Detection intervertebral disc fissures in conventional MRI for clinical decision support using a combination of AI methods  — The International Society for the Study of the Lumbar Spine
  1. Schedule
  2. SESSION #14 - ARTIFICIAL INTELLIGENCE AND NOVEL TECHNIQUE
  3. Detection intervertebral disc fissures in conventional MRI for clinical decision support using a combination of AI methods 

Detection intervertebral disc fissures in conventional MRI for clinical decision support using a combination of AI methods  (#78)

Hanna Hebelka 1 2 , Christian Waldenberg 1 3 , Stefanie Eriksson 1 3 , Helena Brisby 1 4 , Kerstin Lagerstrand 1 3
  1. Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden
  2. Dept. of Radiology, Sahlgrenska University Hospital, Gothenburg, Sweden
  3. Dept. of Radiation Physics, Sahlgrenska Academy, University of Gothenburg, Sweden, Gothenburg, Sweden
  4. Dept. of Orthopaedics, Sahlgrenska University Hospital, Gothenburg, Sweden

Introduction

Low back pain (LBP) is one of the most costly disorders worldwide [1]. Intervertebral disc (IVD) fissures in the outer annulus fibrosus (AF) are believed to be closely related to non-specific LBP [2]. Unfortunately, only a minority of these potentially painful fissures are identified in clinical routine magnetic resonance (MR) images [3]. New non-invasive methods for the detection of fissures have been developed, however, application of these in clinical settings is limited since they are restricted to in vitro measurements or high-field scanners and lack validation against rigorous reference standards [3-6]. This study aims to develop a clinically applicable method capable of determining the occurrence and position of annular fissures using only conventional MR images and artificial intelligence (AI) based techniques.

Methods

A total of 123 IVDs in 43 patients (age 25-63 years, 19 male) suffering from chronic LBP were examined with T1- and T2-weighted conventional MR imaging, low-pressure discography and computed tomography (CT). Using textural features calculated from the MR images, an artificial neural network (ANN) classification model and a method for attention mapping of fissures were developed to identify the occurrence and position of fissures reaching the outer AF. The extension of the fissures was graded according to the Dallas Discogram Description (DDD) [7] using the CT-discograms. Since fissures in the outer AF have been suggested to be markers of LBP [8], IVDs with fissures extending to the outer AF (DDD=2 and 3) were sorted into one group and IVDs with no fissures or fissures not extending to the outer AF (DDD=0 and 1) were sorted into the second group.

Results

The AI-based classifier correctly identified the occurrence of annular fissures in 122 of 123 IVDs (98.9% sensitivity/100% specificity) (Fig.1). A receiver operating characteristic (ROC) analysis displayed an area under the curve (AUC) of 0.9996, indicating outstanding discriminating ability. The true position of the fissures was determined in 87% of the IVDs (Fig.2). Inaccurately localized fissures were diffuse and/or non-delimitable fissures and, in the majority of cases, localized within noticeably degenerated IVDs.

 

Discussion

This is the first study utilizing a combination of texture analysis, ANNs and attention mapping to add invisible but important information within diagnostics. This classifier displayed outstanding discriminating ability and the vast majority of the attention maps determined the true location of the fissures but struggled with noticeably degenerated IVD with several fissures in both the posterior and anterior AF. However, IVDs in the late stages of degeneration have lost their biomechanical properties and viability and are considered more stable with impaired micro-mobility and, thus, less likely to give rise to pain sensations based on direct disc pathology. The novel method shows great promise and can be easily implemented to confidently detect the occurrence and position of potential painful fissures in conventional MR images. As such, it can be used to obtain unique insights into pathology, increase the diagnostic accuracy and allow for new non-invasive research in the clinical setting, both regarding spinal pathology and likely also within other images-based diagnostic areas.

 

 619262e671ebe-fig1.JPG

619262e671ebe-fig2.JPG

  1. Hoy, D., et al., The global burden of low back pain: estimates from the Global Burden of Disease 2010 study. Ann Rheum Dis, 2014. 73(6): p. 968-74.
  2. DePalma, M.J., J.M. Ketchum, and T. Saullo, What Is the Source of Chronic Low Back Pain and Does Age Play a Role? Pain Medicine, 2011. 12(2): p. 224-233.
  3. Berger-Roscher, N., et al., Intervertebral disc lesions: visualisation with ultra-high field MRI at 11.7 T. European spine journal, 2015. 24(11): p. 2488-2495.
  4. Eldaya, R.W., et al., Evaluating the effect of a post-processing algorithm in detection of annular fissure on MR imaging. European Spine Journal, 2021.
  5. Raudner, M., et al., Clinical implementation of accelerated T2 mapping: Quantitative magnetic resonance imaging as a biomarker for annular tear and lumbar disc herniation. European Radiology, 2021. 31(6): p. 3590-3599.
  6. Deneuville, J.-P., et al., Quantitative MRI to Characterize the Nucleus Pulposus Morphological and Biomechanical Variation According to Sagittal Bending Load and Radial Fissure, an ex vivo Ovine Specimen Proof-of-Concept Study. Frontiers in Bioengineering and Biotechnology, 2021. 9(489).
  7. Sachs, B.L., et al., Dallas discogram description. A new classification of CT/discography in low-back disorders. Spine (Phila Pa 1976), 1987. 12(3): p. 287-94.
  8. Peng, B., et al., The pathogenesis and clinical significance of a high-intensity zone (HIZ) of lumbar intervertebral disc on MR imaging in the patient with discogenic low back pain. European Spine Journal, 2006. 15(5): p. 583-587.
#ISSLS2022