Estimating Local Cellular Density in Glioma Using MR Imaging Data.

BACKGROUND AND PURPOSE: Increased cellular density is a hallmark of gliomas, both in the bulk of the tumor and in areas of tumor infiltration into surrounding brain. Altered cellular density causes altered imaging findings, but the degree to which cellular density can be quantitatively estimated from imaging is unknown. The purpose of this study was to discover the best MR imaging and processing techniques to make quantitative and spatially specific estimates of cellular density. MATERIALS AND METHODS: We collected stereotactic biopsies in a prospective imaging clinical trial targeting untreated patients with gliomas at our institution undergoing their first resection. The data included preoperative MR imaging with conventional anatomic, diffusion, perfusion, and permeability sequences and quantitative histopathology on biopsy samples. We then used multiple machine learning methodologies to estimate cellular density using local intensity information from the MR images and quantitative cellular density measurements at the biopsy coordinates as the criterion standard. RESULTS: The random forest methodology estimated cellular density with R 2 = 0.59 between predicted and observed values using 4 input imaging sequences chosen from our full set of imaging data (T2, fractional anisotropy, CBF, and area under the curve from permeability imaging). Limiting input to conventional MR images (T1 pre- and postcontrast, T2, and FLAIR) yielded slightly degraded performance (R2 = 0.52). Outputs were also reported as graphic maps. CONCLUSIONS: Cellular density can be estimated with moderate-to-strong correlations using MR imaging inputs. The random forest machine learning model provided the best estimates. These spatially specific estimates of cellular density will likely be useful in guiding both diagnosis and treatment.

Imaging-Based Algorithm for the Local Grading of Glioma.

BACKGROUND AND PURPOSE: Gliomas are highly heterogeneous tumors, and optimal treatment depends on identifying and locating the highest grade disease present. Imaging techniques for doing so are generally not validated against the histopathologic criterion standard. The purpose of this work was to estimate the local glioma grade using a machine learning model trained on preoperative image data and spatially specific tumor samples. The value of imaging in patients with brain tumor can be enhanced if pathologic data can be estimated from imaging input using predictive models. MATERIALS AND METHODS: Patients with gliomas were enrolled in a prospective clinical imaging trial between 2013 and 2016. MR imaging was performed with anatomic, diffusion, permeability, and perfusion sequences, followed by image-guided stereotactic biopsy before resection. An imaging description was developed for each biopsy, and multiclass machine learning models were built to predict the World Health Organization grade. Models were assessed on classification accuracy, Cohen κ, precision, and recall. RESULTS: Twenty-three patients (with 7/9/7 grade II/III/IV gliomas) had analyzable imaging-pathologic pairs, yielding 52 biopsy sites. The random forest method was the best algorithm tested. Tumor grade was predicted at 96% accuracy (κ = 0.93) using 4 inputs (T2, ADC, CBV, and transfer constant from dynamic contrast-enhanced imaging). By means of the conventional imaging only, the overall accuracy decreased (89% overall, κ = 0.79) and 43% of high-grade samples were misclassified as lower-grade disease. CONCLUSIONS: We found that local pathologic grade can be predicted with a high accuracy using clinical imaging data. Advanced imaging data improved this accuracy, adding value to conventional imaging. Confirmatory imaging trials are justified.

Hepatocellular carcinoma response to transcatheter arterial chemoembolisation using automatically generated pre-therapeutic tumour volumes by a random forest-based segmentation protocol.

AIM: To demonstrate the feasibility of correlating pre-therapeutic volumes and residual liver volume (RLV) with clinical outcomes: time to progression (TTP) and overall survival (OS) in hepatocellular carcinoma (HCC) treated with transcatheter arterial chemoembolisation (TACE). MATERIALS AND METHODS: TTP was calculated from a database of 105 patients, receiving first-line treatment with TACE. TTP cut-off for stratifying patients into responders and non-responders was 28 weeks. Pre-treatment tumour and liver volumes were correlated with the TTP and OS following treatment. Univariate cox-regression model was used to assess whether these volumes could predict TTP and/or OS. Kaplan-Meier analysis with log-rank test was used to compare the TTP between high and low volume groups for viable, necrotic, and total tumour. Kaplan-Meier analysis was performed comparing the OS of 10 patients with the longest TTP (mean=122 weeks) in the responder group and 10 patients with the shortest TTP (mean=7 weeks) in the non-responder group. RESULTS: HCC in high tumour volume groups had a shorter TTP than lesions in low tumour volume groups (p=0.05, p=0.04, p=0.02, for enhancing, non-enhancing, total tumour groups, respectively). A negative (correlation coefficient [CC] 0.3) linear correlation between TTP and tumour volumes, and a positive linear correlation between TTP and residual liver volumes were also demonstrated (CC 0.3). Patients with the longest TTP had a higher OS than with the shortest TTP (p=0.03). CONCLUSION: This demonstrates the feasibility of predicting treatment response of HCC to TACE using volumetric measurements of pre-treatment lesion and the feasibility of correlating RLV with TACE outcome data in HCC patients.

Machine learning-based texture analysis for differentiation of large adrenal cortical tumours on CT.

AIM: To compare the efficacy of computed tomography (CT) texture analysis and conventional evaluation by radiologists for differentiation between large adrenal adenomas and carcinomas. MATERIALS AND METHODS: Quantitative CT texture analysis was used to evaluate 54 histopathologically proven adrenal masses (mean size=5.9 cm; range=4.1-10 cm) from 54 patients referred to Anderson Cancer Center from January 2002 through April 2014. The patient group included 32 women (mean age at mass evaluation=59 years) and 22 men (mean age at mass evaluation=61 years). Adrenal lesions seen on precontrast and venous-phase CT images were labelled by three different readers, and the labels were used to generate intensity- and geometry-based textural features. The textural features and the attenuation values were considered as input values for a random forest-based classifier. Similarly, the adrenal lesions were classified by two different radiologists based on morphological criteria. Prediction accuracy and interobserver agreement were compared. RESULTS: The textural predictive model achieved a mean accuracy of 82%, whereas the mean accuracy for the radiologists was 68.5% (p<0.0001). The interobserver agreements between the predictive model and radiologists 1 and 2 were 0.44 (p<0.0005; 95% confidence interval [CI]: 0.25-0.62) and 0.47 (p<0.0005; 95% CI: 0.28-0.66), respectively. The Dice similarity coefficient between the readers' image labels was 0.875±0.04. CONCLUSION: CT texture analysis of large adrenal adenomas and carcinomas is likely to improve CT evaluation of adrenal cortical tumours.

A compressed sensing approach to immobilized nanoparticle localization for superparamagnetic relaxometry.

Superparamagnetic relaxometry (SPMR) exploits the unique magnetic properties of targeted superparamagnetic iron oxide nanoparticles (SPIOs) to detect small numbers of cancer cells. Reconstruction of the spatial distribution of cancer-bound nanoparticles requires solving an ill-posed inverse problem. The current method, multiple source analysis (MSA), uses a least-squares fit to determine the strength and location of a pre-determined number of magnetic dipoles. In this proof-of-concept study, we propose the application of a sparsity averaged reweighting algorithm (SARA) for volumetric reconstruction of immobilized nanoparticle distributions. We first calibrate the parameters that define the location of the sensors in the forward model of measurement physics. Using this optimized model, we evaluated the performance of the algorithms on various configurations of single and multiple point-source phantoms. We investigated the effect of the data fidelity parameter, voxel size, and iterative reweighting on the reconstruction produced by SARA. We found that the calibrated physics model can predict the detected field values within 5% of the measured data. When only a single source was present, both algorithms were able to detect as little as 0.5 µg of immobilized particles. However, when two sources were measured simultaneously, MSA failed to detect sources containing as much as 10 µg of particles, while SARA detected all of the sources containing at least 5 µg of particles. We show that a suitable data fidelity parameter can be selected objectively, and the total magnitude and location of a point source reconstructed by SARA is not sensitive to voxel size. Detection and localization of multiple small clusters of nanoparticles is a crucial step in SPMR-based diagnostic applications. Our algorithm overcomes the need to know the number of dipoles before reconstruction and improves the sensitivity of the reconstruction when multiple sources are present.

Performance Assessment for Brain MR Imaging Registration Methods.

BACKGROUND AND PURPOSE: Clinical brain MR imaging registration algorithms are often made available by commercial vendors without figures of merit. The purpose of this study was to suggest a rational performance comparison methodology for these products. MATERIALS AND METHODS: Twenty patients were imaged on clinical 3T scanners by using 4 sequences: T2-weighted, FLAIR, susceptibility-weighted angiography, and T1 postcontrast. Fiducial landmark sites (n = 1175) were specified throughout these image volumes to define identical anatomic locations across sequences. Multiple registration algorithms were applied by using the T2 sequence as a fixed reference. Euclidean error was calculated before and after each registration and compared with a criterion standard landmark registration. The Euclidean effectiveness ratio is the fraction of Euclidean error remaining after registration, and the statistical effectiveness ratio is similar, but accounts for dispersion and noise. RESULTS: Before registration, error values for FLAIR, susceptibility-weighted angiography, and T1 postcontrast were 2.07 ± 0.55 mm, 2.63 ± 0.62 mm, and 3.65 ± 2.00 mm, respectively. Postregistration, the best error values for FLAIR, susceptibility-weighted angiography, and T1 postcontrast were 1.55 ± 0.46 mm, 1.34 ± 0.23 mm, and 1.06 ± 0.16 mm, with Euclidean effectiveness ratio values of 0.493, 0.181, and 0.096 and statistical effectiveness ratio values of 0.573, 0.352, and 0.929 for rigid mutual information, affine mutual information, and a commercial GE registration, respectively. CONCLUSIONS: We demonstrate a method for comparing the performance of registration algorithms and suggest the Euclidean error, Euclidean effectiveness ratio, and statistical effectiveness ratio as performance metrics for clinical registration algorithms. These figures of merit allow registration algorithms to be rationally compared.

Accelerated magnetic resonance thermometry in the presence of uncertainties.

A model-based information theoretic approach is presented to perform the task of magnetic resonance (MR) thermal image reconstruction from a limited number of observed samples on k-space. The key idea of the proposed approach is to optimally detect samples of k-space that are information-rich with respect to a model of the thermal data acquisition. These highly informative k-space samples can then be used to refine the mathematical model and efficiently reconstruct the image. The information theoretic reconstruction was demonstrated retrospectively in data acquired during MR-guided laser induced thermal therapy (MRgLITT) procedures. The approach demonstrates that locations with high-information content with respect to a model-based reconstruction of MR thermometry may be quantitatively identified. These information-rich k-space locations are demonstrated to be useful as a guide for k-space undersampling techniques. The effect of interactively increasing the predicted number of data points used in the subsampled model-based reconstruction was quantified using the L2-norm of the distance between the subsampled and fully sampled reconstruction. Performance of the proposed approach was also compared with uniform rectilinear subsampling and variable-density Poisson disk subsampling techniques. The proposed subsampling scheme resulted in accurate reconstructions using a small fraction of k-space points, suggesting that the reconstruction technique may be useful in improving the efficiency of thermometry data temporal resolution.

Design and initial evaluation of a treatment planning software system for MRI-guided laser ablation in the brain.

PURPOSE: An open-source software system for planning magnetic resonance (MR)-guided laser-induced thermal therapy (MRgLITT) in brain is presented. The system was designed to provide a streamlined and operator-friendly graphical user interface (GUI) for simulating and visualizing potential outcomes of various treatment scenarios to aid in decisions on treatment approach or feasibility. METHODS: A portable software module was developed on the 3D Slicer platform, an open-source medical imaging and visualization framework. The module introduces an interactive GUI for investigating different laser positions and power settings as well as the influence of patient-specific tissue properties for quickly creating and evaluating custom treatment options. It also provides a common treatment planning interface for use by both open-source and commercial finite element solvers. In this study, an open-source finite element solver for Pennes' bioheat equation is interfaced to the module to provide rapid 3D estimates of the steady-state temperature distribution and potential tissue damage in the presence of patient-specific tissue boundary conditions identified on segmented MR images. RESULTS: The total time to initialize and simulate an MRgLITT procedure using the GUI was [Formula: see text]5 min. Each independent simulation took [Formula: see text]30 s, including the time to visualize the results fused with the planning MRI. For demonstration purposes, a simulated steady-state isotherm contour [Formula: see text] was correlated with MR temperature imaging (N = 5). The mean Hausdorff distance between simulated and actual contours was 2.0 mm [Formula: see text], whereas the mean Dice similarity coefficient was 0.93 [Formula: see text]. CONCLUSIONS: We have designed, implemented, and conducted initial feasibility evaluations of a software tool for intuitive and rapid planning of MRgLITT in brain. The retrospective in vivo dataset presented herein illustrates the feasibility and potential of incorporating fast, image-based bioheat predictions into an interactive virtual planning environment for such procedures.

An inverse problem approach to recovery of in vivo nanoparticle concentrations from thermal image monitoring of MR-guided laser induced thermal therapy.

Quantification of local variations in the optical properties of tumor tissue introduced by the presence of gold-silica nanoparticles (NP) presents significant opportunities in monitoring and control of NP-mediated laser induced thermal therapy (LITT) procedures. Finite element methods of inverse parameter recovery constrained by a Pennes bioheat transfer model were applied to estimate the optical parameters. Magnetic resonance temperature imaging (MRTI) acquired during a NP-mediated LITT of a canine transmissible venereal tumor in brain was used in the presented statistical inverse problem formulation. The maximum likelihood (ML) value of the optical parameters illustrated a marked change in the periphery of the tumor corresponding with the expected location of NP and area of selective heating observed on MRTI. Parameter recovery information became increasingly difficult to infer in distal regions of tissue where photon fluence had been significantly attenuated. Finite element temperature predictions using the ML parameter values obtained from the solution of the inverse problem are able to reproduce the NP selective heating within 5 °C of measured MRTI estimations along selected temperature profiles. Results indicate the ML solution found is able to sufficiently reproduce the selectivity of the NP mediated laser induced heating and therefore the ML solution is likely to return useful optical parameters within the region of significant laser fluence.

Kalman filtered MR temperature imaging for laser induced thermal therapies.

The feasibility of using a stochastic form of Pennes bioheat model within a 3-D finite element based Kalman filter (KF) algorithm is critically evaluated for the ability to provide temperature field estimates in the event of magnetic resonance temperature imaging (MRTI) data loss during laser induced thermal therapy (LITT). The ability to recover missing MRTI data was analyzed by systematically removing spatiotemporal information from a clinical MR-guided LITT procedure in human brain and comparing predictions in these regions to the original measurements. Performance was quantitatively evaluated in terms of a dimensionless L(2) (RMS) norm of the temperature error weighted by acquisition uncertainty. During periods of no data corruption, observed error histories demonstrate that the Kalman algorithm does not alter the high quality temperature measurement provided by MR thermal imaging. The KF-MRTI implementation considered is seen to predict the bioheat transfer with RMS error < 4 for a short period of time, ∆t < 10 s, until the data corruption subsides. In its present form, the KF-MRTI method currently fails to compensate for consecutive for consecutive time periods of data loss ∆t > 10 sec.

Correlation between the temperature dependence of intrinsic MR parameters and thermal dose measured by a rapid chemical shift imaging technique.

In order to investigate simultaneous MR temperature imaging and direct validation of tissue damage during thermal therapy, temperature-dependent signal changes in proton resonance frequency (PRF) shifts, R(2)* values, and T1-weighted amplitudes are measured from one technique in ex vivo tissue. Using a multigradient echo acquisition and the Stieglitz-McBride algorithm, the temperature sensitivity coefficients of these parameters are measured in each tissue at high spatiotemporal resolutions (1.6 x 1.6 x 4 mm 3,≤ 5sec) at the range of 25-61 °C. Non-linear changes in MR parameters are examined and correlated with an Arrhenius rate dose model of thermal damage. Using logistic regression, the probability of changes in these parameters is calculated as a function of thermal dose to determine if changes correspond to thermal damage. Temperature sensitivity of R(2)* and, in some cases, T1-weighted amplitudes are statistically different before and after thermal damage occurred. Significant changes in the slopes of R(2)* as a function of temperature are observed. Logistic regression analysis shows that these changes could be accurately predicted using the Arrhenius rate dose model (Ω = 1.01 ± 0.03), thereby showing that the changes in R(2)* could be direct markers of protein denaturation. Overall, by using a chemical shift imaging technique with simultaneous temperature estimation, R(2)* mapping and T1-W imaging, it is shown that changes in the sensitivity of R(2)* and, to a lesser degree, T1-W amplitudes are measured in ex vivo tissue when thermal damage is expected to occur. These changes could possibly be used for direct validation of thermal damage in contrast to model-based predictions.

Magnetic resonance temperature imaging validation of a bioheat transfer model for laser-induced thermal therapy.

PURPOSE: Magnetic resonance-guided laser-induced thermal therapy (MRgLITT) is currently undergoing initial safety and feasibility clinical studies for the treatment of intracranial lesions in humans. As studies progress towards evaluation of treatment efficacy, predictive computational models may play an important role for prospective 3D treatment planning. The current work critically evaluates a computational model of laser induced bioheat transfer against retrospective multiplanar MR thermal imaging (MRTI) in a canine model of the MRgLITT procedure in the brain. METHODS: A 3D finite element model of the bioheat transfer that couples Pennes equation to a diffusion theory approximation of light transport in tissue is used. The laser source is modelled conformal with the applicator geometry. Dirichlet boundary conditions are used to model the temperature of the actively cooled catheter. The MRgLITT procedure was performed on n = 4 canines using a 1-cm diffusing tip 15-W diode laser (980 nm). A weighted L2norm is used as the metric of comparison between the spatiotemporal MR-derived temperature estimates and model prediction. RESULTS: The normalised error history between the computational models and MRTI was within 1-4 standard deviations of MRTI noise. Active cooling models indicate that the applicator temperature has a strong effect on the maximum temperature reached, but does not significantly decrease the tissue temperature away from the active tip. CONCLUSIONS: Results demonstrate the computational model of the bioheat transfer may provide a reasonable approximation of the laser-tissue interaction, which could be useful for treatment planning, but cannot readily replace MR temperature imaging in a complex environment such as the brain.

Computational modeling and real-time control of patient-specific laser treatment of cancer.

An adaptive feedback control system is presented which employs a computational model of bioheat transfer in living tissue to guide, in real-time, laser treatments of prostate cancer monitored by magnetic resonance thermal imaging. The system is built on what can be referred to as cyberinfrastructure-a complex structure of high-speed network, large-scale parallel computing devices, laser optics, imaging, visualizations, inverse-analysis algorithms, mesh generation, and control systems that guide laser therapy to optimally control the ablation of cancerous tissue. The computational system has been successfully tested on in vivo, canine prostate. Over the course of an 18 min laser-induced thermal therapy performed at M.D. Anderson Cancer Center (MDACC) in Houston, Texas, the computational models were calibrated to intra-operative real-time thermal imaging treatment data and the calibrated models controlled the bioheat transfer to within 5 degrees C of the predetermined treatment plan. The computational arena is in Austin, Texas and managed at the Institute for Computational Engineering and Sciences (ICES). The system is designed to control the bioheat transfer remotely while simultaneously providing real-time remote visualization of the on-going treatment. Post-operative histology of the canine prostate reveal that the damage region was within the targeted 1.2 cm diameter treatment objective.

Transurethral ultrasound applicators with directional heating patterns for prostate thermal therapy: in vivo evaluation using magnetic resonance thermometry.

A catheter-based transurethral ultrasound applicator with angularly directional heating patterns has been designed for prostate thermal therapy and evaluated in canine prostate in vivo using MRI to monitor and assess performance. The ultrasound transducer array (3.5 mm diameter tubular transducers, 180 degrees active sectors, approximately 7.5 MHz) was integrated to a flexible delivery catheter (4 mm OD), and encapsulated within an expandable balloon (35 mm x 10 mm OD, 80 ml min(-1) ambient water) for coupling and cooling of the prostatic urethra. These devices were used to thermally coagulate targeted portions of the canine prostate (n = 2) while using MR thermal imaging (MRTI) to monitor the therapy. MRI was also used for target definition, positioning of the applicator, and evaluation of target viability post-therapy. MRTI was based upon the complex phase-difference mapping technique using an interleaved gradient echo-planar imaging sequence with lipid suppression. MRTI derived temperature distributions, thermal dose exposures, T1-contrast enhanced MR images, and histology of sectioned prostates were used to define destroyed tissue zones and characterize the three-dimensional heating patterns. The ultrasound applicators produced approximately 180 degrees directed zones of thermal coagulation within targeted tissue which extended 15-20 mm radially to the outer boundary of the prostate within 15 min. Transducer activation lengths of 17 mm and 24 mm produced contiguous zones of coagulation extending axially approximately 18 mm and approximately 25 mm from base to apex, respectively. Peak temperatures around 90 degrees C were measured, with approximately 50 degrees C-52 degrees C corresponding to outer boundary t43 = 240 min at approximately 15 min treatment time. These devices are MRI compatible, and when coupled with multiplanar MRTI provide a means for selectively controlling the length and sector angle of therapeutic thermal treatment in the prostate.

Technical developments for cerebral thermal treatment: water-cooled diffusing laser fibre tips and temperature-sensitive MRI using intersecting image planes.

The aim was to determine if water-cooled diffusing tips could produce larger and safer (better controlled) thermal lesions than non-cooled diffusing tips at 980 nm. Thermal lesions were induced in beef myocardium in vitro with and without water cooling using a 980 nm diode laser at various power levels. Seven intracerebral treatments were performed in six canines using water-cooled diffusing tips with four animals having intracerebral transmissible venereal tumours grown from inoculate. Magnetic resonance thermal imaging (MRTI)-based feedback software using a fast, radio frequency-spoiled gradient echo acquisition with two intersecting image planes was used for on-line monitoring and control of treatment and for the evaluation of in vivo laser lesion production. In cases where two-plane MRTI was employed, the maximum calculated temperature was compared in each plane. Using water-cooled tips and 400 micro m core diameter laser diffusing fibres in in vitro beef myocardium, power of up to 9.5 W was applied for 8 min without tip failure. Without cooling, tip failure occurred in under 4 min at 6 W, in under 2 min at 7 W and instantaneously at 8 W. Additionally, char accompanied lesions made with uncooled tips while cooled application resulted in only minimal char at only the highest thermal dose. Achieved lesion cross-sectional diameters in in vitro samples were up to 26.5 x 23.3 mm when water cooling was used. In canine brain and transmissible venereal tumours, up to 18.1 x 21.4 mm lesions were achieved. It is concluded that water cooling allows safe application of higher power to small core diameter diffusing tip fibres, which results in larger thermal lesions than can be achieved without cooling. Two-plane MRTI enhances on-line monitoring and feedback of thermal treatment.

Nanoshell-mediated near-infrared thermal therapy of tumors under magnetic resonance guidance.

Metal nanoshells are a class of nanoparticles with tunable optical resonances. In this article, an application of this technology to thermal ablative therapy for cancer is described. By tuning the nanoshells to strongly absorb light in the near infrared, where optical transmission through tissue is optimal, a distribution of nanoshells at depth in tissue can be used to deliver a therapeutic dose of heat by using moderately low exposures of extracorporeally applied near-infrared (NIR) light. Human breast carcinoma cells incubated with nanoshells in vitro were found to have undergone photothermally induced morbidity on exposure to NIR light (820 nm, 35 W/cm2), as determined by using a fluorescent viability stain. Cells without nanoshells displayed no loss in viability after the same periods and conditions of NIR illumination. Likewise, in vivo studies under magnetic resonance guidance revealed that exposure to low doses of NIR light (820 nm, 4 W/cm2) in solid tumors treated with metal nanoshells reached average maximum temperatures capable of inducing irreversible tissue damage (DeltaT = 37.4 +/- 6.6 degrees C) within 4-6 min. Controls treated without nanoshells demonstrated significantly lower average temperatures on exposure to NIR light (DeltaT < 10 degrees C). These findings demonstrated good correlation with histological findings. Tissues heated above the thermal damage threshold displayed coagulation, cell shrinkage, and loss of nuclear staining, which are indicators of irreversible thermal damage. Control tissues appeared undamaged.

Monitoring of high-intensity focused ultrasound-induced temperature changes in vitro using an interleaved spiral acquisition.

An interleaved, spoiled gradient-echo spiral acquisition technique was implemented to monitor high-intensity focused ultrasound heating of porcine kidney ex vivo by measuring temperature induced phase shifts in the detected MR signal. Echo time, flip angle, repetition time, number of interleaves, and readout time were varied to observes effects on temperature sensitivity and phase-difference noise. The temperature response of the interleaved spiral acquisition was found to be comparable to a spoiled fast gradient-echo sequence of comparable in-plane spatial resolution. However, when imaging with an optimal echo time, spiral acquisition offers dramatically increased temporal resolution for comparable spatial resolution. Magn Reson Med 43:909-912, 2000.