The subjective analysis of conventional radiography represents the principal method for bone diagnostics in endoprosthetics. Alternative objective quantitative methods are described but not commonly used. Therefore, semi-quantitative methods are tested using digital computation and artificial intelligence to standardize, simplify, and ultimately improve the assessment. This study aimed to evaluate the correlation between relative density progressions and clinical outcomes. Radiographs and clinical examinations before and 24 and 48 weeks after surgery were obtained from sixty-eight patients with a modular hip stem. For the calculation of the relative bone density, the modal gray values of the Gruen zones were measured using ImageJ and were normalized by gray values of the highest and lowest ROI. The clinical outcomes were measured according to the Harris hip score before evaluating them for correlations. Analyses were performed separately for subgroups and bone regions. The Harris hip score increased from 44.15 ± 15.00 pre-operatively to 66.20 ± 13.87 at the latest follow-up. The relative bone density adjustment of Gruen zone 7 showed a significant correlation to its clinical outcome. Other bone adaptations could be realistically reproduced and differences by regional zones and patients' histories visualized. Next to the simplicity and that no additional examination is required, the method provides good semi-quantitative results and visualizes adaptations, which make it suitable for use.
Read full abstract