Learning Multiparametric Biomarkers for Assessing MR-Guided Focused Ultrasound Treatment of Malignant Tumors

Treatment efficacy prediction of focused ultrasound therapies using multi-parametric magnetic resonance imaging

Magnetic resonance guided focused ultrasound (MRgFUS) is one of the most attractive emerging minimally invasive procedures for breast cancer, which induces localized hyperthermia, resulting in tumor cell death. Accurately assessing the post-ablation viability of all treated tumor tissue and surrounding margins immediately after MRgFUS thermal therapy residual tumor tissue is essential for evaluating treatment efficacy. While both thermal and vascular MRI-derived biomarkers are currently used to assess treatment efficacy, currently, no adequately accurate methods exist for the in vivo determination of tissue viability during treatment. The non-perfused volume (NPV) acquired three or more days following MRgFUS thermal ablation treatment is most correlated with the gold standard of histology. However, its delayed timing impedes real-time guidance for the treating clinician during the procedure. We present a robust deep-learning framework that leverages multiparametric MR imaging acquired during treatment to predict treatment efficacy. The network uses qualtitative T1, T2 weighted images and MR temperature image derived metrics to predict the three day post-ablation NPV. To validate the proposed approach, an ablation study was conducted on a dataset (N=6) of VX2 tumor model rabbits that had undergone MRgFUS ablation. Using a deep learning framework, we evaluated which of the acquired MRI inputs were most predictive of treatment efficacy as compared to the expert radiologist annotated 3 day post-treatment images.

A Non-Contrast Multi-Parametric MRI Biomarker for Assessment of MR-Guided Focused Ultrasound Thermal Therapies

OBJECTIVE

We present the development of a non-contrast multi-parametric magnetic resonance (MPMR) imaging biomarker to assess treatment outcomes for magnetic resonance-guided focused ultrasound (MRgFUS) ablations of localized tumors. Images obtained immediately following MRgFUS ablation were inputs for voxel-wise supervised learning classifiers, trained using registered histology as a label for thermal necrosis.

METHODS

VX2 tumors in New Zealand white rabbits quadriceps were thermally ablated using an MRgFUS system under 3 T MRI guidance. Animals were re-imaged three days post-ablation and euthanized. Histological necrosis labels were created by 3D registration between MR images and digitized H&E segmentations of thermal necrosis to enable voxel-wise classification of necrosis. Supervised MPMR classifier inputs included maximum temperature rise, cumulative thermal dose (CTD), post-FUS differences in T2-weighted images, and apparent diffusion coefficient, or ADC, maps. A logistic regression, support vector machine, and random forest classifier were trained in red a leave-one-out strategy in test data from four subjects.

RESULTS

In the validation dataset, the MPMR classifiers achieved higher recall and Dice than a clinically adopted 240 cumulative equivalent minutes at 43 °C (CEM 43) threshold (0.43) in all subjects. The average Dice scores of overlap with the registered histological label for the logistic regression (0.63) and support vector machine (0.63) MPMR classifiers were within 6% of the acute contrast-enhanced non-perfused volume (0.67).

CONCLUSIONS

Voxel-wise registration of MPMR data to histological outcomes facilitated supervised learning of an accurate non-contrast MR biomarker for MRgFUS ablations in a rabbit VX2 tumor model.

Simultaneous proton resonance frequency T1 - MR shear wave elastography for MR-guided focused ultrasound multiparametric treatment monitoring

PURPOSE

To develop an efficient MRI pulse sequence to simultaneously measure multiple parameters that have been shown to correlate with tissue nonviability following thermal therapies.

METHODS

A 3D segmented EPI pulse sequence was used to simultaneously measure proton resonance frequency shift (PRFS) MR thermometry (MRT), T1 relaxation time, and shear wave velocity induced by focused ultrasound (FUS) push pulses. Experiments were performed in tissue mimicking gelatin phantoms and ex vivo bovine liver. Using a carefully designed FUS triggering scheme, a heating duty cycle of approximately 65% was achieved by interleaving FUS ablation pulses with FUS push pulses to induce shear waves in the tissue.

RESULTS

In phantom studies, temperature increases measured with PRFS MRT and increases in T1 correlated with decreased shear wave velocity, consistent with material softening with increasing temperature. During ablation in ex vivo liver, temperature increase measured with PRFS MRT initially correlated with increasing T1 and decreasing shear wave velocity, and after tissue coagulation with decreasing T1 and increasing shear wave velocity. This is consistent with a previously described hysteresis in T1 versus PRFS curves and increased tissue stiffness with tissue coagulation.

CONCLUSION

An efficient approach for simultaneous and dynamic measurements of PRSF, T1 , and shear wave velocity during treatment is presented. This approach holds promise for providing co-registered dynamic measures of multiple parameters, which correlates to tissue nonviability during and following thermal therapies, such as FUS.

Diffusion-weighted MRI with deep learning for visualizing treatment results of MR-guided HIFU ablation of uterine fibroids

OBJECTIVES

No method is available to determine the non-perfused volume (NPV) repeatedly during magnetic resonance-guided high-intensity focused ultrasound (MR-HIFU) ablations of uterine fibroids, as repeated acquisition of contrast-enhanced T1-weighted (CE-T1w) scans is inhibited by safety concerns. The objective of this study was to develop and test a deep learning-based method for translation of diffusion-weighted imaging (DWI) into synthetic CE-T1w scans, for monitoring MR-HIFU treatment progression.

METHODS

The algorithm was retrospectively trained and validated on data from 33 and 20 patients respectively who underwent an MR-HIFU treatment of uterine fibroids between June 2017 and January 2019. Postablation synthetic CE-T1w images were generated by a deep learning network trained on paired DWI and reference CE-T1w scans acquired during the treatment procedure. Quantitative analysis included calculation of the Dice coefficient of NPVs delineated on synthetic and reference CE-T1w scans. Four MR-HIFU radiologists assessed the outcome of MR-HIFU treatments and NPV ratio based on the synthetic and reference CE-T1w scans.

RESULTS

Dice coefficient of NPVs was 71% (± 22%). The mean difference in NPV ratio was 1.4% (± 22%) and not statistically significant (p = 0.79). Absolute agreement of the radiologists on technical treatment success on synthetic and reference CE-T1w scans was 83%. NPV ratio estimations on synthetic and reference CE-T1w scans were not significantly different (p = 0.27).

CONCLUSIONS

Deep learning-based synthetic CE-T1w scans derived from intraprocedural DWI allow gadolinium-free visualization of the predicted NPV, and can potentially be used for repeated gadolinium-free monitoring of treatment progression during MR-HIFU therapy for uterine fibroids.

KEY POINTS

• Synthetic CE-T1w scans can be derived from diffusion-weighted imaging using deep learning. • Synthetic CE-T1w scans may be used for visualization of the NPV without using a contrast agent directly after MR-HIFU ablations of uterine fibroids.

Histology to 3D in vivo MR registration for volumetric evaluation of MRgFUS treatment assessment biomarkers

Advances in imaging and early cancer detection have increased interest in magnetic resonance (MR) guided focused ultrasound (MRgFUS) technologies for cancer treatment. MRgFUS ablation treatments could reduce surgical risks, preserve organ tissue and function, and improve patient quality of life. However, surgical resection and histological analysis remain the gold standard to assess cancer treatment response. For non-invasive ablation therapies such as MRgFUS, the treatment response must be determined through MR imaging biomarkers. However, current MR biomarkers are inconclusive and have not been rigorously evaluated against histology via accurate registration. Existing registration methods rely on anatomical features to directly register in vivo MR and histology. For MRgFUS applications in anatomies such as liver, kidney, or breast, anatomical features that are not caused by the treatment are often insufficient to drive direct registration. We present a novel MR to histology registration workflow that utilizes intermediate imaging and does not rely on anatomical MR features being visible in histology. The presented workflow yields an overall registration accuracy of 1.00 ± 0.13 mm. The developed registration pipeline is used to evaluate a common MRgFUS treatment assessment biomarker against histology. Evaluating MR biomarkers against histology using this registration pipeline will facilitate validating novel MRgFUS biomarkers to improve treatment assessment without surgical intervention. While the presented registration technique has been evaluated in a MRgFUS ablation treatment model, this technique could be potentially applied in any tissue to evaluate a variety of therapeutic options.