Simplified radius, ulna, and short bone-age assessment procedure using grouped-Tanner-Whitehouse method

Area of Science:

• Pediatric endocrinology
• Radiology
• Data mining

Background:

• The Tanner-Whitehouse III (TW3) method is a standard for assessing bone age in children.
• The TW3 method is recognized as time-consuming in clinical practice.
• A grouped-TW algorithm (GTA) was developed to streamline bone age assessment.

Purpose of the Study:

• To develop and evaluate a simplified algorithm for bone age assessment.
• To compare the accuracy of the grouped-TW algorithm (GTA) with established methods like TW3 and Greulich-Pyle (GP).

Main Methods:

• A dataset of 534 left-hand roentgenograms from children aged 2-15 years was used.
• The GTA algorithm classified radius, ulna, and short bones into three groups using data mining.
• Bone age assessment tables were created using linear regression and fuzzy logic.
• Bland-Altman plots were used to compare accuracy between GTA, GP, and TW3.

Main Results:

• The GTA demonstrated high relative accuracy, achieving 96.2% in boys and 95% in girls compared to TW3.
• GTA showed significantly higher accuracy than the GP method (87%) when compared to TW3.
• Even with non-optimal processing, GTA achieved moderate accuracy (78.2% in boys, 79.6% in girls).

Conclusions:

• The GTA method simplifies the TW3 bone age assessment process while maintaining comparable accuracy.
• GTA offers an efficient alternative for clinical bone age evaluation.
• The relative accuracy of GTA compared to GP is also notable.