Table of contents

Objective. Fluorescence molecular tomography (FMT) is a promising non-invasive optical molecular imaging technology with strong specificity and sensitivity that has great potential for preclinical and clinical studies in tumor diagnosis, drug development and therapeutic evaluation. However, the strong scattering of photons and insufficient surface measurements make it very challenging to improve the quality of FMT image reconstruction and its practical application for early tumor detection. Therefore, continuous efforts have been made to explore more effective approaches or solutions in the pursuit of high-quality FMT reconstructions. Approach. This review takes a comprehensive overview of advances in imaging methodology for FMT, mainly focusing on two critical issues in FMT reconstructions: improving the accuracy of solving the forward physical model and mitigating the ill-posed nature of the inverse problem from a methodological point of view. More importantly, numerous impressive and practical strategies and methods for improving the quality of FMT reconstruction are summarized. Notably, deep learning methods are discussed in detail to illustrate their advantages in promoting the imaging performance of FMT thanks to large datasets, the emergence of optimized algorithms and the application of innovative networks. Main results. The results demonstrate that the imaging quality of FMT can be effectively promoted by improving the accuracy of optical parameter modeling, combined with prior knowledge, and reducing dimensionality. In addition, the traditional regularization-based methods and deep neural network-based methods, especially end-to-end deep networks, can enormously alleviate the ill-posedness of the inverse problem and improve the quality of FMT image reconstruction. Significance. This review aims to illustrate a variety of effective and practical methods for the reconstruction of FMT images that may benefit future research. Furthermore, it may provide some valuable research ideas and directions for FMT in the future, and could promote, to a certain extent, the development of FMT and other methods of optical tomography.

Objective. Kilovoltage computed tomography (kVCT) is the cornerstone of radiotherapy treatment planning for delineating tissues and towards dose calculation. For the former, kVCT provides excellent contrast and signal-to-noise ratio. For the latter, kVCT may have greater uncertainty in determining relative electron density (${\rho }_{e}$) and proton stopping power ratio (SPR). Conversely, megavoltage CT (MVCT) may result in superior dose calculation accuracy. The purpose of this work was to convert kVCT HU to MVCT HU using deep learning to obtain higher accuracy ${\rho }_{e}$ and SPR. Approach. Tissue-mimicking phantoms were created to compare kVCT- and MVCT-determined ${\rho }_{e}$ and SPR to physical measurements. Using 100 head-and-neck datasets, an unpaired deep learning model was trained to learn the relationship between kVCTs and MVCTs, creating synthetic MVCTs (sMVCTs). Similarity metrics were calculated between kVCTs, sMVCTs, and MVCTs in 20 test datasets. An anthropomorphic head phantom containing bone-mimicking material with known composition was scanned to provide an independent determination of ${\rho }_{e}$ and SPR accuracy by sMVCT. Main results. In tissue-mimicking bone, ${\rho }_{e}$ errors were 2.20% versus 0.19% and SPR errors were 4.38% versus 0.22%, for kVCT versus MVCT, respectively. Compared to MVCT, in vivo mean difference (MD) values were 11 and 327 HU for kVCT and 2 and 3 HU for sMVCT in soft tissue and bone, respectively. ${\rho }_{e}$ MD decreased from 1.3% to 0.35% in soft tissue and 2.9% to 0.13% in bone, for kVCT and sMVCT, respectively. SPR MD decreased from 1.8% to 0.24% in soft tissue and 6.8% to 0.16% in bone, for kVCT and sMVCT, respectively. Relative to physical measurements, ${\rho }_{e}$ and SPR error in anthropomorphic bone decreased from 7.50% and 7.48% for kVCT to <1% for both MVCT and sMVCT. Significance. Deep learning can be used to map kVCT to sMVCT, suggesting higher accuracy ${\rho }_{e}$ and SPR is achievable with sMVCT versus kVCT.

Objective. Synthetic transmit aperture (STA) ultrasound imaging is well known for ideal focusing in the full field of view. However, it suffers from low signal-to-noise ratio (SNR) and low frame rate, because each transducer element must be activated individually. In our previous study, we encoded all the transducer elements with partial Hadamard matrix and reconstructed the complete STA dataset with compressed sensing (CS) algorithm (CS-STA). As all the elements are activated in each transmission and the number of transmissions is smaller than that of STA, this method can achieve higher SNR and higher frame rate. Its main drawback is the time-consuming CS reconstruction (∼hours). In this study, we propose to accelerate the complete STA dataset reconstruction with minimal l₂-norm least squares method. Approach. Partial Hadamard apodized plane wave (PW) transmissions were performed to acquire the PW dataset. Thereafter, the complete STA dataset can be reconstructed from the PW dataset with minimal l₂-norm least squares method. Due to the orthogonality of partial Hadamard matrix, the minimal l₂-norm least squares solution can be easily calculated. Main results. The proposed method is tested with simulation data and experimental phantom and in-vivo data. The results demonstrate that the proposed method achieves ∼5 × 10³ times faster reconstruction speed than CS algorithm. The simulation results demonstrate that the proposed method is capable of achieving the same accuracy as the conventional CS-STA method for the STA dataset reconstruction. The simulations, phantom and in-vivo experiments show that the proposed method is capable of improving the generalized contrast-to-noise ratio (gCNR) and SNR with maintained spatial resolution and fewer transmissions, compared with STA. Significance. In conclusion, the improved image quality and reduced computational time of LS-STA pave the way for its real-time applications in the clinics.

Objective. To develop a bremsstrahlung target and megavoltage (MV) x-ray irradiation platform for ultrahigh dose-rate (UHDR) irradiation of small-animals on the Advanced Rare Isotope Laboratory (ARIEL) electron linac (e-linac) at TRIUMF. Approach. An electron-to-photon converter design for UHDR radiotherapy (RT) was centered around optimization of a tantalum–aluminum (Ta–Al) explosion-bonded target. Energy deposition within a homogeneous water-phantom and the target itself were evaluated using EGSnrc and FLUKA MC codes, respectively, for various target thicknesses (0.5–1.5 mm), beam energies (E_e− = 8, 10 MeV) and electron (Gaussian) beam sizes ($2\sigma$ = 2–10 mm). Depth dose-rates in a 3D-printed mouse phantom were also calculated to infer the compatibility of the 10 MV dose distributions for FLASH-RT in small-animal models. Coupled thermo-mechanical FEA simulations in ANSYS were subsequently used to inform the stress–strain conditions and fatigue life of the target assembly. Main results. Dose-rates of up to 128 Gy s⁻¹ at the phantom surface, or 85 Gy s⁻¹ at 1 cm depth, were obtained for a 1 × 1 cm² field size, 1 mm thick Ta target and 7.5 cm source-to-surface distance using the FLASH-mode beam (E_e− = 10 MeV, 2 $\sigma$ = 5 mm, P = 1 kW); furthermore, removal of the collimation assembly and using a shorter (3.5 cm) SSD afforded dose-rates >600 Gy s⁻¹, albeit at the expense of field conformality. Target temperatures were maintained below the tantalum, aluminum and cooling-water thresholds of 2000 °C, 300 °C and 100 °C, respectively, while the aluminum strain behavior remained everywhere elastic and helped ensure   the converter survives its prescribed 5 yr operational lifetime. Significance. Effective design iteration, target cooling and failure mitigation have culminated in a robust target compatible with intensive transient (FLASH) and steady-state (diagnostic) applications. The ARIEL UHDR photon source will facilitate FLASH-RT experiments concerned with sub-second, pulsed or continuous beam irradiations at dose rates in excess of 40 Gy s⁻¹.

Objective. To develop two combined clinical-radiomics models of pericoronary adipose tissue (PCAT) for the presence and characterization of non-calcified plaques on non-contrast CT scan. Approach. Altogether, 431 patients undergoing Coronary Computed Tomography Angiography from March 2019 to June 2021 who had complete data were enrolled, including 173 patients with non-calcified plaques of the right coronary artery(RCA) and 258 with no abnormality. PCAT was segmented around the proximal RCA on non-contrast CT scan (calcium score acquisition). Two best models were established by screening features and classifiers respectively using FeAture Explorer software. Model 1 distinguished normal coronary arteries from those with non-calcified plaques, and model 2 distinguished vulnerable plaques in non-calcified plaques. Performance was assessed by the area under the receiver operating characteristic curve (AUC-ROC). Main results. 4 and 9 features were selected for the establishment of the radiomics model respectively through Model 1 and 2. In the test group, the AUC values, sensitivity, specificity and accuracy were 0.833%, 78.3%, 80.8%, 76.6% and 0.7467%, 75.0%, 77.8%, 73.5%, respectively. The combined model including radiomics features and independent clinical factors yielded an AUC, sensitivity, specificity and accuracy of 0.896%, 81.4%, 86.5%, 77.9% for model 1 and 0.752%, 75.0%, 77.8%, 73.5% for model 2. Significance. The combined clinical-radiomics models based on non-contrast CT images of PCAT had good diagnostic efficacy for non-calcified and vulnerable plaques.

Purpose. We present a microscopic mechanism that accounts for the outward burst of 'cold' ion species (IS) in a high-energy particle track due to coupling with 'hot' non-ion species (NIS). IS refers to radiolysis products of ionized molecules, whereas NIS refers to non-ionized excitations of molecules in a medium. The interaction is mediated by a quantized field of acoustic phonons, a channel that allows conversion of thermal energy of NIS to kinetic energy of IS, a flow of heat from the outer to the inner core of the track structure. Methods. We perform step-by-step Monte Carlo (MC) simulations of ionizing radiation track structures in water to score the spatial coordinates and energy depositions that form IS and NIS at atto-second time scales. We subsequently calculate the resulting temperature profiles of the tracks with MC track structure simulations and verify the results analytically using the Rutherford scattering formulation. These temperature profiles are then used as boundary conditions in a series of multi-scale atomistic molecular dynamic (MD) simulations that describe the sudden expansion and enhanced diffusive broadening of tracks initiated by the non-equilibrium spectrum of high-energy IS. We derive a stochastic coarse-grained Langevin equation of motion for IS from first-principle MD to describe the irreversible femto-second flow of thermal energy pumping from NIS to IS, mediated by quantized fields of acoustic phonons. A pair-wise Lennard-Jones potential implemented in a classical MD is then employed to validate the results calculated from the Langevin equation. Results. We demonstrate the coexistence of 'hot' NIS with 'cold' IS in the radiation track structures right after their generation. NIS, concentrated within nano-scale volumes wrapping around IS, are the main source of intensive heat-waves and the outward burst of IS due to femto-second time scale IS-NIS coupling. By comparing the transport of IS coupled to NIS with identical configurations of non-interacting IS in thermal equilibrium at room temperature, we demonstrate that the energy gain of IS due to the surrounding hot nanoscopic volumes of NIS significantly increases their effective diffusion constants. Comparing the average track separation and the time scale calculated for a deposited dose of 10 Gy and a dose rate of 40 Gy s⁻¹, typical values used in FLASH ultra high dose rate (UHDR) experiments, we find that the sudden expansion of tracks and ballistic transport proposed in this work strengthens the hypothesis of inter-track correlations recently introduced to interpret mitigation of the biological responses at the FLASH-UHDR (Abolfath et al 2020 Med. Phys.47, 6551–6561). Conclusions. The much higher diffusion constants predicted in the present model suggest higher inter-track chemical reaction rates at FLASH-UHDR, as well as lower intra-track reaction rates. This study explains why research groups relying on the current Monte Carlo frameworks have reported negligible inter-track overlaps, simply because of underestimation of the diffusion constants. We recommend incorporation of the IS-NIS coupling and heat exchange in all MC codes to enable these tool-kits to appropriately model reaction-diffusion rates at FLASH-UHDR. Novelty. To introduce a hypothetical pathway of outward burst of radiolysis products driven by highly localized thermal spikes wrapping around them and to investigate the interplay of the non-equilibrium spatio-temporal distribution of the chemical activities of diffusive high-energy particle tracks on inter-track correlations at FLASH-UHDR.

Objective. Next generation online and real-time adaptive radiotherapy workflows require precise particle transport simulations in sub-second times, which is unfeasible with current analytical pencil beam algorithms (PBA) or Monte Carlo (MC) methods. We present a deep learning based millisecond speed dose calculation algorithm (DoTA) accurately predicting the dose deposited by mono-energetic proton pencil beams for arbitrary energies and patient geometries. Approach. Given the forward-scattering nature of protons, we frame 3D particle transport as modeling a sequence of 2D geometries in the beam's eye view. DoTA combines convolutional neural networks extracting spatial features (e.g. tissue and density contrasts) with a transformer self-attention backbone that routes information between the sequence of geometry slices and a vector representing the beam's energy, and is trained to predict low noise MC simulations of proton beamlets using 80 000 different head and neck, lung, and prostate geometries. Main results. Predicting beamlet doses in 5 ± 4.9 ms with a very high gamma pass rate of 99.37 ± 1.17% (1%, 3 mm) compared to the ground truth MC calculations, DoTA significantly improves upon analytical pencil beam algorithms both in precision and speed. Offering MC accuracy 100 times faster than PBAs for pencil beams, our model calculates full treatment plan doses in 10–15 s depending on the number of beamlets (800–2200 in our plans), achieving a 99.70 ± 0.14% (2%, 2 mm) gamma pass rate across 9 test patients. Significance. Outperforming all previous analytical pencil beam and deep learning based approaches, DoTA represents a new state of the art in data-driven dose calculation and can directly compete with the speed of even commercial GPU MC approaches. Providing the sub-second speed required for adaptive treatments, straightforward implementations could offer similar benefits to other steps of the radiotherapy workflow or other modalities such as helium or carbon treatments.

Objective. Lead-doped scintillator dosimeters may be well suited for the dosimetry of FLASH-capable x-ray radiotherapy beams. Our study explores the dose rate dependence and temporal resolution of scintillators that makes them promising in the accurate detection of ultrahigh dose-rate (UHDR) x-rays. Approach. We investigated the response of scintillators with four material compositions to UHDR x-rays produced by a conventional x-ray tube. Scintillator output was measured using the HYPERSCINT-RP100 dosimetry research platform. Measurements were acquired at high frame rates (400 fps) which allowed for accurate dose measurements of sub-second radiation exposures from 1 to 100 ms. Dose-rate dependence was assessed by scaling tube current of the x-ray tube. Scintillator measurements were validated against Monte Carlo simulations of the probe geometries and UHDR x-ray system. Calibration factors converting dose-to-medium to dose-to-water were obtained from simulation data of plastic and lead-doped scintillator materials. Main Results. The results of this work suggest that lead-doped scintillators were dose-rate independent for UHDR x-rays from 1.1 to 40.1 Gy s⁻¹ and capable of measuring conventional radiotherapy dose-rates (0.1 Gy s⁻¹) at extended distance from the x-ray focal spot. Dose-to-water measured with a 5% lead-doped scintillator detector agreed with simulations within 0.6%. Significance. Lead-doped scintillators may be a valuable tool for the accurate real-time dosimetry of FLASH-capable UHDR x-ray beams.

Objective. Current segmentation practice for thoracic cancer RT considers the whole heart as a single organ despite increased risks of cardiac toxicities from irradiation of specific cardiac substructures. Segmenting up to 15 different cardiac substructures can be a very time-intensive process, especially due to their different volume sizes and anatomical variations amongst different patients. In this work, a new deep learning (DL)-based mutual enhancing strategy is introduced for accurate and automatic segmentation, especially of smaller substructures such as coronary arteries. Approach. Our proposed method consists of three subnetworks: retina U-net, classification module, and segmentation module. Retina U-net is used as a backbone network architecture that aims to learn deep features from the whole heart. Whole heart feature maps from retina U-net are then transferred to four different sets of classification modules to generate classification localization maps of coronary arteries, great vessels, chambers of the heart, and valves of the heart. Each classification module is in sync with its corresponding subsequent segmentation module in a bootstrapping manner, allowing them to share their encoding paths to generate a mutual enhancing strategy. We evaluated our method on three different datasets: institutional CT datasets (55 subjects) 2) publicly available Multi-Modality Whole Heart Segmentation (MM-WHS) challenge datasets (120 subjects), and Automated Cardiac Diagnosis Challenge (ACDC) datasets (100 subjects). For institutional datasets, we performed five-fold cross-validation on training data (45 subjects) and performed inference on separate hold-out data (10 subjects). For each subject, 15 cardiac substructures were manually contoured by a resident physician and evaluated by an attending radiation oncologist. For the MM-WHS dataset, we trained the network on 100 datasets and performed an inference on a separate hold-out dataset with 20 subjects, each with 7 cardiac substructures. For ACDC datasets, we performed five-fold cross-validation on 100 datasets, each with 3 cardiac substructures. We compared the proposed method against four different network architectures: 3D U-net, mask R-CNN, mask scoring R-CNN, and proposed network without classification module. Segmentation accuracies were statistically compared through dice similarity coefficient, Jaccard, 95% Hausdorff distance, mean surface distance, root mean square distance, center of mass distance, and volume difference. Main results. The proposed method generated cardiac substructure segmentations with significantly higher accuracy (P < 0.05) for small substructures, especially for coronary arteries such as left anterior descending artery (CA-LADA) and right coronary artery (CA-RCA) in comparison to four competing methods. For large substructures (i.e. chambers of the heart), our method yielded comparable results to mask scoring R-CNN method, resulting in significantly (P < 0.05) improved segmentation accuracy in comparison to 3D U-net and mask R-CNN. Significance. A new DL-based mutual enhancing strategy was introduced for automatic segmentation of cardiac substructures. Overall results of this work demonstrate the ability of the proposed method to improve segmentation accuracies of smaller substructures such as coronary arteries without largely compromising the segmentation accuracies of larger substructures. Fast and accurate segmentations of up to 15 substructures can possibly be used as a tool to rapidly generate substructure segmentations followed by physicians' reviews to improve clinical workflow.

Objective. The design of neutron moderators for BNCT treatment units currently relies on parametric approaches, which yield quality results but are ultimately limited by human imagination. Efficient but non-intuitive design solutions may thus be missed out. This limitation needs to be addressed. Approach. To overcome this limitation, we propose to use a topology optimization algorithm coupled with a state-of-the-art Monte-Carlo transport code. This approach recently proved capable of finding complex optimal configurations of particle propagators with limited human intervention. Main results. In this study, we apply this algorithmic solution to optimize some heavy-water neutron moderators for a specific AB-BNCT treatment unit. The moderators thus generated are compact yet succeed in limiting the exposure of patient's healthy tissues to levels below recommended limits. They present subtle, original geometries inaccessible to standard parametric approaches or human intuition. Significance. This approach could be used to automatically fit the design of a BNCT moderator to the location and shape of the tumor or to the morphology of the patient to be treated, opening a path for more targeted BNCT treatment.

Objective. Using Monte-Carlo simulations, we evaluated the physical performance of a hypothetical state-of-the-art clinical PET scanner with adaptive axial field-of-view (AFOV) based on the validated GATE model of the Siemens Biograph Vision^TM PET/CT scanner. Approach. Vision consists of 16 compact PET rings, each consisting of 152 mini-blocks of 5 × 5 Lutetium Oxyorthosilicate crystals (3.2 × 3.2 × 20 mm³). The Vision 25.6 cm AFOV was extended by adopting (i) a sparse mini-block ring (SBR) configuration of 49.6 cm AFOV, with all mini-block rings interleaved with 16 mm axial gaps, or (ii) a sparse mini-block checkerboard (SCB) configuration of 51.2 cm AFOV, with all mini-blocks interleaved with gaps of 16 mm (transaxial) × 16 mm (axial) width in checkerboard pattern. For sparse configurations, a 'limited' continuous bed motion (limited-CBM) acquisition was employed to extend AFOVs by 2.9 cm. Spatial resolution, sensitivity, image quality (IQ), NECR and scatter fraction were assessed per NEMA NU2-2012. Main Results. All IQ phantom spheres were distinguishable with all configurations. SBR and SCB percent contrast recovery (% CR) and background variability (% BV) were similar (p-value > 0.05). Compared to Vision, SBR and SCB %CRs were similar (p-values > 0.05). However, SBR and SCB %BVs were deteriorated by 30% and 26% respectively (p-values < 0.05). SBR, SCB and Vision exhibited system sensitivities of 16.6, 16.8, and 15.8 kcps MBq⁻¹, NECRs of 311 kcps @35 kBq cc⁻¹, 266 kcps @25.8 kBq cc⁻¹, and 260 kcps @27.8 kBq cc⁻¹, and scatter fractions of 31.2%, 32.4%, and 32.6%, respectively. SBR and SCB exhibited a smoother sensitivity reduction and noise enhancement rate from AFOV center to its edges. SBR and SCB attained comparable spatial resolution in all directions (p-value > 0.05), yet, up to 1.5 mm worse than Vision (p-values < 0.05). Significance. The proposed sparse configurations may offer a clinically adoptable solution for cost-effective adaptive AFOV PET with either highly-sensitive or long-AFOV acquisitions.

Objective. In this study, we propose a novel approach designed to take advantage of the Cerenkov light angular dependency to perform a direct measurement of an external beam irradiation angle. Approach. A Cerenkov probe composed of a 10 mm long filtered sensitive volume of clear PMMA optical fibre was built. Both filtered and raw Cerenkov signals from the transport fibre were collected through a single 1 mm diameter transport fibre. An independent plastic scintillation detector composed of 10 mm BCF12 scintillating fibre was also used for simultaneous dose measurements. A first series of measurements aimed at validating the ability to account for the Cerenkov electron energy spectrum dependency by simultaneously measuring the deposited dose, thus isolating signal variations resulting from the angular dependency. Angular calibration curve for fixed dose irradiations and incident angle measurements using electron and photon beams where also achieved. Main results. The beam nominal energy was found to have a significant impact on the shapes of the angular calibration curves. This can be linked to the electron energy spectrum dependency of the Cerenkov emission cone. Irradiation angle measurements exhibit an absolute mean error of 1.86° and 1.02° at 6 and 18 MV, respectively. Similar results were obtained with electron beams and the absolute mean error reaches 1.97°, 1.66°, 1.45° and 0.95° at 9, 12, 16 and 20 MeV, respectively. Reducing the numerical aperture of the Cerenkov probe leads to an increased angular dependency for the lowest energy while no major changes were observed at higher energy. This allowed irradiation angle measurements at 6 MeV with a mean absolute error of 4.82°. Significance. The detector offers promising perspectives as a potential tool for future quality assurance applications in radiotherapy, especially for stereotactic radiosurgery (SRS), magnetic resonance image-guided radiotherapy (MRgRT) and brachytherapy applications.

This paper studies the impact of tiny changes in region-of-interest (ROI) tomography system matrices on the variance of the reconstructed ROI. In small-scale and medium-scale examples, the variance in the reconstructed ROI was estimated for different system matrices. The results revealed a striking and counterintuitive phenomenon: a tiny change in the system matrix can dramatically affect the variance of the ROI estimate. In one of our examples, a decrease of 0.1% in one element out of hundreds of thousands of the system matrix resulted in a systematic reduction of the variance inside the ROI, and by a factor of 5 to 10 for some pixels. Our results agree with a recently proven theorem about the ability of additional measurements to reduce the variance in ROI tomography.