Analysis of the paraspinal muscle morphology of the lumbar spine using a convolutional neural network (CNN)

Abstract

Purpose This single-center study aimed to develop a convolutional neural network to segment multiple consecutive axial magnetic resonance imaging (MRI) slices of the lumbar spinal muscles of patients with lower back pain and automatically classify fatty muscle degeneration. Methods We developed a fully connected deep convolutional neural network (CNN) with a pre-trained U-Net model trained on a dataset of 3,650 axial T2-weighted MRI images from 100 patients with lower back pain. We included all qualities of MRI; the exclusion criteria were fractures, tumors, infection, or spine implants. The training was performed using k-fold cross-validation (k = 10), and performance was evaluated using the dice similarity coefficient (DSC) and cross-sectional area error (CSA error). For clinical correlation, we used a simplified Goutallier classification (SGC) system with three classes. Results The mean DSC was high for overall muscle (0.91) and muscle tissue segmentation (0.83) but showed deficiencies in fatty tissue segmentation (0.51). The CSA error was small for the overall muscle area of 8.42%, and fatty tissue segmentation showed a high mean CSA error of 40.74%. The SGC classification was correctly predicted in 75% of the patients. Conclusion Our fully connected CNN segmented overall muscle and muscle tissue with high precision and recall, as well as good DSC values. The mean predicted SGC values of all available patient axial slices showed promising results. With an overall Error of 25%, further development is needed for clinical implementation. Larger datasets and training of other model architectures are required to segment fatty tissue more accurately.

Three‐dimensional freehand ultrasonography to measure muscle volume of the lumbar multifidus: Reliability of processing technique and validity through comparison to magnetic resonance imaging

“…Obviously, (semi‐) automatic segmentation procedures to define muscle borders would facilitate future studies and use in clinical practice. For MRI, recent advances in artificial intelligence such as convolutional neural networks showed promising results in automatic image segmentation and quantification of FF (Baur et al., 2021; Käser et al., 2001; Paliwal et al., 2021; Shen et al., 2021; Weber et al., 2019, 2021). However, the translation of these automated methods to clinical practice has not yet been realized and to our knowledge, this has not yet been developed for 3DfUS.…”

“…Since the variation in muscle structure is expected to be smaller in healthy subjects, a better agreement between the two modalities could be assumed. ing results in automatic image segmentation and quantification of FF (Baur et al, 2021;Käser et al, 2001;Paliwal et al, 2021;Shen et al, 2021;Weber et al, 2019Weber et al, , 2021. However, the translation of these automated methods to clinical practice has not yet been realized and to our knowledge, this has not yet been developed for 3DfUS.…”

Kommentar zu: Different degeneration patternsof paraspinal muscles in degenerative lumbar disease:a MRI analysis of 154 patients

“…Eine eigene Klassifikation wie nach Goutallier bei der Schulter fehlt aktuell. Moderne KI-Systeme [4] werden in Zukunft Radiologie und Wirbelsäulentherapeuten/-innen unterstützende Informationen liefern. In der Praxis sollte die Betrachtung der paraspinalen Muskulatur fest die Bildanalyse einbezogen werden.…”

A Convolutional Neural Network for Automated Detection of Cervical Ossification of the Posterior Longitudinal Ligament using Magnetic Resonance Imaging

“…Successful applications of CNN to the evaluation of spinal diseases have been reported, including automated lumbar segmentation, obtain radiological parameters, and detection of certain spinal lesions. [10][11][12][13] Although, most of the previous studies were focused on lumbar spine, CNN models adopted in analysis of cervical spinal disorders remains under explored. Recently, 3 studies have evaluated the usefulness of CNN in detecting and diagnosing cervical OPLL on plain radiographs; [14][15][16] however, the performance of CNN models using MRI images to distinguish cervical OPLL from degenerative spinal changes is still unknown.…”

“…Image features are automatically and adaptively learned and extracted to establish a model that cover the complex relationship between these features and diagnosis. Successful applications of CNN to the evaluation of spinal diseases have been reported, including automated lumbar segmentation, obtain radiological parameters, and detection of certain spinal lesions 10–13 . Although, most of the previous studies were focused on lumbar spine, CNN models adopted in analysis of cervical spinal disorders remains under explored.…”