The anterior cruciate ligament (ACL) is the principal restraint to anterior tibial translation and contributes to rotational stability of the knee. It is the most frequently injured knee ligament, typically through non-contact pivoting or deceleration in sport, and injury incidence differs by sex, age, sport and level of participation.(1) Prompt and accurate diagnosis allows timely rehabilitation or reconstruction and reduces the risk of secondary meniscal and chondral damage; a systematic review of acute ACL injury pathways found that delays to diagnosis are common and that reliance on specialist assessment and advanced imaging lengthens the route to a definitive diagnosis.(2)

Clinical assessment rests on the history and on provocative tests such as the Lachman, anterior drawer and pivot shift tests. Their accuracy depends on examiner experience, pain, muscle guarding and the time since injury; a recent meta-analysis reported pooled sensitivities of only 59–76% for these three tests.(3) MRI is therefore the imaging investigation of choice, with high accuracy against arthroscopy,(4,5) and it also depicts associated meniscal, ligamentous and osteochondral injury.(6) However, MRI is costly, often involves long waiting lists and may be contraindicated by claustrophobia or metallic implants. In Pakistan and other resource-limited settings, access to MRI is frequently the bottleneck in the diagnostic pathway.

Ultrasonography is quick, inexpensive, portable, free of ionising radiation and permits dynamic examination. Advances in high-frequency transducers and better understanding of knee sonoanatomy now allow direct and indirect assessment of the ACL.(7) Meta-analyses have reported pooled sensitivities of 84–90% and specificities of 92–97% for ultrasonography in ACL injury,(8–10) but the included studies vary widely in technique, reference standard and patient spectrum, and local evidence comparing ultrasonography with MRI is limited.(11)

The aim of this study was to determine the diagnostic accuracy of ultrasonography for ACL tear in adults with clinically suspected ACL injury, using MRI as the reference standard.

Study design and setting This cross-sectional diagnostic accuracy study was conducted in the Department of Radiology, King Edward Medical University / Mayo Hospital, Lahore, Pakistan, over a six-month period January 2025 to June 2025. It is reported in accordance with the Standards for Reporting of Diagnostic Accuracy Studies (STARD) 2015.(12) Ethical approval The study was approved by the Institutional Review Board of King Edward Medical University, Lahore. Written informed consent was obtained from all participants. The study was conducted in accordance with the Declaration of Helsinki. Participants Patients referred from the Departments of Orthopaedics and Radiology were enrolled by non-probability convenience sampling. Inclusion criteria were age over 18 years, either sex, a history of knee trauma and at least one positive clinical test for anterior or rotational instability (Lachman, anterior drawer or pivot shift test). Exclusion criteria were injury within the preceding four weeks, a positive posterior drawer test, multidirectional instability, previous knee surgery or infection, open knee wounds and any pre-existing knee deformity. All enrolled patients underwent both the index test and the reference standard. Clinical assessment All patients underwent history taking and physical examination. The anterior drawer test was considered positive when anterior tibial translation exceeded 6 mm relative to the contralateral knee; the Lachman test was positive when anterior tibial translation was increased with a soft or absent end point; and the pivot shift test was positive when a sudden shift or clunk was felt as the tibia reduced on the femur. Index test: ultrasonography Ultrasonography was performed with a MyLab Seven system (Esaote, Genoa, Italy) and an 8–12 MHz linear transducer using an anterior approach with the knee flexed. The normal ACL was identified in the longitudinal plane as a thick, hyperechoic, fibrillar band extending from the lateral femoral condyle to the tibial plateau. Abnormalities were graded as grade 1 (hypoechoic area within an otherwise continuous ligament), grade 2 (partial discontinuity of fibres) or grade 3 (non-visualisation of the ligament or abnormal angulation). Any grade was classified as positive for ACL injury. Ultrasonography was performed and interpreted by an experienced radiologist before MRI and without knowledge of the MRI findings. Reference standard: MRI MRI was performed on a 1.5 T scanner (SIGNA Voyager, GE Healthcare, Milwaukee, WI, USA) using the departmental knee protocol, comprising multiplanar sagittal, coronal and axial sequences with a small field of view (14–16 cm), 3 mm slice thickness and a high matrix. ACL injury on MRI was graded as grade 1 (intrasubstance hyperintense signal with intact fibres), grade 2 (partial fibre disruption) or grade 3 (complete fibre disruption). Any grade was classified as positive. MRI examinations were reported by an independent radiologist with five years’ experience in musculoskeletal radiology who was blinded to the ultrasonography findings and to the clinical examination results. Sample size The sample size was calculated with Buderer’s method(13) using an expected sensitivity of 92.6% and specificity of 98% for ultrasonography,(11) an expected prevalence of ACL tear of 30%, an absolute precision of 10% and a 5% significance level. The sensitivity-based estimate (88 patients) exceeded the specificity-based estimate, so 88 patients were enrolled. Statistical analysis Data were analysed with IBM SPSS Statistics version 26 (IBM Corp., Armonk, NY, USA). Age is reported as mean ± standard deviation (SD) and range, and categorical variables as frequencies and percentages. Ultrasonography results were cross-tabulated against MRI in a 2 × 2 table to calculate sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive and negative likelihood ratios (LR+ and LR−) and overall diagnostic accuracy [(TP + TN)/(TP + FP + FN + TN)]. Ninety-five per cent CIs were calculated with the Wilson score method for proportions and the log method for likelihood ratios. Agreement between the two modalities was quantified with Cohen’s kappa and interpreted according to Landis and Koch.(14) The association between ultrasonography and MRI findings was tested with Fisher’s exact test because one expected cell count was below five. A two-sided P value < 0.05 was considered statistically significant.

Participants

Eighty-eight patients were enrolled, and all underwent both ultrasonography and MRI; there were no indeterminate results and no exclusions after enrolment (Figure 1). Mean age was 37.0 ± 15.0 years (range 18–77), 51 (58.0%) patients were male and 70 (79.5%) lived in urban areas. All patients had a history of knee trauma and positive Lachman, anterior drawer and pivot shift tests (Table 1).

Figure 1. Flow of participants through the study (STARD diagram). MRI, magnetic resonance imaging.

Table 1. Baseline characteristics of the study participants (n = 88)

SD, standard deviation.

Ultrasonography and MRI findings

MRI showed ACL injury in 74 (84.1%) patients and a normal ACL in 14 (15.9%). Ultrasonography was positive in 71 (80.7%) and negative in 17 (19.3%). The distribution of injury grades was similar on the two modalities, with grade 2 and grade 3 injuries predominating (Table 2).

Table 2. Grade of anterior cruciate ligament injury on MRI and ultrasonography among positive cases

MRI grading: grade 1, intrasubstance hyperintense signal with intact fibres; grade 2, partial fibre disruption; grade 3, complete fibre disruption. Ultrasonography grading: grade 1, hypoechoic area within the ligament; grade 2, partial discontinuity of fibres; grade 3, non-visualisation or abnormal angulation. The ultrasonography-positive group includes two false-positive cases. MRI, magnetic resonance imaging.

Diagnostic accuracy

Of the 71 ultrasonography-positive patients, 69 had ACL injury on MRI (true positives) and 2 did not (false positives). Of the 17 ultrasonography-negative patients, 12 had a normal ACL on MRI (true negatives) and 5 had ACL injury (false negatives) (Table 3; Figure 1). Ultrasonography therefore had a sensitivity of 93.2% (95% CI 85.1–97.1), specificity of 85.7% (60.1–96.0), PPV of 97.2% (90.3–99.2), NPV of 70.6% (46.9–86.7) and overall accuracy of 92.0% (84.5–96.1). LR+ was 6.53 (1.81–23.58) and LR− was 0.08 (0.03–0.19). Agreement between ultrasonography and MRI was substantial (κ = 0.73, 95% CI 0.53–0.92), and the association between the two was statistically significant (P < 0.001) (Table 4). Figure 2 places these estimates alongside previously published studies.

Table 3. Cross-tabulation of ultrasonography against MRI for anterior cruciate ligament injury (n = 88)

Fisher’s exact test, P < 0.001. FN, false negative; FP, false positive; MRI, magnetic resonance imaging; TN, true negative; TP, true positive.

Table 4. Diagnostic performance of ultrasonography for anterior cruciate ligament injury, with MRI as the reference standard

Confidence intervals for proportions were calculated with the Wilson score method and for likelihood ratios with the log method. Predictive values depend on the prevalence of ACL injury in this cohort (84.1%). CI, confidence interval; MRI, magnetic resonance imaging.

Figure 2. Sensitivity and specificity of ultrasonography for anterior cruciate ligament injury in the present study compared with published meta-analyses and primary studies.(8–10,15,16,18,20) The square shows the present study and circles show published estimates; horizontal lines are 95% confidence intervals where reported (Wilson score intervals for the present study). Breukers et al. values are for complete tears.

In this study of adults with clinically suspected ACL injury, ultrasonography detected 93% of the ACL tears identified on MRI, was correct in 92% of patients overall and showed substantial agreement with MRI. A positive scan was highly reliable (PPV 97%), whereas a negative scan was less informative (NPV 71%). Our sensitivity is at the upper end of previously reported values, while our specificity is somewhat lower (Figure 2). Wang et al. pooled 90% sensitivity and 97% specificity,(8) Lee and Yun reported 88% and 96%,(9) and the most recent meta-analysis of 12 studies, by Schreiner et al., reported 84% and 92%, with marked heterogeneity in sensitivity.(10) The lower and imprecise specificity in our series reflects the small number of MRI-negative patients (n = 14): two false-positive scans reduced specificity by 14 percentage points, and the 95% CI (60–96%) overlaps the pooled estimates. Similarly, the high PPV and modest NPV are largely a consequence of the high prevalence of ACL injury (84%) in a cohort selected for positive clinical tests, and they should not be extrapolated to lower-prevalence settings such as emergency departments. The likelihood ratios, which are independent of prevalence, indicate that a negative ultrasonography result (LR− 0.08) is more useful for lowering the probability of a tear than a positive result is for raising it. Diagnostic performance varies with technique. Studies using dynamic or functional ultrasonography, which measure anterior tibial translation relative to the contralateral knee, have reported sensitivities of 82–95% and specificities of 89–96% where reported,(15–17) whereas a large arthroscopically verified series of dynamic ultrasonography found good performance for complete tears but poor sensitivity for partial tears.(18) Point-of-care knee ultrasonography in the emergency department has also shown high sensitivity and specificity for cruciate ligament tears.(19) In contrast, Barzin et al. reported a sensitivity of only 65% with 100% specificity,(20) and Mir et al. found diagnostic accuracies of 83–95% across menisco-ligamentous injuries, supporting ultrasonography as a screening tool in low-resource settings.(21) Earlier work relied mainly on indirect signs, which are more evident after acute injury.(22) Because our cohort excluded injuries less than four weeks old, our results largely reflect direct visualisation of the distal ACL through the anterior approach. The proximal and mid-substance ligament, where most tears occur, is difficult to see anteriorly; combining anterior and posterior approaches has recently been shown to allow visualisation of the full ligament length,(23) and experimental and arthroscopic work supports the value of standardised sonographic assessment.(7,24) Adding a posterior approach and dynamic testing may reduce false-negative results such as the five observed in this study. Our patients were older (mean 37 years) and more often female (42%) than in a recent Indian reconstruction cohort (mean age 28 years; 91% male).(25) This probably reflects inclusion of non-sporting trauma in a general hospital population rather than a purely athletic or surgical cohort, and it broadens the applicability of our findings. Notably, all 88 patients had positive Lachman, anterior drawer and pivot shift tests, yet 14 had an intact ACL on MRI, consistent with the limited specificity of clinical examination alone(3) and underlining the value of an objective imaging test. From a practical standpoint, ultrasonography can be performed at the first consultation and could shorten the diagnostic pathway,(2) particularly where MRI is scarce. A positive scan in a patient with clinical instability supports early referral for orthopaedic management, while a negative or equivocal scan with persistent clinical suspicion warrants MRI, which also evaluates associated meniscal and chondral injury.(6) Strengths and limitations Strengths of this study include blinded, independent interpretation of the index test and reference standard, performance of both tests in every enrolled patient and reporting in line with STARD 2015. Several limitations should be considered. First, MRI rather than arthroscopy was the reference standard; although MRI is highly accurate,(4,5) its sensitivity for partial tears is lower,(6) so some misclassification is possible, particularly for grade 1 injuries. Second, this was a single-centre study with convenience sampling and a high prevalence of ACL injury, which limits generalisability and gave imprecise estimates of specificity and NPV. Third, ultrasonography was performed by a single operator, and inter-observer reliability was not assessed; ultrasonography is operator dependent. Fourth, agreement on injury grade could not be assessed because grades were not cross-tabulated by patient, and associated meniscal, collateral ligament and cartilage injuries were not evaluated. Finally, acute injuries (< 4 weeks) were excluded, so the findings may not apply to the acute setting. Larger multicentre studies with arthroscopic verification, multiple operators and dynamic techniques are needed.

Ultrasonography showed high sensitivity (93.2%), good overall accuracy (92.0%) and substantial agreement with MRI for the detection of ACL tear in adults with clinically suspected injury. It is a valuable, non-invasive and accessible first-line imaging option, especially where MRI is not readily available. Because a negative scan does not reliably exclude a tear, MRI should be obtained when ultrasonography is negative or equivocal and clinical suspicion persists.