DAART: a deep learning platform for deeply accelerated adaptive radiation therapy for lung cancer.

Purpose: The study aimed to implement a novel, deeply accelerated adaptive radiation therapy (DAART) approach for lung cancer radiotherapy (RT). Lung cancer is the most common cause of cancer-related death, and RT is the preferred medically inoperable treatment for early stage non-small cell lung cancer (NSCLC). In the current lengthy workflow, it takes a median of four weeks from diagnosis to RT treatment, which can result in complete restaging and loss of local control with delay. We implemented the DAART approach, featuring a novel deepPERFECT system, to address unwanted delays between diagnosis and treatment initiation. Materials and methods: We developed a deepPERFECT to adapt the initial diagnostic imaging to the treatment setup to allow initial RT planning and verification. We used data from 15 patients with NSCLC treated with RT to train the model and test its performance. We conducted a virtual clinical trial to evaluate the treatment quality of the proposed DAART for lung cancer radiotherapy. Results: We found that deepPERFECT predicts planning CT with a mean high-intensity fidelity of 83 and 14 HU for the body and lungs, respectively. The shape of the body and lungs on the synthesized CT was highly conformal, with a dice similarity coefficient (DSC) of 0.91, 0.97, and Hausdorff distance (HD) of 7.9 mm, and 4.9 mm, respectively, compared with the planning CT scan. The tumor showed less conformality, which warrants acquisition of treatment Day1 CT and online adaptive RT. An initial plan was designed on synthesized CT and then adapted to treatment Day1 CT using the adapt to position (ATP) and adapt to shape (ATS) method. Non-inferior plan quality was achieved by the ATP scenario, while all ATS-adapted plans showed good plan quality. Conclusion: DAART reduces the common online ART (ART) treatment course by at least two weeks, resulting in a 50% shorter time to treatment to lower the chance of restaging and loss of local control.

Quantifying the impact of SpaceOAR hydrogel on inter-fractional rectal and bladder dose during 0.35 T MR-guided prostate adaptive radiotherapy.

PURPOSE: To investigate the impact of rectal spacing on inter-fractional rectal and bladder dose and the need for adaptive planning in prostate cancer patients undergoing SBRT with a 0.35 T MRI-Linac. MATERIALS AND METHODS: We evaluated and compared SBRT plans from prostate cancer patients with and without rectal spacer who underwent treatment on a 0.35 T MRI-Linac. Each group consisted of 10 randomly selected patients that received prostate SBRT to a total dose of 36.25 Gy in five fractions. Dosimetric differences in planned and delivered rectal and bladder dose and the number of fractions violating OAR constraints were quantified. We also assessed whether adaptive planning was needed to meet constraints for each fraction. RESULTS: On average, rectal spacing reduced the maximum dose delivered to the rectum by more than 8 Gy (p < 0.001). We also found that D3cc received by the rectum could be 12 Gy higher in patients who did not have rectal spacer (p < 9E-7). In addition, the results show that a rectal spacer can reduce the maximum dose and D15cc to the bladder wall by more than 1 (p < 0.004) and 8 (p < 0.009) Gy, respectively. Our study also shows that using a rectal spacer could reduce the necessity for adaptive planning. The incidence of dose constraint violation was observed in almost 91% of the fractions in patients without the rectal spacer and 52% in patients with implanted spacer. CONCLUSION: Inter-fractional changes in rectal and bladder dose were quantified in patients who underwent SBRT with/without rectal SpaceOAR hydrogel. Rectal spacer does not eliminate the need for adaptive planning but reduces its necessity.

Deep learning-based synthetic CT generation for MR-only radiotherapy of prostate cancer patients with 0.35T MRI linear accelerator.

PURPOSE: To develop a deep learning model to generate synthetic CT for MR-only radiotherapy of prostate cancer patients treated with 0.35 T MRI linear accelerator. MATERIALS AND METHODS: A U-NET convolutional neural network was developed to translate 0.35 T TRUFI MRI into electron density map using a novel cost function equalizing the contribution of various tissue types including fat, muscle, bone, and background air in training. The impact of training time, dataset size, image standardization, and data augmentation approaches was also quantified. Mean absolute error (MAE) between synthetic and planning CTs was calculated to measure the goodness of the model. RESULTS: With 20 patients in training, our U-NET model has the potential to generate synthetic CT with a MAE of about 29.68 ± 4.41, 16.34 ± 2.67, 23.36 ± 2.85, and 105.90 ± 22.80 HU over the entire body, fat, muscle, and bone tissues, respectively. As expected, we found that the number of patients used for training and MAE are nonlinearly correlated. Data augmentation and our proposed loss function were effective to improve MAE by ~9% and ~18% in bony voxels, respectively. Increasing the training time and image standardization did not improve the accuracy of the model. CONCLUSION: A U-NET model has been developed and tested numerically to generate synthetic CT from 0.35T TRUFI MRI for MR-only radiotherapy of prostate cancer patients. Dosimetric evaluation using a large and independent dataset warrants the validity of the proposed model and the actual number of patients needed for the safe usage of the model in routine clinical workflow.

Dosimetric evaluation of an atlas-based synthetic CT generation approach for MR-only radiotherapy of pelvis anatomy.

PURPOSE: To investigate the potential of an atlas-based approach in generation of synthetic CT for pelvis anatomy. METHODS: Twenty-three matched pairs of computed tomography (CT) and magnetic resonance imaging (MRI) scans were selected from a pool of prostate cancer patients. All MR scans were preprocessed to reduce scanner- and patient-induced intensity inhomogeneities and to standardize their intensity histograms. Ten (training dataset) of 23 pairs were then utilized to construct the coregistered CT-MR atlas. The synthetic CT for a new patient is generated by appropriately weighting the deformed atlas of CT-MR onto the new patient MRI. The training dataset was used as an atlas to generate the synthetic CT for the rest of the patients (test dataset). The mean absolute error (MAE) between the deformed planning CT and synthetic CT was computed over the entire CT image, bone, fat, and muscle tissues. The original treatment plans were also recomputed on the new synthetic CTs and dose-volume histogram metrics were compared. The results were compared with a commercially available synthetic CT Software (MRCAT) that is routinely used in our clinic. RESULTS: MAE errors (±SD) between the deformed planning CT and our proposed synthetic CTs in the test dataset were 47 ± 5, 116 ± 12, 36 ± 6, and 47 ± 5 HU for the entire image, bone, fat, and muscle tissues respectively. The MAEs were 65 ± 5, 172 ± 9, 43 ± 7, and 42 ± 4 HU for the corresponding tissues in MRCAT CT. The dosimetric comparison showed consistent results for all plans using our synthetic CT, deformed planning CT and MRCAT CT. CONCLUSION: We investigated the potential of a multiatlas approach to generate synthetic CT images for the pelvis. Our results demonstrate excellent results in terms of HU value assignment compared to the original CT and dosimetric consistency.

Multiatlas approach with local registration goodness weighting for MRI-based electron density mapping of head and neck anatomy.

PURPOSE: The growing use of magnetic resonance imaging (MRI) as a substitute for computed tomography-based treatment planning requires the development of effective algorithms to generate electron density maps for treatment planning and patient setup verification. The purpose of this work was to develop a method to synthesize computerized tomography (CT) for MR-only radiotherapy of head and neck cancer patients. METHODS: The algorithm is based on registration of multiple patient datasets containing both MRI and CT images (a "multiatlas" algorithm). Twelve matched pairs of good quality CT and MRI scans (those without apparent motion and blurring artifacts) were selected from a pool of head and neck cancer patients to form the atlas. All atlas MRI scans were preprocessed to reduce scanner- and patient-induced intensity inhomogeneities and to standardize their intensity histograms. Atlas CT and MRIs were coregistered using a novel bone-to-air replacement technique applied to the CT scans that improves the similarity between CTs and MRIs and facilitates the registration process. For each new patient, all atlas MRIs are deformed initially onto the new patients' MRI. We introduce a generalized registration error (GRE) metric that automatically measures the goodness of local registration between MRI pairs. The final synthetic CT value at each point is a nonlinear GRE-weighted average of the atlas CTs. For evaluation, the leave-one-out technique was used for synthetic CT generation and the mean absolute error (MAE) between the original and synthetic CT was computed over the entire CT image. The impact of our proposed CT-MR registration scheme on the accuracy of the final synthetic CT was also studied. The original treatment plans were also recomputed on the new synthetic CTs and dose-volume histogram metrics were compared. In addition, the two-dimensional (2D) gamma analysis at 1%/1 mm and 2%/2 mm dose difference/distance to agreement was also performed to study the dose distribution at the isocenter. RESULTS: MAE error (± standard deviation) between the original and the synthetic CTs was 64 ± 10, 113 ± 12, and 130 ± 28 Hounsfield Unit (HU) for the entire image, air, and bone regions respectively. Our results showed that our proposed bone-suppression based CT-MR fusion and GRE-weighted strategy could lower the overall MAE error between the original and synthetic CTs by ~69% and ~34% respectively. Dose recalculation comparison showed highly consistent results between plans based on the synthetic vs. the original CTs. The 2D gamma analysis revealed the pass rate of 95.44 ± 2.5 and 99.36 ± 0.71 for 1%/1 mm and 2%/2 mm criteria respectively. Due to local registration weighting, the method is robust with respect to MRI imaging artifacts. CONCLUSION: We developed a novel image analysis technique to synthesize CT for head and neck anatomy. Novel methods were introduced to accurately register atlas CTs and MRIs as well as to weight the final electron density maps using local registration goodness estimates. The resulting accuracy is clinically acceptable, at least for these atlas patients.

A Radiation-Induced Hippocampal Vascular Injury Surrogate Marker Predicts Late Neurocognitive Dysfunction.

PURPOSE: We aimed to develop a hippocampal vascular injury surrogate marker for early prediction of late neurocognitive dysfunction in patients receiving brain radiation therapy (RT). METHODS AND MATERIALS: Twenty-seven patients (17 males and 10 females, 31-80 years of age) were enrolled in an institutional review board-approved prospective longitudinal study. Patients received diagnoses of low-grade glioma or benign tumor and were treated by (3D) conformal or intensity-modulated RT with a median dose of 54 Gy (50.4-59.4 Gy in 1.8-Gy fractions). Six dynamic-contrast enhanced MRI scans were performed from pre-RT to 18-month post-RT, and quantified for vascular parameters related to blood-brain barrier permeability, K(trans), and the fraction of blood plasma volume, Vp. The temporal changes in the means of hippocampal transfer constant K(trans) and Vp after starting RT were modeled by integrating the dose effects with age, sex, hippocampal laterality, and presence of tumor or edema near a hippocampus. Finally, the early vascular dose response in hippocampi was correlated with neurocognitive dysfunction at 6 and 18 months post-RT. RESULTS: The mean K(trans) Increased significantly from pre-RT to 1-month post-RT (P<.0004), which significantly depended on sex (P<.0007) and age (P<.00004), with the dose response more pronounced in older females. Also, the vascular dose response in the left hippocampus of females correlated significantly with changes in memory function at 6 (r=-0.95, P<.0006) and 18-months (r=-0.88, P<.02) post-RT. CONCLUSIONS: The early hippocampal vascular dose response could be a predictor of late neurocognitive dysfunction. A personalized hippocampus sparing strategy may be considered in the future.

Arterial perfusion imaging-defined subvolume of intrahepatic cancer.

PURPOSE: To assess whether an increase in a subvolume of intrahepatic tumor with elevated arterial perfusion during radiation therapy (RT) predicts tumor progression after RT. METHODS AND MATERIALS: Twenty patients with unresectable intrahepatic cancers undergoing RT were enrolled in a prospective, institutional review board-approved study. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) was performed before RT (pre-RT), after delivering ∼60% of the planned dose (mid-RT) and 1 month after completion of RT to quantify hepatic arterial perfusion. The arterial perfusions of the tumors at pre-RT were clustered into low-normal and elevated perfusion by a fuzzy clustering-based method, and the tumor subvolumes with elevated arterial perfusion were extracted from the hepatic arterial perfusion images. The percentage changes in the tumor subvolumes and means of arterial perfusion over the tumors from pre-RT to mid-RT were evaluated for predicting tumor progression post-RT. RESULTS: Of the 24 tumors, 6 tumors in 5 patients progressed 5 to 21 months after RT completion. Neither tumor volumes nor means of tumor arterial perfusion at pre-RT were predictive of treatment outcome. The mean arterial perfusion over the tumors increased significantly at mid-RT in progressive tumors compared with the responsive tumors (P=.006). From pre-RT to mid-RT, the responsive tumors had a decrease in the tumor subvolumes with elevated arterial perfusion (median, -14%; range, -75% to 65%), whereas the progressive tumors had an increase of the subvolumes (median, 57%; range, -7% to 165%) (P=.003). Receiver operating characteristic analysis of the percentage change in the subvolume for predicting tumor progression post-RT had an area under the curve of 0.90. CONCLUSION: The increase in the subvolume of the intrahepatic tumor with elevated arterial perfusion during RT has the potential to be a predictor for tumor progression post-RT. The tumor subvolume could be a radiation boost candidate for response-driven adaptive RT.

DCE-MRI defined subvolumes of a brain metastatic lesion by principle component analysis and fuzzy-c-means clustering for response assessment of radiation therapy.

PURPOSE: To develop a pharmacokinetic modelfree framework to analyze the dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) data for assessment of response of brain metastases to radiation therapy. METHODS: Twenty patients with 45 analyzable brain metastases had MRI scans prior to whole brain radiation therapy (WBRT) and at the end of the 2-week therapy. The volumetric DCE images covering the whole brain were acquired on a 3T scanner with approximately 5 s temporal resolution and a total scan time of about 3 min. DCE curves from all voxels of the 45 brain metastases were normalized and then temporally aligned. A DCE matrix that is constructed from the aligned DCE curves of all voxels of the 45 lesions obtained prior to WBRT is processed by principal component analysis to generate the principal components (PCs). Then, the projection coefficient maps prior to and at the end of WBRT are created for each lesion. Next, a pattern recognition technique, based upon fuzzy-c-means clustering, is used to delineate the tumor subvolumes relating to the value of the significant projection coefficients. The relationship between changes in different tumor subvolumes and treatment response was evaluated to differentiate responsive from stable and progressive tumors. Performance of the PC-defined tumor subvolume was also evaluated by receiver operating characteristic (ROC) analysis in prediction of nonresponsive lesions and compared with physiological-defined tumor subvolumes. RESULTS: The projection coefficient maps of the first three PCs contain almost all response-related information in DCE curves of brain metastases. The first projection coefficient, related to the area under DCE curves, is the major component to determine response while the third one has a complimentary role. In ROC analysis, the area under curve of 0.88 ± 0.05 and 0.86 ± 0.06 were achieved for the PC-defined and physiological-defined tumor subvolume in response assessment. CONCLUSIONS: The PC-defined subvolume of a brain metastasis could predict tumor response to therapy similar to the physiological-defined one, while the former is determined more rapidly for clinical decision-making support.

Investigation of the diffusion abnormality index as a new imaging biomarker for early assessment of brain tumor response to radiation therapy.

BACKGROUND: Diffusion MRI, although having the potential to be a biomarker for early assessment of tumor response to therapy, could be confounded by edema and necrosis in or near the brain tumors. This study aimed to develop and investigate the ability of the diffusion abnormality index (DAI) to be a new imaging biomarker for early assessment of brain metastasis response to radiation therapy (RT). METHODS: Patients with either radiosensitive or radioresistant brain metastases that were treated by whole brain RT alone or combined with bortezomib as a radiation sensitizer had diffusion-weighted (DW) MRI pre-RT and 2 weeks (2W) after starting RT. A patient-specific diffusion abnormality probability function (DAProF) was created to account for abnormal low and high apparent diffusion coefficients differently, reflecting respective high cellularity and edema/necrosis. The DAI of a lesion was then calculated by the integral of DAProF-weighted tumor apparent diffusion coefficient histogram. The changes in DAI from pre-RT to 2W were evaluated for differentiating the responsive, stable, and progressive tumors and compared with the changes in gross tumor volume and conventional diffusion metrics during the same time interval. RESULTS: In lesions treated with whole brain RT, the DAI performed the best among all metrics in predicting the posttreatment response of brain metastases to RT. In lesions treated with whole brain RT + bortezomib, although DAI was the best predictor, the performance of all metrics worsened compared with the first group. CONCLUSIONS: The ability of DAI for early assessment of brain metastasis response to RT depends upon treatment regimes.

Physiological imaging-defined, response-driven subvolumes of a tumor.

PURPOSE: To develop an image analysis framework to delineate the physiological imaging-defined subvolumes of a tumor in relating to treatment response and outcome. METHODS AND MATERIALS: Our proposed approach delineates the subvolumes of a tumor based on its heterogeneous distributions of physiological imaging parameters. The method assigns each voxel a probabilistic membership function belonging to the physiological parameter classes defined in a sample of tumors, and then calculates the related subvolumes in each tumor. We applied our approach to regional cerebral blood volume (rCBV) and Gd-DTPA transfer constant (K(trans)) images of patients who had brain metastases and were treated by whole-brain radiation therapy (WBRT). A total of 45 lesions were included in the analysis. Changes in the rCBV (or K(trans))-defined subvolumes of the tumors from pre-RT to 2 weeks after the start of WBRT (2W) were evaluated for differentiation of responsive, stable, and progressive tumors using the Mann-Whitney U test. Performance of the newly developed metrics for predicting tumor response to WBRT was evaluated by receiver operating characteristic (ROC) curve analysis. RESULTS: The percentage decrease in the high-CBV-defined subvolumes of the tumors from pre-RT to 2W was significantly greater in the group of responsive tumors than in the group of stable and progressive tumors (P<.007). The change in the high-CBV-defined subvolumes of the tumors from pre-RT to 2W was a predictor for post-RT response significantly better than change in the gross tumor volume observed during the same time interval (P=.012), suggesting that the physiological change occurs before the volumetric change. Also, K(trans) did not add significant discriminatory information for assessing response with respect to rCBV. CONCLUSION: The physiological imaging-defined subvolumes of the tumors delineated by our method could be candidates for boost target, for which further development and evaluation is warranted.

An approach for computer-aided detection of brain metastases in post-Gd T1-W MRI.

PURPOSE: To develop an approach for computer-aided detection (CAD) of small brain metastases in post-Gd T1-weighted magnetic resonance imaging (MRI). METHOD: A set of unevenly spaced 3D spherical shell templates was optimized to localize brain metastatic lesions by cross-correlation analysis with MRI. Theoretical and simulation analyses of effects of lesion size and shape heterogeneity were performed to optimize the number and size of the templates and the cross-correlation thresholds. Also, effects of image factors of noise and intensity variation on the performance of the CAD system were investigated. A nodule enhancement strategy to improve sensitivity of the system and a set of criteria based upon the size, shape and brightness of lesions were used to reduce false positives. An optimal set of parameters from the FROC curves was selected from a training dataset, and then the system was evaluated on a testing dataset including 186 lesions from 2753 MRI slices. Reading results from two radiologists are also included. RESULTS: Overall, a 93.5% sensitivity with 0.024 of intra-cranial false positive rate (IC-FPR) was achieved in the testing dataset. Our investigation indicated that nodule enhancement was very effective in improving both sensitivity and specificity. The size and shape criteria reduced the IC-FPR from 0.075 to 0.021, and the brightness criterion decreases the extra-cranial FPR from 0.477 to 0.083 in the training dataset. Readings from the two radiologists had sensitivities of 60% and 67% in the training dataset and 70% and 80% in the testing dataset for the metastatic lesions <5 mm in diameter. CONCLUSION: Our proposed CAD system has high sensitivity and fairly low FPR for detection of the small brain metastatic lesions in MRI compared to the previous work and readings of neuroradiologists. The potential of this method for assisting clinical decision- making warrants further evaluation and improvements.

An image analysis approach for automatic malignancy determination of prostate pathological images.

BACKGROUND: Determining malignancy of prostate pathological samples is important for treatment planning of prostate cancer. Traditionally, this is performed by expert pathologists who evaluate the structure of prostate glands in the biopsy samples. However, this is a subjective task due to inter- and intra-observer differences among pathologists. Also, it is time-consuming and difficult to some extent. Therefore, automatic determination of malignancy of prostate pathological samples is of interest. METHODS: A texture-based technique is first used to segment the prostate glands in the image. Features related to size and shape of these glands are then extracted and combined to generate an index, which is proportional to malignancy of cancer. A linear classifier is employed to classify the specimens into benign (low potential for malignancy) and malignant. RESULTS: The leave-one-out technique is employed to evaluate the method using two datasets. The first has 91 images with similar magnifications and illuminations while the second has 199 images with different magnifications and illuminations. In the experiments, accuracies of about 98 and 95% have been achieved for these two datasets, respectively. CONCLUSIONS: An image analysis approach is employed to evaluate prostate pathological images. Experimental results show that the proposed method can successfully classify the prostate biopsy samples into benign and malignant. They also show that the proposed method is robust to variations in magnification and illumination.