Comparison of deep learning, radiomics and subjective assessment of chest CT findings in SARS-CoV-2 pneumonia.

PURPOSE: Comparison of deep learning algorithm, radiomics and subjective assessment of chest CT for predicting outcome (death or recovery) and intensive care unit (ICU) admission in patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. METHODS: The multicenter, ethical committee-approved, retrospective study included non-contrast-enhanced chest CT of 221 SARS-CoV-2 positive patients from Italy (n = 196 patients; mean age 64 ± 16 years) and Denmark (n = 25; mean age 69 ± 13 years). A thoracic radiologist graded presence, type and extent of pulmonary opacities and severity of motion artifacts in each lung lobe on all chest CTs. Thin-section CT images were processed with CT Pneumonia Analysis Prototype (Siemens Healthineers) which yielded segmentation masks from a deep learning (DL) algorithm to derive features of lung abnormalities such as opacity scores, mean HU, as well as volume and percentage of all-attenuation and high-attenuation (opacities >-200 HU) opacities. Separately, whole lung radiomics were obtained for all CT exams. Analysis of variance and multiple logistic regression were performed for data analysis. RESULTS: Moderate to severe respiratory motion artifacts affected nearly one-quarter of chest CTs in patients. Subjective severity assessment, DL-based features and radiomics predicted patient outcome (AUC 0.76 vs AUC 0.88 vs AUC 0.83) and need for ICU admission (AUC 0.77 vs AUC 0.0.80 vs 0.82). Excluding chest CT with motion artifacts, the performance of DL-based and radiomics features improve for predicting ICU admission. CONCLUSION: DL-based and radiomics features of pulmonary opacities from chest CT were superior to subjective assessment for differentiating patients with favorable and adverse outcomes.

Multicenter Assessment of CT Pneumonia Analysis Prototype for Predicting Disease Severity and Patient Outcome.

To perform a multicenter assessment of the CT Pneumonia Analysis prototype for predicting disease severity and patient outcome in COVID-19 pneumonia both without and with integration of clinical information. Our IRB-approved observational study included consecutive 241 adult patients (> 18 years; 105 females; 136 males) with RT-PCR-positive COVID-19 pneumonia who underwent non-contrast chest CT at one of the two tertiary care hospitals (site A: Massachusetts General Hospital, USA; site B: Firoozgar Hospital Iran). We recorded patient age, gender, comorbid conditions, laboratory values, intensive care unit (ICU) admission, mechanical ventilation, and final outcome (recovery or death). Two thoracic radiologists reviewed all chest CTs to record type, extent of pulmonary opacities based on the percentage of lobe involved, and severity of respiratory motion artifacts. Thin-section CT images were processed with the prototype (Siemens Healthineers) to obtain quantitative features including lung volumes, volume and percentage of all-type and high-attenuation opacities (≥ -200 HU), and mean HU and standard deviation of opacities within a given lung region. These values are estimated for the total combined lung volume, and separately for each lung and each lung lobe. Multivariable analyses of variance (MANOVA) and multiple logistic regression were performed for data analyses. About 26% of chest CTs (62/241) had moderate to severe motion artifacts. There were no significant differences in the AUCs of quantitative features for predicting disease severity with and without motion artifacts (AUC 0.94-0.97) as well as for predicting patient outcome (AUC 0.7-0.77) (p > 0.5). Combination of the volume of all-attenuation opacities and the percentage of high-attenuation opacities (AUC 0.76-0.82, 95% confidence interval (CI) 0.73-0.82) had higher AUC for predicting ICU admission than the subjective severity scores (AUC 0.69-0.77, 95% CI 0.69-0.81). Despite a high frequency of motion artifacts, quantitative features of pulmonary opacities from chest CT can help differentiate patients with favorable and adverse outcomes.

A multi-platform approach to image guided radiation therapy (IGRT).

Siemens Medical Solutions, Oncology Care Systems Group (SMSOCSG) is supporting the development of several technologies that enable image acquisition and decision making processes required for IGRT in various clinical settings. Four such technologies are presented including: (i) the integration of a traditional multi-slice computed tomography (CT) scanner "on rails" with a C-arm gantry linear accelerator; (ii) the development of a high sensitivity, fast, megavoltage (MV) electronic portal imaging device capable of clinical MV Conebeam CT (MVCBCT) reconstruction and fluoroscopy mounted on a C-arm gantry linear accelerator; (iii) the modification of a mobile C-arm with flat panel kilovoltage (kV) diagnostic imager; and (iv) the development of an in-line megavoltage and kilovoltage flat panel imaging system that has the potential to image both anatomical and dosimetric information in "real-time" utilizing the traditional C-arm gantry linear accelerator geometry. Each method of IGRT has unique as well as complementary qualities which are discussed from both a clinical and technical perspective.

Respiratory correlated cone-beam computed tomography on an isocentric C-arm.

A methodology for 3D image reconstruction from retrospectively gated cone-beam CT projection data has been developed. A mobile x-ray cone-beam device consisting of an isocentric C-arm equipped with a flat panel detector was used to image a moving phantom. Frames for reconstruction were retrospectively selected from complete datasets based on the known rotation of the C-arm and a signal from a respiratory monitor. Different sizes of gating windows were tested. A numerical criterion for blur on the reconstructed image was suggested. The criterion is based on minimization of an Ising energy function, similar to approaches used in image segmentation or restoration. It is shown that this criterion can be used for the determination of the optimal gating window size. Images reconstructed from the retrospectively gated projection sequences using the optimal gating window data showed a significant improvement compared to images reconstructed from the complete projection datasets.

Low-dose megavoltage cone-beam CT for radiation therapy.

PURPOSE: The objective of this work was to demonstrate the feasibility of acquiring low-exposure megavoltage cone-beam CT (MV CBCT) three-dimensional (3D) image data of sufficient quality to register the CBCT images to kilovoltage planning CT images for patient alignment and dose verification purposes. METHODS AND MATERIALS: A standard clinical 6-MV Primus linear accelerator, operating in arc therapy mode, and an amorphous-silicon (a-Si) flat-panel electronic portal-imaging device (EPID) were employed. The dose-pulse rate of 6-MV Primus accelerator beam was windowed to expose an a-Si flat panel by using only 0.02 to 0.08 monitor unit (MUs) per image. A triggered image-acquisition mode was designed to produce a high signal-to-noise ratio without pulsing artifacts. Several data sets were acquired for an anthropomorphic head phantom and frozen sheep and pig cadaver head, as well as for a head-and-neck cancer patient on intensity-modulated radiotherapy (IMRT). For each CBCT image, a set of 90 to 180 projection images incremented by 1 degree to 2 degrees was acquired. The two-dimensional (2D) projection images were then synthesized into a 3D image by use of cone-beam CT reconstruction. The resulting MV CBCT image set was used to visualize the 3D bony anatomy and some soft-tissue details. The 3D image registration with the kV planning CT was performed either automatically by application of a maximization of mutual information (MMI) algorithm or manually by aligning multiple 1D slices. RESULTS: Low-noise 3D MV CBCT images without pulsing artifacts were acquired with a total delivered dose that ranged from 5 to 15 cGy. Acquisition times, including image readout, were on the order of 90 seconds for 180 projection images taken through a continuous gantry rotation of 180 degrees. The processing time of the data required an additional 90 seconds for the reconstruction of a 256(3) cube with 1.0-mm voxel size. Implanted gold markers (1 mm x 3 mm) were easily visible or all exposure levels without artifacts. In general, the presence of high Z materials such as tooth fillings or implanted markers did not result in visible streak artifacts. The registration of structures such as the spinal canal and the nasopharynx in the MV CBCT and kV CT data sets was possible with millimeter and degree accuracy as assessed by displacement simulations and subsequent visual evaluation. CONCLUSIONS: We believe that the quality of these images, along with the rapid acquisition and reconstruction times, demonstrates that MV CBCT performed by use of a standard linear accelerator equipped with a flat-panel imager can be applied clinically for patient alignment.