MRI-only based material mass density and relative stopping power estimation via deep learning for proton therapy: a preliminary study.

Magnetic Resonance Imaging (MRI) is increasingly being used in treatment planning due to its superior soft tissue contrast, which is useful for tumor and soft tissue delineation compared to computed tomography (CT). However, MRI cannot directly provide mass density or relative stopping power (RSP) maps, which are required for calculating proton radiotherapy doses. Therefore, the integration of artificial intelligence (AI) into MRI-based treatment planning to estimate mass density and RSP directly from MRI has generated significant interest. A deep learning (DL) based framework was developed to establish a voxel-wise correlation between MR images and mass density as well as RSP. To facilitate the study, five tissue substitute phantoms were created, representing different tissues such as skin, muscle, adipose tissue, 45% hydroxyapatite (HA), and spongiosa bone. The composition of these phantoms was based on information from ICRP reports. Additionally, two animal tissue phantoms, simulating pig brain and liver, were prepared for DL training purposes. The phantom study involved the development of two DL models. The first model utilized clinical T1 and T2 MRI scans as input, while the second model incorporated zero echo time (ZTE) MRI scans. In the patient application study, two more DL models were trained: one using T1 and T2 MRI scans as input, and another model incorporating synthetic dual-energy computed tomography (sDECT) images to provide accurate bone tissue information. The DECT empirical model was used as a reference to evaluate the proposed models in both phantom and patient application studies. The DECT empirical model was selected as the reference for evaluating the proposed models in both phantom and patient application studies. In the phantom study, the DL model based on T1, and T2 MRI scans demonstrated higher accuracy in estimating mass density and RSP for skin, muscle, adipose tissue, brain, and liver. The mean absolute percentage errors (MAPE) were 0.42%, 0.14%, 0.19%, 0.78%, and 0.26% for mass density, and 0.30%, 0.11%, 0.16%, 0.61%, and 0.23% for RSP, respectively. The DL model incorporating ZTE MRI further improved the accuracy of mass density and RSP estimation for 45% HA and spongiosa bone, with MAPE values of 0.23% and 0.09% for mass density, and 0.19% and 0.07% for RSP, respectively. These results demonstrate the feasibility of using an MRI-only approach combined with DL methods for mass density and RSP estimation in proton therapy treatment planning. By employing this approach, it is possible to obtain the necessary information for proton radiotherapy directly from MRI scans, eliminating the need for additional imaging modalities.

Multimodal imaging-based material mass density estimation for proton therapy using supervised deep learning.

OBJECTIVE: Mapping CT number to material property dominates the proton range uncertainty. This work aims to develop a physics-constrained deep learning-based multimodal imaging (PDMI) framework to integrate physics, deep learning, MRI, and advanced dual-energy CT (DECT) to derive accurate patient mass density maps. METHODS: Seven tissue substitute MRI phantoms were used for validation including adipose, brain, muscle, liver, skin, spongiosa, 45% hydroxyapatite (HA) bone. MRI images were acquired using T1 weighted Dixon and T2 weighted short tau inversion recovery sequences. Training inputs are from MRI and twin-beam dual-energy images acquired at 120 kVp with gold/tin filters. The feasibility investigation included an empirical model and four residual networks (ResNet) derived from different training inputs and strategies by PDMI framework. PRN-MR-DE and RN-MR-DE denote ResNet (RN) trained with and without a physics constraint (P) using MRI (MR) and DECT (DE) images. PRN-DE stands for RN trained with a physics constraint using only DE images. A retrospective study using institutional patient data was also conducted to investigate the feasibility of the proposed framework. RESULTS: For the tissue surrogate study, PRN-MR-DE, PRN-DE, and RN-MR-DE result in mean mass density errors: -0.72%/2.62%/-3.58% for adipose; -0.03%/-0.61%/-0.18% for muscle; -0.58%/-1.36%/-4.86% for 45% HA bone. The retrospective patient study indicated that PRN-MR-DE predicted the densities of soft tissue and bone within expected intervals based on the literature survey, while PRN-DE generated large density deviations. CONCLUSION: The proposed PDMI framework can generate accurate mass density maps using MRI and DECT images. The supervised learning can further enhance model efficacy, making PRN-MR-DE outperform RN-MR-DE. The patient investigation also shows that the framework can potentially improve proton range uncertainty with accurate patient mass density maps. ADVANCES IN KNOWLEDGE: PDMI framework is proposed for the first time to inform deep learning models by physics insights and leverage the information from MRI to derive accurate mass density maps.

Quantifying progression and regression of thrombotic risk in experimental atherosclerosis.

Currently, there are no generally applicable noninvasive methods for defining the relationship between atherosclerotic vascular damage and risk of focal thrombosis. Herein, we demonstrate methods to delineate the progression and regression of vascular damage in response to an atherogenic diet by quantifying the in vivo accumulation of semipermeable 200-300 nm perfluorocarbon core nanoparticles (PFC-NP) in ApoE null mouse plaques with [(19)F] magnetic resonance spectroscopy (MRS). Permeability to PFC-NP remained minimal until 12 weeks on diet, then increased rapidly following 12 weeks, but regressed to baseline within 8 weeks after diet normalization. Markedly accelerated clotting (53.3% decrease in clotting time) was observed in carotid artery preparations of fat-fed mice subjected to photochemical injury as defined by the time to flow cessation. For all mice on and off diet, an inverse linear relationship was observed between the permeability to PFC-NP and accelerated thrombosis (P = 0.02). Translational feasibility for quantifying plaque permeability and vascular damage in vivo was demonstrated with clinical 3 T MRI of PFC-NP accumulating in plaques of atherosclerotic rabbits. These observations suggest that excessive permeability to PFC-NP may indicate prothrombotic risk in damaged atherosclerotic vasculature, which resolves within weeks after dietary therapy.

Antithrombin nanoparticles improve kidney reperfusion and protect kidney function after ischemia-reperfusion injury.

In the extension phase of acute kidney injury, microvascular thrombosis, inflammation, vasoconstriction, and vascular endothelial cell dysfunction promote progressive damage to renal parenchyma after reperfusion. In this study, we hypothesized that direct targeting and pharmaceutical knockdown of activated thrombin at the sites of injury with a selective nanoparticle (NP)-based thrombin inhibitor, PPACK (phenylalanine-proline-arginine-chloromethylketone), would improve kidney reperfusion and protect renal function after transient warm ischemia in rodent models. Saline- or plain NP-treated animals were employed as controls. In vivo 19F magnetic resonance imaging revealed that kidney nonreperfusion was evident within 3 h after global kidney reperfusion at 34 ± 13% area in the saline group and 43 ± 12% area in the plain NP group and substantially reduced to 17 ± 4% (∼50% decrease, P < 0.05) in the PPACK NP pretreatment group. PPACK NP pretreatment prevented an increase in serum creatinine concentration within 24 h after ischemia-reperfusion, reflecting preserved renal function. Histologic analysis illustrated substantially reduced intrarenal thrombin accumulation within 24 h after reperfusion for PPACK NP-treated kidneys (0.11% ± 0.06%) compared with saline-treated kidneys (0.58 ± 0.37%). These results suggest a direct role for thrombin in the pathophysiology of AKI and a nanomedicine-based preventative strategy for improving kidney reperfusion after transient warm ischemia.

Balanced UTE-SSFP for 19F MR imaging of complex spectra.

PURPOSE: A novel technique for highly sensitive detection of multiresonant fluorine imaging agents was designed and tested with the use of dual-frequency 19F/1H ultrashort echo times (UTE) sampled with a balanced steady-state free precession (SSFP) pulse sequence and three-dimensional (3D) radial readout. METHODS: Feasibility of 3D radial balanced UTE-SSFP imaging was demonstrated for a phantom comprising liquid perfluorooctyl bromide (PFOB). Sensitivity of the pulse sequence was measured and compared with other sequences imaging the PFOB (CF2 )6 line group including UTE radial gradient-echo (GRE) at α = 30°, as well as Cartesian GRE, balanced SSFP, and fast spin-echo (FSE). The PFOB CF3 peak was also sampled with FSE. RESULTS: The proposed balanced UTE-SSFP technique exhibited a relative detection sensitivity of 51 µmolPFOB(-1) min(-1/2) (α = 30°), at least twice that of other sequence types with either 3D radial (UTE GRE: 20 µmolPFOB(-1) min(-1/2) ) or Cartesian k-space filling (GRE: 12 µmolPFOB(-1) min(-1/2) ; FSE: 16 µmolPFOB(-1) min(-1/2) ; balanced SSFP: 23 µmolPFOB(-1) min(-1/2) ). In vivo imaging of angiogenesis-targeted PFOB nanoparticles was demonstrated in a rabbit model of cancer on a clinical 3 Tesla scanner. CONCLUSION: A new dual 19F/1H balanced UTE-SSFP sequence manifests high SNR, with detection sensitivity more than two-fold better than traditional techniques, and alleviates imaging problems caused by dephasing in complex spectra.

Improved quantitative (19) F MR molecular imaging with flip angle calibration and B1 -mapping compensation.

PURPOSE: To improve (19) F flip angle calibration and compensate for B1 inhomogeneities in quantitative (19) F MRI of sparse molecular epitopes with perfluorocarbon (PFC) nanoparticle (NP) emulsion contrast agents. MATERIALS AND METHODS: Flip angle sweep experiments on PFC-NP point source phantoms with three custom-designed (19) F/(1) H dual-tuned coils revealed a difference in required power settings for (19) F and (1) H nuclei, which was used to calculate a calibration ratio specific for each coil. An image-based correction technique was developed using B1 -field mapping on (1) H to correct for (19) F and (1) H images in two phantom experiments. RESULTS: Optimized (19) F peak power differed significantly from that of (1) H power for each coil (P < 0.05). A ratio of (19) F/(1) H power settings yielded a coil-specific and spatially independent calibration value (surface: 1.48 ± 0.06; semicylindrical: 1.71 ± 0.02, single-turn-solenoid: 1.92 ± 0.03). (1) H-image-based B1 correction equalized the signal intensity of (19) F images for two identical (19) F PFC-NP samples placed in different parts of the field, which were offset significantly by ~66% (P < 0.001), before correction. CONCLUSION: (19) F flip angle calibration and B1 -mapping compensations to the (19) F images employing the more abundant (1) H signal as a basis for correction resulted in a significant change in the quantification of sparse (19) F MR signals from targeted PFC NP emulsions.

A fibrin-specific thrombolytic nanomedicine approach to acute ischemic stroke.

AIM: To develop a fibrin-specific urokinase nanomedicine thrombolytic agent. MATERIALS & METHODS: In vitro fibrin-clot dissolution studies were utilized to develop and characterize simultaneous coupling and loading of anti-fibrin monoclonal antibody and urokinase onto perfluorocarbon nanoparticle (NP) surface. In vivo pharmacokinetics and fibrin-specific targeting of the nanolytic agent was studied in dogs. RESULTS: Simultaneous coupling of up to 40 anti-fibrin antibodies and 400 urokinase enzymes per perfluorocarbon NP produced an effective targeted nanolytic agent with no significant surface protein-protein interference. Fibrin clot dissolution was not improved by increasing homing capacity from 10 to 40 antibodies/NP, but increasing enzymatic payload from 100 to 400/NP resulted in maximized lytic effect. Fluorescent microscopy showed that rhodamine-labeled urokinase nanoparticles densely decorated the intraluminal thrombus in canine clots in vivo analogous to the fibrin pattern, while an irrelevant-targeted agent had negligible binding. CONCLUSION: This agent offers a vascularly constrained, simple to administer, low-dose nanomedicine approach that may present an attractive alternative for treating acute stroke victims.