Chuan Zhou PhD, Presenter: Nothing to Disclose

Qian Dong MD, Abstract Co-Author: Nothing to Disclose

Heang-Ping Chan PhD, Abstract Co-Author: Institutional research collaboration, General Electric Company

Daniel R. Couriel, Abstract Co-Author: Nothing to Disclose

Lubomir M. Hadjiiski PhD, Abstract Co-Author: Nothing to Disclose

Jun Wei PhD, Abstract Co-Author: Nothing to Disclose

Attaphol Pawarode, Abstract Co-Author: Nothing to Disclose

Bo He, Abstract Co-Author: Nothing to Disclose

Studies revealed that MRI patterns of multiple myeloma (MM) infiltration in the bone marrow have prognostic significance and are useful for the evaluation of tumor burdens and treatment response for MM patients. We are developing quantitative MRI analysis method for treatment response assessment of MM. This study investigated the feasibility of stratifying MRI patterns of MM infiltration.

With IRB approval, 50 pairs of spinal MR scans performed before and after bone marrow transplant were collected retrospectively from 50 patients with MM disease. An experienced musculoskeletal radiologist visually examined each vertebra and provided the descriptor of its pattern (normal, focal, variegated (salt-pepper), and diffuse) as reference standard. A 3D dynamic intensity entropy transformation (DIET) method was developed to transform MR T1-weighted signal voxel by voxel to a quantitative entropy enhancement value (qEEV), defined as the intensity entropy at the voxel normalized by the median intensity entropy in the adjacent intervertebral discs. The mean of qEEV (m-qEEV) in each vertebral body was used as a predictor for three classification tasks: (1) diffuse vs group of the normal, focal, variegated (salt-pepper) patterns, (2) normal vs group of the focal, variegated and diffuse patterns, and (3) group of variegated and diffuse patterns vs group of normal and focal patterns. The classification performance was evaluated by ROC analysis.

Of the 1022 vertebras, 229, 241, 372 and 180 vertebras were identified by radiologist as normal, focal, variegated and diffuse pattern, respectively. The m-qEEV values measured in the set of the four patterns were 2.01±0.26, 1.84±0.26, 1.70±0.27 and 1.32±0.14, respectively. The area under the ROC curve (AUC) achieved 0.960±0.006, 0.807±0.014 and 0.804±0.013 for the three classification tasks, respectively. 

DIET method accurately classified MRI patterns with different grades of MM infiltration, especially the high-grade diffuse pattern that tend to have higher tumor burden than those with normal or focal patterns. This has the potential to quantitatively track the regression or progression of MM during or after treatment.

Quantitative image-based biomarker may improve the accuracy and efficacy for staging and assessing treatment response for MM, allowing clinicians to optimize therapy of individual patients.

Zhou, C, Dong, Q, Chan, H, Couriel, D, Hadjiiski, L, Wei, J, Pawarode, A, He, B, Quantitative Analysis of MRI for Treatment Response Assessment of Multiple Myeloma (MM): Stratification of MM Infiltration Pattern in Bone Marrow Using Dynamic Intensity Entropy Transformation (DIET) Method.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045702.html