Artificial intelligence-based radiotherapy machine parameter optimization using reinforcement learning.

PURPOSE: To develop and evaluate a volumetric modulated arc therapy (VMAT) machine parameter optimization (MPO) approach based on deep-Q reinforcement learning (RL) capable of finding an optimal machine control policy using previous prostate cancer patient CT scans and contours, and applying the policy to new cases to rapidly produce deliverable VMAT plans in a simplified beam model. METHODS: A convolutional deep-Q network was employed to control the dose rate and multileaf collimator of a C-arm linear accelerator model using the current dose distribution and machine parameter state as input. A Q-value was defined as the discounted cumulative cost based on dose objectives, and experience-replay RL was performed to determine a policy to minimize the Q-value. A two-dimensional network design was employed which optimized each opposing leaf pair independently while monitoring the corresponding dose plane blocked by those leaves. This RL approach was applied to CT and contours from 40 retrospective prostate cancer patients. The dataset was split into training (15 patients) and validation (5 patients) groups to optimize the network, and its performance was tested in an independent cohort of 20 patients by comparing RL-based dose distributions to conformal arcs and clinical intensity modulated radiotherapy (IMRT) delivering a prescription dose of 78 Gy in 40 fractions. RESULTS: Mean ± SD execution time of the RL VMAT optimization was 1.5 ± 0.2 s per slice. In the test cohort, mean ± SD (P-value) planning target volume (PTV), bladder, and rectum dose were 80.5 ± 2.0 Gy (P < 0.001), 44.2 ± 14.6 Gy (P < 0.001), and 43.7 ± 11.1 Gy (P < 0.001) for RL VMAT compared to 81.6 ± 1.1 Gy, 51.6 ± 12.9 Gy, and 36.0 ± 12.3 Gy for clinical IMRT. CONCLUSIONS: RL was applied to VMAT MPO using clinical patient contours without independently optimized treatment plans for training and achieved comparable target and normal tissue dose to clinical plans despite the application of a relatively simple network design originally developed for video-game control. These results suggest that extending a RL approach to a full three-dimensional beam model could enable rapid artificial intelligence-based optimization of deliverable treatment plans, reducing the time required for radiotherapy planning without requiring previous plans for training.

A semiautomatic segmentation method for interstitial needles in intraoperative 3D transvaginal ultrasound images for high-dose-rate gynecologic brachytherapy of vaginal tumors.

PURPOSE: The purpose of this study was to evaluate the use of a semiautomatic algorithm to simultaneously segment multiple high-dose-rate (HDR) gynecologic interstitial brachytherapy (ISBT) needles in three-dimensional (3D) transvaginal ultrasound (TVUS) images, with the aim of providing a clinically useful tool for intraoperative implant assessment. METHODS AND MATERIALS: A needle segmentation algorithm previously developed for HDR prostate brachytherapy was adapted and extended to 3D TVUS images from gynecologic ISBT patients with vaginal tumors. Two patients were used for refining/validating the modified algorithm and five patients (8-12 needles/patient) were reserved as an unseen test data set. The images were filtered to enhance needle edges, using intensity peaks to generate feature points, and leveraged the randomized 3D Hough transform to identify candidate needle trajectories. Algorithmic segmentations were compared against manual segmentations and calculated dwell positions were evaluated. RESULTS: All 50 test data set needles were successfully segmented with 96% of algorithmically segmented needles having angular differences <3° compared with manually segmented needles and the maximum Euclidean distance was <2.1 mm. The median distance between corresponding dwell positions was 0.77 mm with 86% of needles having maximum differences <3 mm. The mean segmentation time using the algorithm was <30 s/patient. CONCLUSIONS: We successfully segmented multiple needles simultaneously in intraoperative 3D TVUS images from gynecologic HDR-ISBT patients with vaginal tumors and demonstrated the robustness of the algorithmic approach to image artifacts. This method provided accurate segmentations within a clinically efficient timeframe, providing the potential to be translated into intraoperative clinical use for implant assessment.

Accuracy and variability of high-dose-rate prostate brachytherapy needle tip localization using live two-dimensional and sagittally reconstructed three-dimensional ultrasound.

PURPOSE: To measure the accuracy and variability of manual high-dose-rate (HDR) prostate brachytherapy (BT) needle tip localization using sagittally reconstructed three-dimensional (3D) transrectal ultrasound (TRUS) augmented with live two-dimensional (2D) sagittal TRUS. METHODS AND MATERIALS: Ten prostate cancer patients underwent HDR-BT during which the sagittally assisted sagittally reconstructed (SASR) segmentation technique was completed in parallel with commercially available sagittally assisted axially reconstructed (SAAR) TRUS for comparison. The SASR technique makes use of live 2D ultrasound intraoperatively and allows needle tip updates using the final 3D image in the absence of image artifacts. These updates were repeated offline twice by two separate users. Needle end-length measurements were used to calculate insertion depth errors (IDEs) for each technique. RESULTS: Images of 147 needles were analyzed. For the SASR technique, both users were confident in tip positions on the final 3D image within 3 mm for 52% of needles, so these tip positions were updated. For the remaining 48% of needles, the tip positions from the live 2D images were used. This SASR technique enabled the localization of all needles with IDEs within ±3 mm for 84% of needles and IDE range of [-6.2 mm, 5.9 mm], compared with 57% and [-8.1 mm, 7.7 mm] when using the commercially available SAAR technique. CONCLUSIONS: The SASR technique mitigates the impact of 3D TRUS image artifacts on HDR-BT needle tip localization by incorporating live 2D sagittal TRUS intraoperatively and provides a statistically significant reduction in IDE variance compared with the routine SAAR technique.

ANG1 treatment reduces muscle pathology and prevents a decline in perfusion in DMD mice.

Vascular endothelial growth factor (VEGF) and other pro-angiogenic growth factors have been investigated to enhance muscle tissue perfusion and repair in Duchenne muscular dystrophy (DMD). Current understanding is limited by a lack of functional data following in vivo delivery of these growth factors. We previously used dynamic contrast-enhanced computed tomography to monitor disease progression in murine models of DMD, but no study to date has utilized this imaging technique to assess vascular therapy in a preclinical model of DMD. In the current study, we locally delivered VEGF and ANG1 alone or in combination to dystrophic hind limb skeletal muscle. Using functional imaging, we found the combination treatment as well as ANG1 alone prevented decline in muscle perfusion whereas VEGF alone had no effect compared to controls. These findings were validated histologically as demonstrated by increased alpha-smooth muscle actin-positive vessels in muscles that received either VEGF+ANG1 or ANG1 alone compared to the sham group. We further show that ANG1 alone slows progression of fibrosis compared to either sham or VEGF treatment. The findings from this study shed new light on the functional effects of vascular therapy and suggest that ANG1 alone may be a candidate therapy in the treatment of DMD.

Simultaneous automatic segmentation of multiple needles using 3D ultrasound for high-dose-rate prostate brachytherapy.

PURPOSE: Sagittally reconstructed 3D (SR3D) ultrasound imaging shows promise for improved needle localization for high-dose-rate prostate brachytherapy (HDR-BT); however, needles must be manually segmented intraoperatively while the patient is anesthetized to create a treatment plan. The purpose of this article was to describe and validate an automatic needle segmentation algorithm designed for HDR-BT, specifically capable of simultaneously segmenting all needles in an HDR-BT implant using a single SR3D image with ~5 mm interneedle spacing. MATERIALS AND METHODS: The segmentation algorithm involves regularized feature point classification and line trajectory identification based on the randomized 3D Hough transform modified to handle multiple straight needles in a single image simultaneously. Needle tips are identified based on peaks in the derivative of the signal intensity profile along the needle trajectory. For algorithm validation, 12 prostate cancer patients underwent HDR-BT during which SR3D images were acquired with all needles in place. Needles present in each of the 12 images were segmented manually, providing a gold standard for comparison, and using the algorithm. Tip errors were assessed in terms of the 3D Euclidean distance between needle tips, and trajectory error was assessed in terms of 2D distance in the axial plane and angular deviation between trajectories. RESULTS: In total, 190 needles were investigated. Mean execution time of the algorithm was 11.0 s per patient, or 0.7 s per needle. The algorithm identified 82% and 85% of needle tips with 3D errors ≤3 mm and ≤5 mm, respectively, 91% of needle trajectories with 2D errors in the axial plane ≤3 mm, and 83% of needle trajectories with angular errors ≤3°. The largest tip error component was in the needle insertion direction. CONCLUSIONS: Previous work has indicated HDR-BT needles may be manually segmented using SR3D images with insertion depth errors ≤3 mm and ≤5 mm for 83% and 92% of needles, respectively. The algorithm shows promise for reducing the time required for the segmentation of straight HDR-BT needles, and future work involves improving needle tip localization performance through improved image quality and modeling curvilinear trajectories.

Three-dimensional transrectal ultrasound guided high-dose-rate prostate brachytherapy: A comparison of needle segmentation accuracy with two-dimensional image guidance.

PURPOSE: Conventional transrectal ultrasound guided high-dose-rate prostate brachytherapy (HDR-BT) uses an axially acquired image set for organ segmentation and 2D sagittal images for needle segmentation. Sagittally reconstructed 3D (SR3D) transrectal ultrasound enables both organ and needle segmentation and has the potential to reduce organ-needle alignment uncertainty. This study compares the accuracy of needle tip localization between the conventional 2D sagittally assisted axially reconstructed (SAAR) and SR3D approaches. METHODS AND MATERIALS: Twelve patients underwent SAAR-guided HDR-BT, during which SR3D images were acquired for subsequent segmentation and analysis. A total of 183 needles were investigated. Needle end-length measurements were taken, providing a gold standard for insertion depths. Dosimetric impact of insertion depth errors (IDEs) on clinical treatment plans was assessed. RESULTS: SR3D guidance provided statistically significantly smaller IDEs than SAAR guidance with a mean ± SD of -0.6 ± 3.2 mm and 2.8 ± 3.2 mm, respectively (p < 0.001). Shadow artifacts were found to obstruct the view of some needle tips in SR3D images either partially (12%) or fully (10%); however, SR3D IDEs had a statistically significantly smaller impact on prostate V100% than SAAR IDEs with mean ± SD decreases of -1.2 ± 1.3% and -6.5 ± 6.7%, respectively (p < 0.05). CONCLUSIONS: SR3D-guided HDR-BT eliminates a source of systematic uncertainty from the SAAR-guided approach, providing decreased IDEs for most needles, leading to a significant decrease in dosimetric uncertainty. Although imaging artifacts can limit the accuracy of tip localization in a subset of needles, we identified a method to mitigate these artifacts for clinical implementation.