Prognostic value of different discretization parameters in 18fluorodeoxyglucose positron emission tomography radiomics of oropharyngeal squamous cell carcinoma.

Purpose: We aim to interrogate the role of positron emission tomography (PET) image discretization parameters on the prognostic value of radiomic features in patients with oropharyngeal cancer. Approach: A prospective clinical trial (NCT01908504) enrolled patients with oropharyngeal squamous cell carcinoma (N=69; mixed HPV status) undergoing definitive radiotherapy and evaluated intra-treatment 18fluorodeoxyglucose PET as a potential imaging biomarker of early metabolic response. The primary tumor volume was manually segmented by a radiation oncologist on PET/CT images acquired two weeks into treatment (20 Gy). From this, 54 radiomic texture features were extracted. Two image discretization techniques-fixed bin number (FBN) and fixed bin size (FBS)-were considered to evaluate systematic changes in the bin number ({32, 64, 128, 256} gray levels) and bin size ({0.10, 0.15, 0.22, 0.25} bin-widths). For each discretization-specific radiomic feature space, an LASSO-regularized logistic regression model was independently trained to predict residual and/or recurrent disease. The model training was based on Monte Carlo cross-validation with a 20% testing hold-out, 50 permutations, and minor-class up-sampling to account for imbalanced outcomes data. Performance differences among the discretization-specific models were quantified via receiver operating characteristic curve analysis. A final parameter-optimized logistic regression model was developed by incorporating different settings parameterizations into the same model. Results: FBN outperformed FBS in predicting residual and/or recurrent disease. The four FBN models achieved AUC values of 0.63, 0.61, 0.65, and 0.62 for 32, 64, 128, and 256 gray levels, respectively. By contrast, the average AUC of the four FBS models was 0.53. The parameter-optimized model, comprising features joint entropy (FBN = 64) and information measure correlation 1 (FBN = 128), achieved an AUC of 0.70. Kaplan-Meier analyses identified these features to be associated with disease-free survival (p=0.0158 and p=0.0180, respectively; log-rank test). Conclusions: Our findings suggest that the prognostic value of individual radiomic features may depend on feature-specific discretization parameter settings.

Radiomics on spatial-temporal manifolds via Fokker-Planck dynamics.

BACKGROUND: Delta radiomics is a high-throughput computational technique used to describe quantitative changes in serial, time-series imaging by considering the relative change in radiomic features of images extracted at two distinct time points. Recent work has demonstrated a lack of prognostic signal of radiomic features extracted using this technique. We hypothesize that this lack of signal is due to the fundamental assumptions made when extracting features via delta radiomics, and that other methods should be investigated. PURPOSE: The purpose of this work was to show a proof-of-concept of a new radiomics paradigm for sparse, time-series imaging data, where features are extracted from a spatial-temporal manifold modeling the time evolution between images, and to assess the prognostic value on patients with oropharyngeal cancer (OPC). METHODS: To accomplish this, we developed an algorithm to mathematically describe the relationship between two images acquired at time t = 0 $t = 0$ and t > 0 $t > 0$ . These images serve as boundary conditions of a partial differential equation describing the transition from one image to the other. To solve this equation, we propagate the position and momentum of each voxel according to Fokker-Planck dynamics (i.e., a technique common in statistical mechanics). This transformation is driven by an underlying potential force uniquely determined by the equilibrium image. The solution generates a spatial-temporal manifold (3 spatial dimensions + time) from which we define dynamic radiomic features. First, our approach was numerically verified by stochastically sampling dynamic Gaussian processes of monotonically decreasing noise. The transformation from high to low noise was compared between our Fokker-Planck estimation and simulated ground-truth. To demonstrate feasibility and clinical impact, we applied our approach to 18F-FDG-PET images to estimate early metabolic response of patients (n = 57) undergoing definitive (chemo)radiation for OPC. Images were acquired pre-treatment and 2-weeks intra-treatment (after 20 Gy). Dynamic radiomic features capturing changes in texture and morphology were then extracted. Patients were partitioned into two groups based on similar dynamic radiomic feature expression via k-means clustering and compared by Kaplan-Meier analyses with log-rank tests (p < 0.05). These results were compared to conventional delta radiomics to test the added value of our approach. RESULTS: Numerical results confirmed our technique can recover image noise characteristics given sparse input data as boundary conditions. Our technique was able to model tumor shrinkage and metabolic response. While no delta radiomics features proved prognostic, Kaplan-Meier analyses identified nine significant dynamic radiomic features. The most significant feature was Gray-Level-Size-Zone-Matrix gray-level variance (p = 0.011), which demonstrated prognostic improvement over its corresponding delta radiomic feature (p = 0.722). CONCLUSIONS: We developed, verified, and demonstrated the prognostic value of a novel, physics-based radiomics approach over conventional delta radiomics via data assimilation of quantitative imaging and differential equations.

Quantifying U-Net uncertainty in multi-parametric MRI-based glioma segmentation by spherical image projection.

BACKGROUND: Uncertainty quantification in deep learning is an important research topic. For medical image segmentation, the uncertainty measurements are usually reported as the likelihood that each pixel belongs to the predicted segmentation region. In potential clinical applications, the uncertainty result reflects the algorithm's robustness and supports the confidence and trust of the segmentation result when the ground-truth result is absent. For commonly studied deep learning models, novel methods for quantifying segmentation uncertainty are in demand. PURPOSE: To develop a U-Net segmentation uncertainty quantification method based on spherical image projection of multi-parametric MRI (MP-MRI) in glioma segmentation. METHODS: The projection of planar MRI data onto a spherical surface is equivalent to a nonlinear image transformation that retains global anatomical information. By incorporating this image transformation process in our proposed spherical projection-based U-Net (SPU-Net) segmentation model design, multiple independent segmentation predictions can be obtained from a single MRI. The final segmentation is the average of all available results, and the variation can be visualized as a pixel-wise uncertainty map. An uncertainty score was introduced to evaluate and compare the performance of uncertainty measurements. The proposed SPU-Net model was implemented on the basis of 369 glioma patients with MP-MRI scans (T1, T1-Ce, T2, and FLAIR). Three SPU-Net models were trained to segment enhancing tumor (ET), tumor core (TC), and whole tumor (WT), respectively. The SPU-Net model was compared with (1) the classic U-Net model with test-time augmentation (TTA) and (2) linear scaling-based U-Net (LSU-Net) segmentation models in terms of both segmentation accuracy (Dice coefficient, sensitivity, specificity, and accuracy) and segmentation uncertainty (uncertainty map and uncertainty score). RESULTS: The developed SPU-Net model successfully achieved low uncertainty for correct segmentation predictions (e.g., tumor interior or healthy tissue interior) and high uncertainty for incorrect results (e.g., tumor boundaries). This model could allow the identification of missed tumor targets or segmentation errors in U-Net. Quantitatively, the SPU-Net model achieved the highest uncertainty scores for three segmentation targets (ET/TC/WT): 0.826/0.848/0.936, compared to 0.784/0.643/0.872 using the U-Net with TTA and 0.743/0.702/0.876 with the LSU-Net (scaling factor = 2). The SPU-Net also achieved statistically significantly higher Dice coefficients, underscoring the improved segmentation accuracy. CONCLUSION: The SPU-Net model offers a powerful tool to quantify glioma segmentation uncertainty while improving segmentation accuracy. The proposed method can be generalized to other medical image-related deep-learning applications for uncertainty evaluation.

Performance evaluation of rural water environment governance based on AHP: a case study of the Beitang River Basin.

Rural water environment governance in China still lacks a systematic and comprehensive assessment protocol to help analyze and improve such governance performance. The Analytic Hierarchy Process (AHP) method was employed in this study to build a governance assessment system that integrates ecological conditions, water pollution control, and public satisfaction. To cover these topics, the assessment system is composed of an indicator layer that is customized to rural water environment governance in China. The Beitang River, located in the rural region of Hangzhou, Zhejiang, China, was presented as a case study. Field investigation provided raw data for this assessment. A questionnaire survey was conducted to interview local residents on the governance performance. An additional survey with executives who played major roles in the governance was performed to reconstruct a water environment assessment on the Beitang River prior to the governance, in order to highlight the effects of the governance through contrast. The results showed consistency in the questionnaire survey and the assessment system. The AHP assessment system was able to reflect the improvement in the water quality, river ecology, and residential welfare after the governance, and suggested limits and future directions in the following upgrade programs for the river basin.

Development of a multi-feature-combined model: proof-of-concept with application to local failure prediction of post-SBRT or surgery early-stage NSCLC patients.

Objective: To develop a Multi-Feature-Combined (MFC) model for proof-of-concept in predicting local failure (LR) in NSCLC patients after surgery or SBRT using pre-treatment CT images. This MFC model combines handcrafted radiomic features, deep radiomic features, and patient demographic information in an integrated machine learning workflow. Methods: The MFC model comprised three key steps. (1) Extraction of 92 handcrafted radiomic features from the GTV segmented on pre-treatment CT images. (2) Extraction of 512 deep radiomic features from pre-trained U-Net encoder. (3) The extracted handcrafted radiomic features, deep radiomic features, along with 4 patient demographic information (i.e., gender, age, tumor volume, and Charlson comorbidity index), were concatenated as a multi-dimensional input to the classifiers for LR prediction. Two NSCLC patient cohorts from our institution were investigated: (1) the surgery cohort includes 83 patients with segmentectomy or wedge resection (7 LR), and (2) the SBRT cohort includes 84 patients with lung SBRT (9 LR). The MFC model was developed and evaluated independently for both cohorts, and was subsequently compared against the prediction models based on only handcrafted radiomic features (R models), patient demographic information (PI models), and deep learning modeling (DL models). ROC with AUC was adopted to evaluate model performance with leave-one-out cross-validation (LOOCV) and 100-fold Monte Carlo random validation (MCRV). The t-test was performed to identify the statistically significant differences. Results: In LOOCV, the AUC range (surgery/SBRT) of the MFC model was 0.858-0.895/0.868-0.913, which was higher than the three other models: 0.356-0.480/0.322-0.650 for PI models, 0.559-0.618/0.639-0.682 for R models, and 0.809/0.843 for DL models. In 100-fold MCRV, the MFC model again showed the highest AUC results (surgery/SBRT): 0.742-0.825/0.888-0.920, which were significantly higher than PI models: 0.464-0.564/0.538-0.628, R models: 0.557-0.652/0.551-0.732, and DL models: 0.702/0.791. Conclusion: We successfully developed an MFC model that combines feature information from multiple sources for proof-of-concept prediction of LR in patients with surgical and SBRT early-stage NSCLC. Initial results suggested that incorporating pre-treatment patient information from multiple sources improves the ability to predict the risk of local failure.

Generation, validation, and benchmarking of a commercial independent Monte Carlo calculation beam model for multi-target SRS.

BACKGROUND: Dosimetric validation of single isocenter multi-target radiosurgery plans is difficult due to conditions of electronic disequilibrium and the simultaneous irradiation of multiple off-axis lesions dispersed throughout the volume. Here we report the benchmarking of a customizable Monte Carlo secondary dose calculation algorithm specific for multi-target radiosurgery which future users may use to guide their commissioning and clinical implementation. PURPOSE: To report the generation, validation, and clinical benchmarking of a volumetric Monte Carlo (MC) dose calculation beam model for single isocenter radiosurgery of intracranial multi-focal disease. METHODS: The beam model was prepared within SciMoCa (ScientificRT, Munich Germany), a commercial independent dose calculation software, with the aim of broad availability via the commercial software for use with single isocenter radiosurgery. The process included (1) definition & acquisition of measurement data required for beam modeling, (2) tuning model parameters to match measurements, (3) validation of the beam model via independent measurements and end-to-end testing, and finally, (4) clinical benchmarking and validation of beam model utility in a patient specific QA setting. We utilized a 6X Flattening-Filter-Free photon beam from a TrueBeam STX linear accelerator (Siemens Healthineers, Munich Germany). RESULTS: In addition to the measured data required for standard IMRT/VMAT (depth dose, central axis profiles & output factors, leaf gap), beam modeling and validation for single-isocenter SRS required central axis and off axis (5 cm & 9 cm) small field output factors and comparison between measurement and simulation of backscatter with aperture for jaw much greater than MLCs. Validation end-to-end measurements included SRS MapCHECK in StereoPHAN geometry (2%/1 mm Gamma = 99.2% ±â€¯2.2%), and OSL & scintillator measurements in anthropomorphic STEEV phantom (6 targets, volume = 0.1-4.1cc, distance from isocenter = 1.2-7.9 cm) for which mean difference was -1.9% ±â€¯2.2%. For 10 patient cases, MC for individual PTVs was -0.8% ±â€¯1.5%, -1.3% ±â€¯1.7%, and -0.5% ±â€¯1.8% for mean dose, D95%, and D1%, respectively. This corresponded to custom passing rates action limits per AAPM TG-218 guidelines of ±5.2%, ±6.4%, and ±6.3%, respectively. CONCLUSIONS: The beam modeling, validation, and clinical action criteria outlined here serves as a benchmark for future users of the customized beam model within SciMoCa for single isocenter radiosurgery of multi-focal disease.

A radiomics-incorporated deep ensemble learning model for multi-parametric MRI-based glioma segmentation.

Objective.To develop a deep ensemble learning (DEL) model with radiomics spatial encoding execution for improved glioma segmentation accuracy using multi-parametric magnetic resonance imaging (mp-MRI).Approach.This model was developed using 369 glioma patients with a four-modality mp-MRI protocol: T1, contrast-enhanced T1 (T1-Ce), T2, and FLAIR. In each modality volume, a 3D sliding kernel was implemented across the brain to capture image heterogeneity: 56 radiomic features were extracted within the kernel, resulting in a fourth-order tensor. Each radiomic feature can then be encoded as a 3D image volume, namely a radiomic feature map (RFM). For each patient, all RFMs extracted from all four modalities were processed using principal component analysis for dimension reduction, and the first four principal components (PCs) were selected. Next, a DEL model comprised of four U-Net sub-models was trained for the segmentation of a region-of-interest: each sub-model utilizes the mp-MRI and one of the four PCs as a five-channel input for 2D execution. Last, four softmax probability results given by the DEL model were superimposed and binarized using Otsu's method as the segmentation results. Three DEL models were trained to segment the enhancing tumor (ET), tumor core (TC), and whole tumor (WT), respectively. The segmentation results given by the proposed ensemble were compared to the mp-MRI-only U-Net results.Main Results.All three radiomics-incorporated DEL models were successfully implemented: compared to the mp-MRI-only U-net results, the dice coefficients of ET (0.777 → 0.817), TC (0.742 → 0.757), and WT (0.823 → 0.854) demonstrated improvement. The accuracy, sensitivity, and specificity results demonstrated similar patterns.Significance.The adopted radiomics spatial encoding execution enriches the image heterogeneity information that leads to the successful demonstration of the proposed DEL model, which offers a new tool for mp-MRI-based medical image segmentation.

Stereotactic Radiosurgery vs Conventional Radiotherapy for Localized Vertebral Metastases of the Spine: Phase 3 Results of NRG Oncology/RTOG 0631 Randomized Clinical Trial.

Importance: Spine metastasis can be treated with high-dose radiation therapy with advanced delivery technology for long-term tumor and pain control. Objective: To assess whether patient-reported pain relief was improved with stereotactic radiosurgery (SRS) as compared with conventional external beam radiotherapy (cEBRT) for patients with 1 to 3 sites of vertebral metastases. Design, Setting, and Participants: In this randomized clinical trial, patients with 1 to 3 vertebral metastases were randomized 2:1 to the SRS or cEBRT groups. This NRG 0631 phase 3 study was performed as multi-institutional enrollment within NRG Oncology. Eligibility criteria included the following: (1) solitary vertebral metastasis, (2) 2 contiguous vertebral levels involved, or (3) maximum of 3 separate sites. Each site may involve up to 2 contiguous vertebral bodies. A total of 353 patients enrolled in the trial, and 339 patients were analyzed. This analysis includes data extracted on March 9, 2020. Interventions: Patients randomized to the SRS group were treated with a single dose of 16 or 18 Gy (to convert to rad, multiply by 100) given to the involved vertebral level(s) only, not including any additional spine levels. Patients assigned to cEBRT were treated with 8 Gy given to the involved vertebra plus 1 additional vertebra above and below. Main Outcomes and Measures: The primary end point was patient-reported pain response defined as at least a 3-point improvement on the Numerical Rating Pain Scale (NRPS) without worsening in pain at the secondary site(s) or the use of pain medication. Secondary end points included treatment-related toxic effects, quality of life, and long-term effects on vertebral bone and spinal cord. Results: A total of 339 patients (mean [SD] age of SRS group vs cEBRT group, respectively, 61.9 [13.1] years vs 63.7 [11.9] years; 114 [54.5%] male in SRS group vs 70 [53.8%] male in cEBRT group) were analyzed. The baseline mean (SD) pain score at the index vertebra was 6.06 (2.61) in the SRS group and 5.88 (2.41) in the cEBRT group. The primary end point of pain response at 3 months favored cEBRT (41.3% for SRS vs 60.5% for cEBRT; difference, -19 percentage points; 95% CI, -32.9 to -5.5; 1-sided P = .99; 2-sided P = .01). Zubrod score (a measure of performance status ranging from 0 to 4, with 0 being fully functional and asymptomatic, and 4 being bedridden) was the significant factor influencing pain response. There were no differences in the proportion of acute or late adverse effects. Vertebral compression fracture at 24 months was 19.5% with SRS and 21.6% with cEBRT (P = .59). There were no spinal cord complications reported at 24 months. Conclusions and Relevance: In this randomized clinical trial, superiority of SRS for the primary end point of patient-reported pain response at 3 months was not found, and there were no spinal cord complications at 2 years after SRS. This finding may inform further investigation of using spine radiosurgery in the setting of oligometastases, where durability of cancer control is essential. Trial Registration: ClinicalTrials.gov Identifier: NCT00922974.

A neural ordinary differential equation model for visualizing deep neural network behaviors in multi-parametric MRI-based glioma segmentation.

PURPOSE: To develop a neural ordinary differential equation (ODE) model for visualizing deep neural network behavior during multi-parametric MRI-based glioma segmentation as a method to enhance deep learning explainability. METHODS: By hypothesizing that deep feature extraction can be modeled as a spatiotemporally continuous process, we implemented a novel deep learning model, Neural ODE, in which deep feature extraction was governed by an ODE parameterized by a neural network. The dynamics of (1) MR images after interactions with the deep neural network and (2) segmentation formation can thus be visualized after solving the ODE. An accumulative contribution curve (ACC) was designed to quantitatively evaluate each MR image's utilization by the deep neural network toward the final segmentation results. The proposed Neural ODE model was demonstrated using 369 glioma patients with a 4-modality multi-parametric MRI protocol: T1, contrast-enhanced T1 (T1-Ce), T2, and FLAIR. Three Neural ODE models were trained to segment enhancing tumor (ET), tumor core (TC), and whole tumor (WT), respectively. The key MRI modalities with significant utilization by deep neural networks were identified based on ACC analysis. Segmentation results by deep neural networks using only the key MRI modalities were compared to those using all four MRI modalities in terms of Dice coefficient, accuracy, sensitivity, and specificity. RESULTS: All Neural ODE models successfully illustrated image dynamics as expected. ACC analysis identified T1-Ce as the only key modality in ET and TC segmentations, while both FLAIR and T2 were key modalities in WT segmentation. Compared to the U-Net results using all four MRI modalities, the Dice coefficient of ET (0.784→0.775), TC (0.760→0.758), and WT (0.841→0.837) using the key modalities only had minimal differences without significance. Accuracy, sensitivity, and specificity results demonstrated the same patterns. CONCLUSION: The Neural ODE model offers a new tool for optimizing the deep learning model inputs with enhanced explainability. The presented methodology can be generalized to other medical image-related deep-learning applications.

Uncertainties in the dosimetric heterogeneity correction and its potential effect on local control in lung SBRT.

Objective. Dose calculation in lung stereotactic body radiation therapy (SBRT) is challenging due to the low density of the lungs and small volumes. Here we assess uncertainties associated with tissue heterogeneities using different dose calculation algorithms and quantify potential associations with local failure for lung SBRT.Approach. 164 lung SBRT plans were used. The original plans were prepared using Pencil Beam Convolution (PBC, n = 8) or Anisotropic Analytical Algorithm (AAA, n = 156). Each plan was recalculated with AcurosXB (AXB) leaving all plan parameters unchanged. A subset (n = 89) was calculated with Monte Carlo to verify accuracy. Differences were calculated for the planning target volume (PTV) and internal target volume (ITV) Dmean[Gy], D99%[Gy], D95%[Gy], D1%[Gy], and V100%[%]. Dose metrics were converted to biologically effective doses (BED) usingα/ß= 10Gy. Regression analysis was performed for AAA plans investigating the effects of various parameters on the extent of the dosimetric differences. Associations between the magnitude of the differences for all plans and outcome were investigated using sub-distribution hazards analysis.Main results. For AAA cases, higher energies increased the magnitude of the difference (ΔDmean of -3.6%, -5.9%, and -9.1% for 6X, 10X, and 15X, respectively), as did lung volume (ΔD99% of -1.6% per 500cc). Regarding outcome, significant hazard ratios (HR) were observed for the change in the PTV and ITV D1% BEDs upon univariate analysis (p = 0.042, 0.023, respectively). When adjusting for PTV volume and prescription, the HRs for the change in the ITV D1% BED remained significant (p = 0.039, 0.037, respectively).Significance. Large differences in dosimetric indices for lung SBRT can occur when transitioning to advanced algorithms. The majority of the differences were not associated with local failure, although differences in PTV and ITV D1% BEDs were associated upon univariate analysis. This shows uncertainty in near maximal tumor dose to potentially be predictive of treatment outcome.

View-sharing for 4D magnetic resonance imaging with randomized projection-encoding enables improvements of respiratory motion imaging for treatment planning in abdominothoracic radiotherapy.

Background and Purpose: The accuracy and precision of radiation therapy are dependent on the characterization of organ-at-risk and target motion. This work aims to demonstrate a 4D magnetic resonance imaging (MRI) method for improving spatial and temporal resolution in respiratory motion imaging for treatment planning in abdominothoracic radiotherapy. Materials and Methods: The spatial and temporal resolution of phase-resolved respiratory imaging is improved by considering a novel sampling function based on quasi-random projection-encoding and peripheral k-space view-sharing. The respiratory signal is determined directly from k-space, obviating the need for an external surrogate marker. The average breathing curve is used to optimize spatial resolution and temporal blurring by limiting the extent of data sharing in the Fourier domain. Improvements in image quality are characterized by evaluating changes in signal-to-noise ratio (SNR), resolution, target detection, and level of artifact. The method is validated in simulations, in a dynamic phantom, and in-vivo imaging. Results: Sharing of high-frequency k-space data, driven by the average breathing curve, improves spatial resolution and reduces artifacts. Although equal sharing of k-space data improves resolution and SNR in stationary features, phases with large temporal changes accumulate significant artifacts due to averaging of high frequency features. In the absence of view-sharing, no averaging and detection artifacts are observed while spatial resolution is degraded. Conclusions: The use of a quasi-random sampling function, with view-sharing driven by the average breathing curve, provides a feasible method for self-navigated 4D-MRI at improved spatial resolution.

Input feature design and its impact on the performance of deep learning models for predicting fluence maps in intensity-modulated radiation therapy.

Objective. Deep learning (DL) models for fluence map prediction (FMP) have great potential to reduce treatment planning time in intensity-modulated radiation therapy (IMRT) by avoiding the lengthy inverse optimization process. This study aims to improve the rigor of input feature design in a DL-FMP model by examining how different designs of input features influence model prediction performance.Approach. This study included 231 head-and-neck intensity-modulated radiation therapy patients. Three input feature designs were investigated. The first design (D1) assumed that information of all critical structures from all beam angles should be combined to predict fluence maps. The second design (D2) assumed that local anatomical information was sufficient for predicting radiation intensity of a beamlet at a respective beam angle. The third design (D3) assumed the need for both local anatomical information and inter-beam modulation to predict radiation intensity values of the beamlets that intersect at a voxel. For each input design, we tailored the DL model accordingly. All models were trained using the same set of ground truth plans (GT plans). The plans generated by DL models (DL plans) were analyzed using key dose-volume metrics. One-way ANOVA with multiple comparisons correction (Bonferroni method) was performed (significance level = 0.05).Main results. For PTV-related metrics, all DL plans had significantly higher maximum dose (p < 0.001), conformity index (p < 0.001), and heterogeneity index (p < 0.001) compared to GT plans, with D2 being the worst performer. Meanwhile, except for cord+5 mm (p < 0.001), DL plans of all designs resulted in OAR dose metrics that are comparable to those of GT plans.Significance. Local anatomical information contains most of the information that DL models need to predict fluence maps for clinically acceptable OAR sparing. Input features from beam angles are needed to achieve the best PTV coverage. These results provide valuable insights for further improvement of DL-FMP models and DL models in general.

Impact of image quality on radiomics applications.

Radiomics features extracted from medical images have been widely reported to be useful in the patient specific outcome modeling for variety of assessment and prediction purposes. Successful application of radiomics features as imaging biomarkers, however, is dependent on the robustness of the approach to the variation in each step of the modeling workflow. Variation in the input image quality is one of the main sources that impacts the reproducibility of radiomics analysis when a model is applied to broader range of medical imaging data. The quality of medical image is generally affected by both the scanner related factors such as image acquisition/reconstruction settings and the patient related factors such as patient motion. This article aimed to review the published literatures in this field that reported the impact of various imaging factors on the radiomics features through the change in image quality. The literatures were categorized by different imaging modalities and also tabulated based on the imaging parameters and the class of radiomics features included in the study. Strategies for image quality standardization were discussed based on the relevant literatures and recommendations for reducing the impact of image quality variation on the radiomics in multi-institutional clinical trial were summarized at the end of this article.

Post-Radiotherapy PET Image Outcome Prediction by Deep Learning Under Biological Model Guidance: A Feasibility Study of Oropharyngeal Cancer Application.

Purpose: To develop a method of biologically guided deep learning for post-radiation 18FDG-PET image outcome prediction based on pre-radiation images and radiotherapy dose information. Methods: Based on the classic reaction-diffusion mechanism, a novel biological model was proposed using a partial differential equation that incorporates spatial radiation dose distribution as a patient-specific treatment information variable. A 7-layer encoder-decoder-based convolutional neural network (CNN) was designed and trained to learn the proposed biological model. As such, the model could generate post-radiation 18FDG-PET image outcome predictions with breakdown biological components for enhanced explainability. The proposed method was developed using 64 oropharyngeal patients with paired 18FDG-PET studies before and after 20-Gy delivery (2 Gy/day fraction) by intensity-modulated radiotherapy (IMRT). In a two-branch deep learning execution, the proposed CNN learns specific terms in the biological model from paired 18FDG-PET images and spatial dose distribution in one branch, and the biological model generates post-20-Gy 18FDG-PET image prediction in the other branch. As in 2D execution, 718/233/230 axial slices from 38/13/13 patients were used for training/validation/independent test. The prediction image results in test cases were compared with the ground-truth results quantitatively. Results: The proposed method successfully generated post-20-Gy 18FDG-PET image outcome prediction with breakdown illustrations of biological model components. Standardized uptake value (SUV) mean values in 18FDG high-uptake regions of predicted images (2.45 ± 0.25) were similar to ground-truth results (2.51 ± 0.33). In 2D-based Gamma analysis, the median/mean Gamma Index (<1) passing rate of test images was 96.5%/92.8% using the 5%/5 mm criterion; such result was improved to 99.9%/99.6% when 10%/10 mm was adopted. Conclusion: The developed biologically guided deep learning method achieved post-20-Gy 18FDG-PET image outcome predictions in good agreement with ground-truth results. With the breakdown biological modeling components, the outcome image predictions could be used in adaptive radiotherapy decision-making to optimize personalized plans for the best outcome in the future.

Fast four-dimensional cone-beam computed tomography reconstruction using deformable convolutional networks.

BACKGROUND: Although four-dimensional cone-beam computed tomography (4D-CBCT) is valuable to provide onboard image guidance for radiotherapy of moving targets, it requires a long acquisition time to achieve sufficient image quality for target localization. To improve the utility, it is highly desirable to reduce the 4D-CBCT scanning time while maintaining high-quality images. Current motion-compensated methods are limited by slow speed and compensation errors due to the severe intraphase undersampling. PURPOSE: In this work, we aim to propose an alternative feature-compensated method to realize the fast 4D-CBCT with high-quality images. METHODS: We proposed a feature-compensated deformable convolutional network (FeaCo-DCN) to perform interphase compensation in the latent feature space, which has not been explored by previous studies. In FeaCo-DCN, encoding networks extract features from each phase, and then, features of other phases are deformed to those of the target phase via deformable convolutional networks. Finally, a decoding network combines and decodes features from all phases to yield high-quality images of the target phase. The proposed FeaCo-DCN was evaluated using lung cancer patient data. RESULTS: (1) FeaCo-DCN generated high-quality images with accurate and clear structures for a fast 4D-CBCT scan; (2) 4D-CBCT images reconstructed by FeaCo-DCN achieved 3D tumor localization accuracy within 2.5 mm; (3) image reconstruction is nearly real time; and (4) FeaCo-DCN achieved superior performance by all metrics compared to the top-ranked techniques in the AAPM SPARE Challenge. CONCLUSION: The proposed FeaCo-DCN is effective and efficient in reconstructing 4D-CBCT while reducing about 90% of the scanning time, which can be highly valuable for moving target localization in image-guided radiotherapy.

Quantification of lung function on CT images based on pulmonary radiomic filtering.

PURPOSE: To develop a radiomics filtering technique for characterizing spatial-encoded regional pulmonary ventilation information on lung computed tomography (CT). METHODS: The lung volume was segmented on 46 CT images, and a 3D sliding window kernel was implemented across the lung volume to capture the spatial-encoded image information. Fifty-three radiomic features were extracted within the kernel, resulting in a fourth-order tensor object. As such, each voxel coordinate of the original lung was represented as a 53-dimensional feature vector, such that radiomic features could be viewed as feature maps within the lungs. To test the technique as a potential pulmonary ventilation biomarker, the radiomic feature maps were compared to paired functional images (Galligas PET or DTPA-SPECT) based on the Spearman correlation (ρ) analysis. RESULTS: The radiomic feature maps GLRLM-based Run-Length Non-Uniformity and GLCOM-based Sum Average are found to be highly correlated with the functional imaging. The achieved ρ (median [range]) for the two features are 0.46 [0.05, 0.67] and 0.45 [0.21, 0.65] across 46 patients and 2 functional imaging modalities, respectively. CONCLUSIONS: The results provide evidence that local regions of sparsely encoded heterogeneous lung parenchyma on CT are associated with diminished radiotracer uptake and measured lung ventilation defects on PET/SPECT imaging. These findings demonstrate the potential of radiomics to serve as a complementary tool to the current lung quantification techniques and provide hypothesis-generating data for future studies.

A geometry-guided multi-beamlet deep learning technique for CT reconstruction.

Purpose. Previous studies have proposed deep-learning techniques to reconstruct CT images from sinograms. However, these techniques employ large fully-connected (FC) layers for projection-to-image domain transformation, producing large models requiring substantial computation power, potentially exceeding the computation memory limit. Our previous work proposed a geometry-guided-deep-learning (GDL) technique for CBCT reconstruction that reduces model size and GPU memory consumption. This study further develops the technique and proposes a novel multi-beamlet deep learning (GMDL) technique of improved performance. The study compares the proposed technique with the FC layer-based deep learning (FCDL) method and the GDL technique through low-dose real-patient CT image reconstruction.Methods. Instead of using a large FC layer, the GMDL technique learns the projection-to-image domain transformation by constructing many small FC layers. In addition to connecting each pixel in the projection domain to beamlet points along the central beamlet in the image domain as GDL does, these smaller FC layers in GMDL connect each pixel to beamlets peripheral to the central beamlet based on the CT projection geometry. We compare ground truth images with low-dose images reconstructed with the GMDL, the FCDL, the GDL, and the conventional FBP methods. The images are quantitatively analyzed in terms of peak-signal-to-noise-ratio (PSNR), structural-similarity-index-measure (SSIM), and root-mean-square-error (RMSE).Results. Compared to other methods, the GMDL reconstructed low-dose CT images show improved image quality in terms of PSNR, SSIM, and RMSE. The optimal number of peripheral beamlets for the GMDL technique is two beamlets on each side of the central beamlet. The model size and memory consumption of the GMDL model is less than 1/100 of the FCDL model.Conclusion. Compared to the FCDL method, the GMDL technique is demonstrated to be able to reconstruct real patient low-dose CT images of improved image quality with significantly reduced model size and GPU memory requirement.

Enhancement of 4-D Cone-Beam Computed Tomography (4D-CBCT) Using a Dual-Encoder Convolutional Neural Network (DeCNN).

4D-CBCT is a powerful tool to provide respiration-resolved images for the moving target localization. However, projections in each respiratory phase are intrinsically under-sampled under the clinical scanning time and imaging dose constraints. Images reconstructed by compressed sensing (CS)-based methods suffer from blurred edges. Introducing the average-4D-image constraint to the CS-based reconstruction, such as prior-image-constrained CS (PICCS), can improve the edge sharpness of the stable structures. However, PICCS can lead to motion artifacts in the moving regions. In this study, we proposed a dual-encoder convolutional neural network (DeCNN) to realize the average-image-constrained 4D-CBCT reconstruction. The proposed DeCNN has two parallel encoders to extract features from both the under-sampled target phase images and the average images. The features are then concatenated and fed into the decoder for the high-quality target phase image reconstruction. The reconstructed 4D-CBCT using of the proposed DeCNN from the real lung cancer patient data showed (1) qualitatively, clear and accurate edges for both stable and moving structures; (2) quantitatively, low-intensity errors, high peak signal-to-noise ratio, and high structural similarity compared to the ground truth images; and (3) superior quality to those reconstructed by several other state-of-the-art methods including the back-projection, CS total-variation, PICCS, and the single-encoder CNN. Overall, the proposed DeCNN is effective in exploiting the average-image constraint to improve the 4D-CBCT image quality.