For years, the image processing experts have applied texture analysis methods to medical images in order to extract indicators of composition and heterogeneity of the regions analysed. Texture analysis is applied in many imaging fields such as brain, muscle, bone, oncology, lung parenchyma and so on. Therefore it seems to be evident that it is a cross-specialty technique that can be applied to medical images to enhance the qualitative information that is frequently appreciated with quantitative data about composition and heterogenity. The applications in the oncology field are specially promising, with some clear conclusions in the Radiology journal indicating that texture analysis provides a very good prognosis evaluation of rectal cancer patients survival.
In the last 8 years, a texture-derived biomarker called Trabecular Bone Score (TBS) has been increasingly applied to the evaluation of fracture risk in patients. The TBS is calculated from the texture analysis applied to the densitometry (DXA) acquisitions of the column and it has been published to be related to osteoporosis and complementary to FRAX® in the follow-up of Osteoporosis in a significant number of studies.
Nevertheless, the applicability of texture analysis to any field has to be carefully analysed. The gold standard for fracture risk assessment is to experimentally evaluate the mechanical resistance by laboratory tests and it has to be noted that a significant discrepancy exists between the TBS performance in assessing fracture risk and the evidence of mechanical resistance evaluation of bone, as stated in this study:
“its discriminative power in fracture studies remains incomprehensible because prior biomechanical tests found no correlation with vertebral strength. To verify this result possibly owing to an unrealistic setup and to cover a wide range of loading scenarios, the data from three previous biomechanical studies using different experimental settings were used. They involved the compressive failure of 62 human lumbar vertebrae loaded 1) via intervertebral discs to mimic the in vivo situation (“full vertebra”); 2) via the classical endplate embedding (“vertebral body”); or 3) via a ball joint to induce anterior wedge failure (“vertebral section”). High-resolution peripheral quantitative computed tomography (HR-pQCT) scans acquired from prior testing were used to simulate anterior-posterior DXA from which areal bone mineral density (aBMD) and the initial slope of the variogram (ISV), the early definition of TBS, were evaluated. Finally, the relation of aBMD and ISV with failure load (Fexp) and apparent failure stress (σexp) was assessed, and their relative contribution to a multilinear model was quantified via ANOVA. We found that, unlike aBMD, ISV did not significantly correlate with Fexp and σexp, except for the “vertebral body” case (r2 = 0.396, p = 0.028). Aside from the “vertebra section” setup where it explained only 6.4% of σexp (p = 0.037), it brought no significant improvement to aBMD. These results indicate that ISV, a replica of TBS, is a poor surrogate for vertebral strength no matter the testing setup, which supports the prior observations and raises a fortiori the question of the deterministic factors underlying the statistical relationship between TBS and vertebral fracture risk. Maquer, G., Lu, Y., Dall’Ara, E., Chevalier, Y., Krause, M., Yang, L., Eastell, R., Lippuner, K. and Zysset, P. K. (2016), The Initial Slope of the Variogram, Foundation of the Trabecular Bone Score, Is Not or Is Poorly Associated With Vertebral Strength. J Bone Miner Res, 31: 341–346. doi:10.1002/jbmr.2610″
In QUIBIM we believe that a “trabecular score” should only be named in this way when enough depiction of the microarchitecture is achieved by imaging methods, like the existing in X-rays, computed tomography (CT) with collimations for high spatial resolution, high spatial resolution peripheral quantitative computed tomoraphy (HR-pQCT) or high spatial resolution magnetic resonance imaging (MRI). Trabeculae depiction can not be achieved with current DXA acquisitions, therefore we propose to name the TBS process simply by DXA texture analysis in order to avoid confusion.
Our computational algorithms allow for the quantitative characterization of trabecular bone properties from high spatial resolution imaging methods, including plain radiographs, therefore at similar dose and cost than current DXA. A complete structured report with the most important morphometry characteristics (Bone Volume to Total Volume – BV/TV; Trabecular thickness – Tb.Th; Trabecular Separation – Tb.Sp; Trabecular Number – Tb.N), irregularity indicators (Fractal Dimension in 2D and 3D – D2D, D3D) and mechanical analysis by the finite element method to calculate the Young’s modulus (Eapp). These methods have been validated against gold standards and high spatial resolution techniques. All these parameters are fused into a Trabecular Bone Architecture Quality Index to stratify fracture risk in patients. An intuitive structured report is generated for the clinician that allows to follow-up patients not only under clinical routine but also in the frame of clinical trials for the therapy response evaluation.
We have established the following plug-ins and methods for the assessment of Osteoporosis and bone diseases:
- 2D Trabecular Bone characterisation (X-ray, CT and MRI)
- 3D Trabecular Bone characterisation (CT and MRI)
- Vertebral fractures detection (X-ray, CT and MRI)
- DXA textural and signal analysis (DXA)
- Phantomless BMD analysis from CT scans (CT)
Try our solution! Upload a case of X-rays, DXA, CT, or MRI to us through our QUIBIM-Precision® platform: https://precision.quibim.com