Artikel
Automated image analysis of video fluoroscopy in the cervical spine
Suche in Medline nach
Autoren
-
Mats Leif Moskopp
- Institut für Physiologie der TU Dresden, Dresden, Deutschland
-
Dag Moskopp
- Vivantes-Klinikum im Friedrichshain, Klinik für Neurochirurgie, Berlin, Deutschland
| Veröffentlicht: | 18. Juni 2018 |
|---|
Gliederung
Text
Objective: Functional cervical x-ray diagnostics are an established tool in the diagnosis of cervical instability. However, this technique relies mostly on the static data of endpoint flexion and extension measurements. Video fluoroscopy (VF) data is publicly available on a large scale and might help to get deeper insights into the movement pattern of the cervical vertebrae and the spine as a whole. It is the aim of this study to investigate the benefit of automated imaging to assess movement patterns in the cervical spine.
Methods: Video hosting services were systematically searched for "cervical video fluoroscopy" and "cervical fluoroscopy". Publicly available videos were included into this study based on the following criteria:
- Content shows a sequence of flexion and extension of the cervical spine.
- Data is sufficiently anonymized.
- Data has sufficient temporal and spatial resolution.
An imaging algorithm was implemented in python programming language. The algorithm relies on pattern recognition and local contrast enhancement. Therefore, every vertebra is tracked individually using the 2D projection of the four corners of the vertebra body plus its spinous process.
Results: The algorithm is able to track vertebrae in the moving cervical spine even on low resolution images. Measurements error are mainly defined by the data quality. The internal quality of the measurement can be assessed by an internal evaluation of multiple tracking points per vertebra. Individual trajectories of all vertebrae allow the analysis of single vertebra as well as their proportion to the movement of the whole spine. Due to the simplicity of this universal approach calculations can be performed on a standard desktop PC even for big datasets in less than 1 min. After the recording of the VF no additional hardware is required.Measurements on a healthy subject allow to quantify the movement range of individual cervical Junghanns motion segments (Flex.: range: 1.3° – 32.1°, mean: 20.0°±5.4°. Ext.: range: -6.5° – 10.2°, mean: 0.6°±3.2° (N=5)).
Conclusion: The results indicate a wide variety of possible usage for the presented algorithm ranging from a diagnostic instrument in the assessment of cervical instability to a tool that might provide information in the design of individually adjusted prosthetics. Further, this physiological pilot study might provide a new approach in early diagnosis of monosegmental adjacent level disease.
Figure 1 [Fig. 1]
