJ and may be more of a developmental variant at the AHBJ. Additionally, given that the greatest region of extrusion is not at the mid-coronal slice, traditional clinical methods may underestimate the degree of displacement. For the lateral meniscus, mean extrusion was greatest in the AB region for the torn and extruded group. Further comprehensive regional evaluation of meniscal pathology as it relates to meniscal position and extrusion mapping should provide insight regarding appropriate thresholds for extrusion in different positions. Such assessments may lead to better prognostication of OA development and may help to identify which meniscal pathologies would benefit from interventions such as meniscal centralization or partial meniscectomy.

4.5. Limitations

Our study had several limitations. The readers performed measurements on the same fixed rMPR image datasets; therefore, this study primarily validates the feasibility of the circumferential mapping tool and quantifies reader-related measurement variability as a measure of reproducibility across readers, rather than full scan–rescan reproducibility. Future studies will evaluate true reproducibility across broader acquisition conditions, including repeat imaging with patient repositioning, different scan sessions, and potentially different scanners or sites. Our cohort consisted of relatively younger patients with a history of anterior cruciate ligament reconstruction, which does not represent the broader population, limiting the generalizability of our findings. The sample size was relatively small and did not include all types of meniscal pathology. Therefore, the comparisons among the MOAKS groups lack statistical power and require further investigation and validation. As with most MRI studies of extrusion, the images were acquired using a standard clinical method in which the knee is non-weightbearing in near full extension. Meniscal position changes with weight bearing during flexion-extension, which has been shown to affect meniscal extrusion measurements; non-weightbearing imaging may underestimate extrusion when compared to functional loading [25, 26].

4.6. Future Extensions

In future work, we aim to automate this entire process, thus removing the need for manual point selection. This would allow the program to be provided as part of the clinical images and for analysis of large datasets without manual input, serving as a support tool for patient care and the development of risk models for OA prediction, surgical outcomes, or joint degeneration metrics. More broadly, as direct segmentation of menisci and tibia continues to improve, rMPR-based frameworks provide a natural analytic layer on top of segmentation, translating pixel-level data into the regional, pattern-based features most relevant to meniscal function and OA prognostication.

Integrating rMPR mapping with automated segmentation, longitudinal imaging, and biomechanical modeling has the potential to move meniscal assessment from a binary threshold (extruded/not extruded) toward a phenotyped characterization of meniscal dysfunction that can guide interventions such as repair, transplant, and centralization, perhaps directing the surgeon to the meniscal region requiring attention. rMPR imaging may also help with the evaluation of meniscal tears. Although conventional imaging planes provide reasonably detailed information for grading, radial reconstructions may offer enhanced characterization of tear morphology, including more precise assessment of tear type, length, width, and degree of circumferential fiber disruption, potentially enabling targeted surgical intervention and identification of meaningful patterns of injury that allow earlier detection of OA [27]. When combined with postoperative outcome data, this approach may also identify meniscal tear patterns associated with limited benefit from partial meniscectomy.

5. Conclusion

We found that rMPR images generated from 3D knee MRI provide a reader-consistent method for both visual assessment and quantification of meniscal extrusion throughout the entire meniscus. This method demonstrated regional variations in meniscal margin position that would be difficult to visualize with standard MRI planes. As automated analysis of meniscal extrusion becomes more clinically accessible, frameworks such as this will be critical for understanding the patterns of meniscal dysfunction that precede and drive OA and for identifying patients most likely to benefit from intervention. Further investigation in larger patient cohorts is required to determine whether circumferential measurement of meniscal extrusion can stratify a patient’s likelihood of knee OA development and progression.

Statement of Clinical Significance:

Meniscus extrusion circumferential mapping using radially oriented multiplanar reconstruction images is feasible and consistent across readers and allows a more complete assessment of meniscal position, which may be valuable for further investigation of meniscal injury and osteoarthritis.