Knee Bone Age using MRI: Validation of a Novel Method to Reduce Hand Bone Age Radiographs

Abstract

Objectives:

Determining bone age in skeletally immature patients is critical for proper management and surgical planning. Pennock et al. recently created and validated a bone age atlas using the ossification pattern of the knee on MRI in pediatric patients, obfuscating the need for a hand radiograph and its associated cost, radiation exposure, and clinical inefficiency. This aim of this study was to create and validate a shorthand method of bone age determination using knee MRI.

Methods:

We identified patients who underwent both knee MRI and hand bone age radiograph within a ninety-day period. From the findings of Pennock et al., a stepwise algorithm for predicting bone age on knee MRI was developed for males and females separately, in collaboration with a pediatric musculoskeletal radiologist. Six raters at varying levels of training used the algorithm to assign a bone age to each knee MRI. Intraclass correlation coefficient (ICC) was used to compare each rater’s predicted knee bone age to the Greulich and Pyle (G&P) hand bone age and validate the shorthand algorithm. Inter- (two-way random, absolute agreement) and intra-rater (two-way mixed, absolute agreement ) reliability were also calculated using ICC.

Results:

Thirty-eight patients (44.7% females) underwent a knee MRI at a mean age of 12.8 years (range 9.3-15.7). The mean time between hand bone age x-ray and knee MRI was 20.2 days (range 0-88). The inter-rater reliability for the application of our shorthand algorithm was 0.81 (95% CI: 0.72 – 0.88), indicating good inter-observer agreement. The shorthand algorithm was also shown to be a reliable predictor of hand bone age across level of training, as medical students (ICC 0.77, 95% CI 0.60-0.88), residents (ICC 0.80, 95% CI 0.65-0.89), and attending physicians (ICC 0.80, 95% CI 0.63-0.89) all achieved strong correlation between predicted knee bone age and G&P hand bone age. All respondents also demonstrated consistency, with intra-rater ICCs between 0.85-0.93.

Conclusion:

This shorthand algorithm is a consistent, reliable and valid way to determine skeletal maturity using knee MRI in patients aged 10 to 15 and can be utilized across different levels of orthopaedic and radiographic expertise. We believe this method is readily applicable in a clinical setting and may reduce the need for routine hand bone age radiographs.

Figure 1.

Table 1.

Correlation between knee MRI bone age and hand radiograph bone age (G&P) for each rater, as determined using intraclass correlation coefficients (ICC). Note: Intra-rater reliability has not been competed for “Resident 2” at this time.

	Validity of MRI Knee Bone Age and Hand Bone Age		Intra-Rater Reliability
Rater	ICC	95% CI	ICC	95% CI
Student 1	0.75	0.57-0.86	0.93	0.79-0.97
Student 2	0.73	0.53-0.86	0.93	0.87-0.97
Resident 1	0.73	0.54-0.85	0.85	0.73-0.92
Resident 2	0.78	0.60-0.88	-	-
Attending 1	0.74	0.54-0.85	0.86	0.70-0.93
Attending 2	0.80	0.63-0.89	0.86	0.72-0.93