Role of computed tomography texture analysis using dual-energy-based bone marrow imaging for multiple myeloma characterization: comparison with histology and established serologic parameters

Abstract

Objective To identify textural features on dual-energy CT (DECT)–based bone marrow images in myeloma which correlate with serum markers of myeloma activity and the degree of medullary involvement. Methods A total of 110 patients (63.0 ± 11.0 years, 51 female) who underwent unenhanced whole-body DECT between September 2015 and February 2019 were retrospectively included, which was approved by our institutional ethics committee with a waiver of the informed consent requirement. All patients had current hematologic laboratory tests. Using DECT post-processing, non-calcium bone marrow images were reconstructed. The vertebral bodies T10–L5 were segmented for quantification of textural features, which were compared with serologic parameters and myeloma stages by the Mann-Whitney U test. In a subgroup of 56/110 patients with current bone marrow biopsies, textural features were correlated with the degree of bone marrow infiltration. Results First-order features were higher in patients with advanced stage of myeloma (p < .02), whereas the 2nd-order “gray-level co-occurrence matrix (GLCM) cluster prominence” was lower (p < .04). In patients with elevated serum-free light chains (SFLC) or kappa/lambda SFLC ratio above 1.56, the “entropy” and 2nd-order GLCM features were lower (p < .03). The degree of bone marrow infiltration correlated with 1st-order features (e.g., “uniformity”; rP = 0.49; p < .0001), whereas “entropy” and 2nd-order GLCM features were negatively correlated (e.g., “difference entropy”; rP = − 0.54; p < .0001). Conclusions CT textural features applied on non-calcium bone marrow images correlate well with myeloma-related serologic parameters and histology showing a more uniform tissue structure and higher attenuation with increasing medullary infiltration and could therefore be used as additional imaging biomarkers for non-invasive assessment of medullary involvement. Key Points • Texture analysis applied on dual-energy reconstructed non-calcium bone marrow images provides information about marrow structure and attenuation. • Myeloma-related serologic parameters and the degree of myeloma cell infiltration correlate with 1st- and 2nd-order features which could be useful as additional imaging biomarkers for non-invasive assessment of medullary involvement.

Feasibility of artificial intelligence–supported assessment of bone marrow infiltration using dual-energy computed tomography in patients with evidence of monoclonal protein — a retrospective observational study

“…The peak at the left end of both histograms can be explained by the virtual calcium removal, which results in extreme negative attenuation of structures with dense calcification. Further, the large overlap of both histograms in the range of negative HU values might reflect the resemblance of bone mineral density of both subsets: after virtual calcium removal, the bone mineral density might crucially influence the negative part of the attenuation histogram A recent study correlated BM texture analysis in dualenergy CT (DECT) with BM infiltration in MM [34]. The study extracted 92 PyRadiomics features for each of 56 patients, which yielded six significant features for Pearson's correlation with BM infiltration.…”

“…Appropriate sample size of the multivariate regression was calculated by the software G*Power for a desired power level of 0.8, significance level of 0.05, and an estimated medium to large effect size (f 2 = 0.32) of the main Fig. 2 Inclusion chart of the study population independent variable, based on reported data (supplementary data 3) [34,35].…”

MRI-Based Bone Marrow Radiomics Nomogram for Prediction of Overall Survival in Patients With Multiple Myeloma

“…Nonetheless, the potentially useful information of MRI has yet not been fully exploited. Several studies have shown some efficiency of the bone marrow radiomics, a study had revealed that dual-energy CT textural features correlate well with MM-related serologic parameters and histology ( 25 ). Another study confirmed the predictive value of radiomics based on PET-CT imaging in MM ( 26 ).…”

Texture Features of Computed Tomography Image under the Artificial Intelligence Algorithm and Its Predictive Value for Colorectal Liver Metastasis

“…Texture analysis can quantify the intuitive quality described by terms such as roughness, smoothness, or bump as a function of spatial variation in pixel intensity. The heterogeneity of different results can be explored through texture analysis to identify texture features that can be used for diagnosis and prediction [ 27 ]. At present, texture analysis has been widely used in tumor imaging, has important value in the judgment of benign and malignant tumors, metastasis, and prognosis, and can guide clinical treatment.…”