#Download ibm spss statistics 21 manual

However, the latter focussed on the L4–L5 intervertebral disc level and thus volumetric measures of the paraspinal muscles traversing the entire lumbar spine were not available. (2021) demonstrated high performance of CNN for the segmentation of lumbar paraspinal muscles from T 2-weighted axial images 13. (2019) provided evidence that CNNs can be used to automate the calculation of both muscle volume and fat measures of the cervical spine extensor muscles with high segmentation performance (i.e., accuracy and reliability) using MR fat–water imaging 14. shapes, localization) using filters whose receptive fields capture more global information 16. First, shallow layers collect low-level features (e.g., edges, contrasts) while deeper layers collect high-level features (e.g. CNNs are able to learn hierarchical spatial features 15 with an increasing level of abstraction as the imaging inputs are processed through the network layers 16. Recent applications of deep learning methods, in particular convolutional neural networks (CNN), have shown potential for automating the segmentation of the cervical and lumbar paraspinal muscles from MRI 13, 14. Similar to other research fields (e.g., spinal cord injury and knee osteoarthritis) 10, 11, the development of time-efficient and fully-automated tissue segmentation techniques are needed to realise the potential implementation of (muscle) morphometric and compositional measures 12 when clinically warranted.

#Download ibm spss statistics 21 manual

However, quantitative musculoskeletal MRI measurements require manual segmentation of the muscle borders, which is time-consuming and user-dependent, representing a significant barrier to the translation of these quantitative MRI methods to clinical practice.

Paraspinal muscle morphometric and compositional measures are preferably quantified by magnetic resonance imaging (MRI) due to high soft tissue contrast 9. Beyond differences between study samples, such conflicting results could be the consequence of differences in imaging and quantification methods to assess lumbar paraspinal muscle health 7, 8. In some studies, a decrease in muscle volume or increase of fatty infiltration of the paraspinal muscles was highly associated with the presence and severity of LBP 5, 6, but other studies disagree 4. However, the magnitude of, and association between, paraspinal muscle health and the clinical course of LBP remains largely unknown 4. Various parameters of paraspinal muscle health (e.g., size, shape, and composition) have been acknowledged as potentially important biological markers in people with LBP 3. Low back pain (LBP) is the leading cause of disability worldwide 1 driven by a complex multifactorial inter-relationship between biological, psychological and social systems 2. CNNs can be used to efficiently and accurately extract measures of paraspinal muscle health from MRI. All trained CNN models demonstrated high accuracy and excellent reliability for segmenting lumbar paraspinal muscles. Increasing network depth did not improve accuracy ( p = 0.771).

The 2D model trained without data augmentation demonstrated the highest average segmentation accuracy. Overall, across all muscles, 2D models performed better than 3D models ( p = 0.012), and training without data augmentation outperformed training with data augmentation ( p < 0.001). Each model showed high segmentation accuracy (Sørensen-Dice index ≥ 0.885) and excellent reliability (ICC 2,1 ≥ 0.941). The effect of increasing network depth on segmentation accuracy was also investigated. Segmentation accuracy was compared between the models using the Sørensen-Dice index as the primary outcome measure. Using cross-validation, we compared 2D and 3D CNNs with and without data augmentation. T 2-weighted MRI axial images from 76 participants (46 female age (SD): 45.6 (12.8) years) with low back pain were used to train CNN models to segment the multifidus, erector spinae, and psoas major muscles (left and right segmented separately). Here we trained a CNN for the automatic segmentation of lumbar paraspinal muscles and determined the impact of CNN architecture and training choices on segmentation performance. Convolutional neural networks (CNNs) provide alternate timesaving, state-of-the-art performance measures, which could realise clinical translation. Measures of size, shape, and composition have required time-consuming and rater-dependent manual segmentation techniques. The size, shape, and composition of paraspinal muscles have been widely reported in disorders of the cervical and lumbar spine.