Objective: Multiple sclerosis (MS) lesions are heterogeneous with regard to inflammation, demyelination, axonal injury, and neuronal loss. We previously developed a diffusion basis spectrum imaging (DBSI) technique to better address MS lesion heterogeneity. We hypothesized that the profiles of multiple DBSI metrics can identify lesion-defining patterns. Here we test this hypothesis by combining a deep learning algorithm using deep neural network (DNN) with DBSI and other imaging methods. Methods: Thirty-eight MS patients were scanned with diffusion-weighted imaging, magnetization transfer imaging, and standard conventional MRI sequences (cMRI). A total of 499 regions of interest were identified on standard MRI and labeled as persistent black holes (PBH), persistent gray holes (PGH), acute black holes (ABH), acute gray holes (AGH), nonblack or gray holes (NBH), and normal appearing white matter (NAWM). DBSI, diffusion tensor imaging (DTI), and magnetization transfer ratio (MTR) were applied to the 43,261 imaging voxels extracted from these ROIs. The optimized DNN with 10 fully connected hidden layers was trained using the imaging metrics of the lesion subtypes and NAWM. Results: Concordance, sensitivity, specificity, and accuracy were determined for the different imaging methods. DBSI-DNN derived lesion classification achieved 93.4% overall concordance with predetermined lesion types, compared with 80.2% for DTI-DNN model, 78.3% for MTR-DNN model, and 74.2% for cMRI-DNN model. DBSI-DNN also produced the highest specificity, sensitivity, and accuracy. Conclusions: DBSI-DNN improves the classification of different MS lesion subtypes, which could aid clinical decision making. The efficacy and efficiency of DBSI-DNN shows great promise for clinical applications in automatic MS lesion detection and classification.

Purpose: Glioblastoma (GBM) is one of the deadliest cancers with no cure. While conventional MRI has been widely adopted to examine GBM clinically, accurate neuroimaging assessment of tumor histopathology for improved diagnosis, surgical planning, and treatment evaluation remains an unmet need in the clinical management of GBMs.Experimental Design: We employ a novel diffusion histology imaging (DHI) approach, combining diffusion basis spectrum imaging (DBSI) and machine learning, to detect, differentiate, and quantify areas of high cellularity, tumor necrosis, and tumor infiltration in GBM.Results: Gadolinium-enhanced T1-weighted or hyperintense fluid-attenuated inversion recovery failed to reflect the morphologic complexity underlying tumor in patients with GBM. Contrary to the conventional wisdom that apparent diffusion coefficient (ADC) negatively correlates with increased tumor cellularity, we demonstrate disagreement between ADC and histologically confirmed tumor cellularity in GBM specimens, whereas DBSI-derived restricted isotropic diffusion fraction positively correlated with tumor cellularity in the same specimens. By incorporating DBSI metrics as classifiers for a supervised machine learning algorithm, we accurately predicted high tumor cellularity, tumor necrosis, and tumor infiltration with 87.5%, 89.0%, and 93.4% accuracy, respectively.Conclusions: Our results suggest that DHI could serve as a favorable alternative to current neuroimaging techniques in guiding biopsy or surgery as well as monitoring therapeutic response in the treatment of GBM.

High-grade pediatric brain tumors exhibit the highest cancer mortality rates in children. While conventional MRI has been widely adopted for examining pediatric high-grade brain tumors clinically, accurate neuroimaging detection and differentiation of tumor histopathology for improved diagnosis, surgical planning, and treatment evaluation, remains an unmet need in their clinical management. We employed a novel Diffusion Histology Imaging (DHI) approach employing diffusion basis spectrum imaging (DBSI) derived metrics as the input classifiers for deep neural network analysis. DHI aims to detect, differentiate, and quantify heterogeneous areas in pediatric high-grade brain tumors, which include normal white matter (WM), densely cellular tumor, less densely cellular tumor, infiltrating edge, necrosis, and hemorrhage. Distinct diffusion metric combination would thus indicate the unique distributions of each distinct tumor histology features. DHI, by incorporating DBSI metrics and the deep neural network algorithm, classified pediatric tumor histology with an overall accuracy of 85.8%. Receiver operating analysis (ROC) analysis suggested DHI’s great capability in distinguishing individual tumor histology with AUC values (95% CI) of 0.984 (0.982–0.986), 0.960 (0.956–0.963), 0.991 (0.990–0.993), 0.950 (0.944–0.956), 0.977 (0.973–0.981) and 0.976 (0.972–0.979) for normal WM, densely cellular tumor, less densely cellular tumor, infiltrating edge, necrosis and hemorrhage, respectively. Our results suggest that DBSI-DNN, or DHI, accurately characterized and classified multiple tumor histologic features in pediatric high-grade brain tumors. If these results could be further validated in patients, the novel DHI might emerge as a favorable alternative to the current neuroimaging techniques to better guide biopsy and resection as well as monitor therapeutic response in patients with high-grade brain tumors.

PurposeGlioblastoma (GBM) is one of the deadliest cancers with no cure. While conventional MRI has been widely adopted for examining GBM clinically, accurate neuroimaging assessment of tumor histopathology for improved diagnosis, surgical planning, and treatment evaluation, remains an unmet need in the clinical management of GBMs.Experimental DesignWe employ a novel Diffusion Histology Imaging (DHI) approach, combining diffusion basis spectrum imaging (DBSI) and machine learning, to detect, differentiate, and quantify areas of high cellularity, tumor necrosis, and tumor infiltration in GBM.ResultsGd-enhanced T1W or hyper-intense FLAIR failed to reflect the morphological complexity underlying tumor in GBM patients. Contrary to the conventional wisdom that apparent diffusion coefficient (ADC) negatively correlates with increased tumor cellularity, we demonstrate disagreement between ADC and histologically confirmed tumor cellularity in glioblastoma specimens, whereas DBSI-derived restricted isotropic diffusion fraction positively correlated with tumor cellularity in the same specimens. By incorporating DBSI metrics as classifiers for a supervised machine learning algorithm, we accurately predicted high tumor cellularity, tumor necrosis, and tumor infiltration with 87.5%, 89.0% and 93.4% accuracy, respectively.ConclusionOur results suggest that DHI could serve as a favorable alternative to current neuroimaging techniques for guiding biopsy or surgery as well as monitoring therapeutic response in the treatment of glioblastoma.Translational RelevanceCurrent clinical diagnosis, surgical planning, and assessment of treatment response for GBM patients relies heavily on gadolinium-enhanced T1-weighted MRI, which is non-specific for tumor growth and merely reflects a disrupted blood-brain barrier. The complex tumor microenvironment and spatial heterogeneity make GBM difficult to characterize using current clinical imaging modalities. In this study, we developed a novel imaging technique to characterize and accurately predict key histological features of GBM - high tumor cellularity, tumor necrosis, and tumor infiltration. While further validation in a larger cohort of patients is needed, the current proof-of-concept approach could provide a solution to resolve important clinical questions such as the identification of true tumor progression vs. pseudoprogression or radiation necrosis.

High-grade pediatric brain tumors constitute the highest mortality of cancer-death in children.While conventional MRI has been widely adopted for examining pediatric high-grade brain tumor clinically, accurate neuroimaging detection and differentiation of tumor histopathology for improved diagnosis, surgical planning, and treatment evaluation, remains an unmet need in the clinical management of pediatric brain tumor. We employed a novel Diffusion Histology Imaging (DHI) approach that incorporates diffusion basis spectrum imaging (DBSI) and deep neural network. DHI aims to detect, differentiate, and quantify heterogenous areas in pediatric highgrade brain tumors, which include normal white matter (WM), densely cellular tumor (DC tumor), less densely cellular tumor (LDC tumor), infiltrating edge, necrosis, and hemorrhage. Distinct diffusion metric combination would thus indicate the unique distributions of each distinct tumor histology features. DHI, by incorporating DBSI metrics and the deep neural network algorithm, classified pediatric tumor histology with an overall accuracy of 83.3%. Receiver operating analysis (ROC) analysis suggested DHI's great capability in distinguishing individual tumor histology with AUC values (95%CI) of 0.983 (0.985-0.989), 0.961 (0.957-0.964), 0.993 (0.992-0.994), 0.953 (0.947-0.958), 0.974 (0.970-0.978) and 0.980 (0.977-0.983) for normal WM, DC tumor, LDCtumor, infiltrating edge, necrosis and hemorrhage, respectively. Our results suggest that DBSI-DNN, or DHI, accurately characterized and classified multiple tumor histologic features in pediatric high-grade brain tumors. If further validated in patients, the novel DHI might emerge as a favorable alternative to the current neuroimaging techniques to better guide biopsy and resection as well as monitor therapeutic response in patients with high-grade brain tumors.