Improved 18-FDG PET/CT diagnosis of multiple myeloma diffuse disease by radiomics analysis

Abstract

In Multiple Myeloma (MM) patients, diffuse infiltration of bone marrow can be diagnosed on MRI and is associated with poorer prognosis. On 18-FDG PET/CT, the other important imaging modality in MM, the diagnosis of diffuse disease by visual analysis can be challenging.Radiomics allows the extraction of large amount of data from images to individualize disease specific diagnostic or prognostic patterns. We aimed to develop a radiomics-based model derived from PET and CT images, that could improve the diagnosis of multiple myeloma diffuse disease on 18-FDG PET/CT. MethodsWe prospectively performed PET/CT and Whole-Body MRI in 30 MM patients at initial diagnosis. MRI was the standard of reference for diffuse disease assessment. Twenty patients were assigned to the training set and 10 to the independent test set. Visual analysis of PET/CT was done by two nuclear medicine physicians separately, then by consensual reading.Spine volumes on CT and PET were automatically segmented and a total of 174 IBSI-compliant radiomics features from both volumes were extracted. Selection of best features in the training set was performed with Random Forest features importance combined with correlation analysis.Machine-learning algorithms were then trained on the selected features with cross-validation.Finally, the model was evaluated on an independent test set. ResultsOut of the 30 patients, 18 had established diffuse disease on MRI. After consensus of PET/CT images, the sensitivity, specificity and accuracy of visual analysis was 67%, 75% and 70%, respectively, with a moderate kappa coefficient of agreement of 0.6. Five radiomics features were selected. On the training set, Random Forest classifier reached the highest mean accuracy of 0.91 (95% CI 0.90-0.92) with an AUC of 0.90 (95% CI 0.89-0.91) for diffuse disease diagnosis. On the independent test set, the model achieved an accuracy of 80%.Conclusions Radiomics analysis of 18-FDG PET/CT images with machine-learning overcame the limitations of visual analysis, providing a highly accurate and more reliable diagnosis of diffuse bone marrow infiltration in multiple myeloma patients.

Comparison of FDG PET/CT and Bone Marrow Biopsy Results in Patients with Diffuse Large B Cell Lymphoma with Subgroup Analysis of PET Radiomics

“…Recently, in hematologic application, FDG PET radiomics was utilized for differential diagnosis of primary lymphoma and other solid tumors [ 12 ]. Studies have looked into applying FDG PET radiomics of the BM to diagnose or predict the clinical outcome in patients with leukemia and multiple myeloma [ 13 , 14 , 15 , 16 ].…”

11C-Methionine PET/CT in Assessment of Multiple Myeloma Patients: Comparison to 18F-FDG PET/CT and Prognostic Value

“…Given that metabolism is considered a hallmark of cancer and glycolysis is one of the main metabolic pathways that are deregulated [ 11 ], new metabolic biomarkers, such as total lesion glycolysis (TLG) and total metabolically active tumor lesion volume (TMTV) have been investigated [ 12 , 13 ]. However, the diagnostic value of FDG decreases when only diffuse BM infiltration or only small lytic skeletal lesions (smaller than 10 mm) are present [ 14 , 15 , 16 ].…”

The application of machine learning to imaging in hematological oncology: A scoping review

“…Of the 33 studies that applied ML techniques to diagnose a hematological malignancy or to differentiate it from another disease state or malignancy ( Table 1 ), 18 were designed to establish and train ML models to discriminate gliomas [predominantly GBM from PCNSL ( 29 – 31 , 34 – 37 , 40 – 42 , 44 , 45 , 49 , 50 , 53 – 56 )] using features extracted from FDG-PET [one study ( 29 ),] or MRI ( 30 , 31 , 34 – 37 , 40 – 42 , 44 , 45 , 49 , 50 , 53 – 56 ) images. The remaining studies belonged to two major categories: those developing models to discriminate solid hematological malignancies from other benign and malignant lesions at other sites [nasopharyngeal carcinomas from nasopharyngeal lymphoma ( 46 , 48 ), idiopathic orbital inflammation from ocular adnexal lymphoma ( 33 ), thymic neoplasm from thymic lymphoma ( 14 ), breast carcinoma from breast lymphoma ( 15 ), lymphoma from normal nodes ( 43 ), or multiple myeloma from bone metastases ( 51 )] and those that detect the location of hematological malignancies either at diagnosis or during the disease course [location of ( 18 ) or evolving/residual lymphoma ( 32 ) or leukemia ( 17 ) or bone marrow involvement with multiple myeloma ( 16 , 38 , 47 , 52 ) or mantle cell lymphoma ( 39 )].…”

“…A second major application of ML techniques is to improve the detection and monitoring of hematological malignancies for accurate diagnosis, treatment, and staging. For example, the FDG-PET/CT images of lymphomas and multiple myeloma can be difficult to interpret due to low avidity, unusual distribution patterns [particularly diffuse disease in multiple myeloma ( 16 ) and leukemia ( 17 )], or motion/attenuation artifacts, especially for inexperienced readers. Algorithms to support diagnostic decision-making would therefore be helpful ( 16 , 18 ).…”