Preliminary data on artificial intelligence tool in magnetic resonance imaging assessment of degenerative pathologies of lumbar spine

Radiological Reporting of Brain Atrophy in MRI: Real-Life Comparison Between Narrative Reports, Semiquantitative Scales and Automated Software-Based Volumetry

Background: Accurate assessment of brain atrophy is essential in the diagnosis and monitoring of brain aging and neurodegenerative disorders. Radiological methods range from narrative reporting to semi-quantitative visual rating scales (VRSs) and fully automated volumetric software. However, their integration and consistency in clinical practice remain limited. Methods: In this retrospective study, brain MRI images of 43 patients were evaluated. Brain atrophy was assessed by extrapolating findings from narrative radiology reports, three validated VRSs (MTA, Koedam, Pasquier), and Pixyl.Neuro.BV, a commercially available volumetric software platform. Agreement between methods was assessed using intraclass correlation coefficients (ICCs), Cohen's kappa, Spearman's correlation, and McNemar tests. Results: Moderate correlation was found between narrative reports and VRSs (ρ = 0.55-0.69), but categorical agreement was limited (kappa = 0.21-0.30). Visual scales underestimated atrophy relative to software (mean scores: VRSs = 0.196; software = 0.279), while reports tended to overestimate. Agreement between VRSs and software was poor (kappa = 0.14-0.33), though MTA showed a significant correlation with hippocampal volume. Agreement between reports and software was lowest for global atrophy. Conclusions: Narrative reports, while common in practice, show low consistency with structured scales and quantitative software, especially in subtle cases. VRSs improve standardization but remain subjective and less sensitive. Integrating structured scales and volumetric tools into clinical workflows may enhance diagnostic accuracy and consistency in dementia imaging.

Recent topics in musculoskeletal imaging focused on clinical applications of AI: How should radiologists approach and use AI?

The advances in artificial intelligence (AI) technology in recent years have been remarkable, and the field of radiology is at the forefront of applying and implementing these technologies in daily clinical practice. Radiologists must keep up with this trend and continually update their knowledge. This narrative review discusses the application of artificial intelligence in the field of musculoskeletal imaging. For image generation, we focused on the clinical application of deep learning reconstruction and the recently emerging MRI-based cortical bone imaging. For automated diagnostic support, we provided an overview of qualitative diagnosis, including classifications essential for daily practice, and quantitative diagnosis, which can serve as imaging biomarkers for treatment decision making and prognosis prediction. Finally, we discussed current issues in the use of AI, the application of AI in the diagnosis of rare diseases, and the role of AI-based diagnostic imaging in preventive medicine as part of our outlook for the future.

Automated spinopelvic measurements on radiographs with artificial intelligence: a multi-reader study

PURPOSE

To develop an artificial intelligence (AI) algorithm for automated measurements of spinopelvic parameters on lateral radiographs and compare its performance to multiple experienced radiologists and surgeons.

METHODS

On lateral full-spine radiographs of 295 consecutive patients, a two-staged region-based convolutional neural network (R-CNN) was trained to detect anatomical landmarks and calculate thoracic kyphosis (TK), lumbar lordosis (LL), sacral slope (SS), and sagittal vertical axis (SVA). Performance was evaluated on 65 radiographs not used for training, which were measured independently by 6 readers (3 radiologists, 3 surgeons), and the median per measurement was set as the reference standard. Intraclass correlation coefficient (ICC), mean absolute error (MAE), and standard deviation (SD) were used for statistical analysis; while, ANOVA was used to search for significant differences between the AI and human readers.

RESULTS

Automatic measurements (AI) showed excellent correlation with the reference standard, with all ICCs within the range of the readers (TK: 0.92 [AI] vs. 0.85-0.96 [readers]; LL: 0.95 vs. 0.87-0.98; SS: 0.93 vs. 0.89-0.98; SVA: 1.00 vs. 0.99-1.00; all p < 0.001). Analysis of the MAE (± SD) revealed comparable results to the six readers (TK: 3.71° (± 4.24) [AI] v.s 1.86-5.88° (± 3.48-6.17) [readers]; LL: 4.53° ± 4.68 vs. 2.21-5.34° (± 2.60-7.38); SS: 4.56° (± 6.10) vs. 2.20-4.76° (± 3.15-7.37); SVA: 2.44 mm (± 3.93) vs. 1.22-2.79 mm (± 2.42-7.11)); while, ANOVA confirmed no significant difference between the errors of the AI and any human reader (all p > 0.05). Human reading time was on average 139 s per case (range: 86-231 s).

CONCLUSION

Our AI algorithm provides spinopelvic measurements accurate within the variability of experienced readers, but with the potential to save time and increase reproducibility.