Deep Learning-Based Localization and Orientation Estimation of Pedicle Screws in Spinal Fusion Surgery.

Objective: This study investigated the application of a deep learning-based object detection model for accurate localization and orientation estimation of spinal fixation surgical instruments during surgery. Methods: We employed the You Only Look Once (YOLO) object detection framework with oriented bounding boxes (OBBs) to address the challenge of non-axis-aligned instruments in surgical scenes. The initial dataset of 100 images was created using brochure and website images from 11 manufacturers of commercially available pedicle screws used in spinal fusion surgeries, and data augmentation was used to expand 300 images. The model was trained, validated, and tested using 70%, 20%, and 10% of the images of lumbar pedicle screws, with the training process running for 100 epochs. Results: The model testing results showed that it could detect the locations of the pedicle screws in the surgical scene as well as their direction angles through the OBBs. The F1 score of the model was 0.86 (precision: 1.00, recall: 0.80) at each confidence level and mAP50. The high precision suggests that the model effectively identifies true positive instrument detections, although the recall indicates a slight limitation in capturing all instruments present. This approach offers advantages over traditional object detection in bounding boxes for tasks where object orientation is crucial, and our findings suggest the potential of YOLOv8 OBB models in real-world surgical applications such as instrument tracking and surgical navigation. Conclusion: Future work will explore incorporating additional data and the potential of hyperparameter optimization to improve overall model performance.

Feasibility of fast, four-dimensional computed tomography-based O-ring LINAC plans for lung stereotactic body radiotherapy in patients with poor performance status.

Purpose: We aimed to retrospectively analyzed the feasibility of fast four-dimensional computed tomography (4DCT)-based O-ring LINAC treatment for patients with an average respiratory amplitude was< 0.5 cm and who cannot endure long treatment times due to poor performance status in lung 4D-stereotactic body radiotherapy (SBRT). Methods: This study included data of 38 patients who received lung 4D-SBRT and had average respiratory amplitude< 0.5 cm in the full phase. C-arm LINAC plans were based on 4DCT data obtained at phase values ranging from 20-70% using a C-arm LINAC. O-ring LINAC plans were retrospectively established based on 4DCT data obtained at phase values of 0-90% using an O-ring LINAC. The conformity index (CI), homogeneity index (HI), and gradient measurement of the planning target volumes (PTV) were analyzed to compare dosimetric data between C-arm LINAC and O-ring LINAC plans. Organs at risk were analyzed in accordance with the Radiation Therapy Oncology Group 0915 protocol. Treatment delivery time and total monitor units were analyzed to compare the efficiency of treatment delivery. Statistical comparisons were performed using the Wilcoxon signed-rank test (P< 0.05). Results: For the PTV, there was no significant difference in the CI or HI between C-arm LINAC and O-ring LINAC plans. For organs-at-risk, all plans met the criteria for dose constraint. There was a significant difference between C-arm LINAC and O-ring LINAC plans except in the spinal cord. Treatment delivery time was 92% longer for C-arm LINAC plans than for O-ring LINAC plans. The total MU value for C-arm LINAC plans was 9.6% higher than that for O-ring LINAC plans. Conclusion: We verified the feasibility of fast 4DCT-based O-ring LINAC treatment for patients with average respiratory amplitude< 0.5 cm and who cannot endure long treatment times due to poor performance status in lung 4D-SBRT.

The Usefulness of Trabecular CT Attenuation Measurement at L4 Level to Predict Screw Loosening After Degenerative Lumbar Fusion Surgery: Consider Number of Fused Levels and Postoperative Sagittal Balance.

STUDY DESIGN: Retrospective study. OBJECTIVE: To evaluate the absolute value of L4 trabecular region-of-interest (t-ROI) computed tomography (CT) attenuation, which can predict pedicle screw loosening, and determine the changes in value according to number of fused levels and sagittal balance in patients undergoing lumbar fusion surgery. SUMMARY OF BACKGROUND DATA: Although osteoporosis was not diagnosed in spinal dual x-ray absorptiometry preoperatively, we encountered several cases of screw loosening within 1 year of lumbar fusion surgery. METHODS: We enrolled 478 patients and analyzed factors related to screw loosening. We evaluated the association between L4 t-ROI CT attenuation and screw loosening and determined the best cutoff value of t L4 t-ROI CT attenuation for predicting screw loosening. RESULTS: The number of fused levels, postoperative C7-S1 sagittal vertical axis (SVA), and L4 t-ROI CT attenuation were independently correlated with screw loosening. According to number of fused level and postoperative C7-T1 SVA (≥36.9 mm or <36.9 mm), in patients with one-level fusion and C7-S1 SVA less than 36.9 mm, the optimal cutoff point of the L4 t-ROI CT attenuation predicting screw loosening was 106.5 Hounsfield unit (HU). L4 t-ROI attenuation did not change until two-level fusions. In patients with three-level fusions and C7-S1 SVA less than 36.9 mm, the optimal cutoff point of the L4 t-ROI CT attenuation predicting screw loosening was 159.0 HU. The optimal cutoff point of L4 t-ROI CT attenuation in patients with three-level fusions and C7-S1 SVA more than or equal to 36.9 mm was 191.0 HU. CONCLUSION: L4 t-ROI CT attenuation value considering number of fused levels and sagittal balance is an accurate measurement method to predict screw loosening. Spine surgeons should be aware of the L4 t-ROI attenuation before surgery to improve the fusion rate and reduce instrument-related complications of lumbar spine surgery in osteoporotic patients. LEVEL OF EVIDENCE: 3.

Accuracy evaluation of surface registration algorithm using normal distribution transform in stereotactic body radiotherapy/radiosurgery: A phantom study.

PURPOSE: To evaluate a feasibility of normal distribution transform (NDT) algorithm compared with the iterative closest point (ICP) method as a useful surface registration in stereotactic body radiotherapy (SBRT)/stereotactic radiosurgery (SRS). METHODS: Point cloud images using the 3D triangulation technology were obtained from a depth camera-based optical imaging (OSI) system equipped in a radiosurgery room. Two surface registration algorithms, NDT and ICP, were used to measure and compare the discrepancy values between the reference and the current surfaces during the positioning of the patient. The performance evaluation was investigated by calculating the registration error and root-mean-square (RMS) values for the surface model, reposition, and target accuracy, which were analyzed statistically using a paired t-test. RESULTS: For surface model accuracy, the average of the registration error and RMS values were measured as 3.56 ± 2.20 mm and 6.98 ± 1.89 mm for ICP method, and 1.76 ± 1.32 mm and 3.58 ± 1.30 mm for NDT method (p < 0.05). For reposition accuracy, the average registration error and RMS values were calculated as 1.41 ± 0.98 mm and 2.53 ± 1.64 mm using ICP method, and 0.92 ± 0.61 mm and 1.75 ± 0.80 mm using NDT method (p = 0.005). The overall target accuracy using the NDT method reduced the average of the reposition error and overall RMS value by 0.71 and 1.32 mm, respectively, compared to the ICP method (p = 0.03). CONCLUSIONS: We found that the surface registration algorithm based on NDT method provides more reliable accuracy in the values of surface model, reposition, and target accuracies than the classic ICP method. The NDT method in OSI systems offers reasonable accuracy in SBRT/SRS.

Dosimetric comparison of VitalBeam® and HalcyonTM 2.0 for hypofractionated VMAT with simultaneous integrated boost treatment of early-stage left-sided breast cancer.

PURPOSE: This study compared the quality of treatment plans for early-stage, left-sided breast cancer, as planned for and delivered by the HalcyonTM and VitalBeam® . MATERIALS AND METHODS: Fifteen patients diagnosed with early-stage left-sided breast cancer, who had received VMAT with hypofractionated SIB, were recruited. All cases were planned using HalcyonTM comprising a dual-layer MLC (DL-MLC) and VitalBeam® with a Millennium 120 MLC (VB-MLC). For the PTVs, the quality of coverage (QC), conformity index (CI), and homogeneity index (HI) were calculated for each plan. The dosimetric differences between the two treatment plans were statistically compared using the Wilcoxon signed-rank test (p < 0.05). To evaluate delivery efficiency, the average delivery time for each patient's treatment plan was recorded and compared. RESULTS: For the PTVs, the two plans (DL-MLC and VB-MLC) were comparable in terms of the QC, CI, and HI. However, V30Gy and Dmean for the heart in the DL-MLC plan were significantly reduced by 0.49% and 14.6%, respectively, compared with those in the VB-MLC plan (p < 0.05). The Dmean value for the ipsilateral lung in the DL-MLC plan significantly decreased by 5.5%, compared with that in the VB-MLC plan (p < 0.05). In addition, the delivery times for the DL-MLC and VB-MLC plans were 79 ± 10 and 101 ± 11 s, respectively. CONCLUSIONS: DL-MLC plans were found to improve OAR sparing. In particular, when treating left-sided breast cancer via DL-MLC plans, the risk of heart toxicity is expected to be reduced.

Evaluation of the dosimetric effect of scattered protons in clinical practice in passive scattering proton therapy.

The present study verified and evaluated the dosimetric effects of protons scattered from a snout and an aperture in clinical practice, when a range compensator was included. The dose distribution calculated by a treatment planning system (TPS) was compared with the measured dose distribution and the dose distribution calculated by Monte Carlo simulation at several depths. The difference between the measured and calculated results was analyzed using Monte Carlo simulation with filtration of scattering in the snout and aperture. The dependence of the effects of scattered protons on snout size, beam range, and minimum thickness of the range compensator was also investigated using the Monte Carlo simulation. The simulated and measured results showed that the additional dose compared with the results calculated by the TPS at shallow depths was mainly due to protons scattered by the snout and aperture. This additional dose was filtered by the structure of the range compensator so that it was observed under the thin region of the range compensator. The maximum difference was measured at a depth of 16 mm (8.25%), with the difference decreasing with depth. Analysis of protons contributing to the additional dose showed that the contribution of protons scattered from the snout was greater than that of protons scattered from the aperture when a narrow snout was used. In the Monte Carlo simulation, this effect of scattered protons was reduced when wider snouts and longer-range proton beams were used. This effect was also reduced when thicker range compensator bases were used, even with a narrow snout. This study verified the effect of scattered protons even when a range compensator was included and emphasized the importance of snout-scattered protons when a narrow snout is used for small fields. It indicated that this additional dose can be reduced by wider snouts, longer range proton beams, and thicker range compensator bases. These results provide a better understanding of the additional dose from scattered protons in clinical practice.

Facial expression monitoring system for predicting patient's sudden movement during radiotherapy using deep learning.

PURPOSE: Imaging, breath-holding/gating, and fixation devices have been developed to minimize setup errors so that the prescribed dose can be exactly delivered to the target volume in radiotherapy. Despite these efforts, additional patient monitoring devices have been installed in the treatment room to view patients' whole-body movement. We developed a facial expression recognition system using deep learning with a convolutional neural network (CNN) to predict patients' advanced movement, enhancing the stability of the radiation treatment by giving warning signs to radiation therapists. MATERIALS AND METHODS: Convolutional neural network model and extended Cohn-Kanade datasets with 447 facial expressions of source images for training were used. Additionally, a user interface that can be used in the treatment control room was developed to monitor real-time patient's facial expression in the treatment room, and the entire system was constructed by installing a camera in the treatment room. To predict the possibility of patients' sudden movement, we categorized facial expressions into two groups: (a) uncomfortable expressions and (b) comfortable expressions. We assumed that the warning sign about the sudden movement was given when the uncomfortable expression was recognized. RESULTS: We have constructed the facial expression monitoring system, and the training and test accuracy were 100% and 85.6%, respectively. In 10 patients, their emotions were recognized based on their comfortable and uncomfortable expressions with 100% detection rate. The detected various emotions were represented by a heatmap and motion prediction accuracy was analyzed for each patient. CONCLUSION: We developed a system that monitors the patient's facial expressions and predicts patient's advanced movement during the treatment. It was confirmed that our patient monitoring system can be complementarily used with the existing monitoring system. This system will help in maintaining the initial setup and improving the accuracy of radiotherapy for the patients using deep learning in radiotherapy.

Comparative clinical evaluation of atlas and deep-learning-based auto-segmentation of organ structures in liver cancer.

BACKGROUND: Accurate and standardized descriptions of organs at risk (OARs) are essential in radiation therapy for treatment planning and evaluation. Traditionally, physicians have contoured patient images manually, which, is time-consuming and subject to inter-observer variability. This study aims to a) investigate whether customized, deep-learning-based auto-segmentation could overcome the limitations of manual contouring and b) compare its performance against a typical, atlas-based auto-segmentation method organ structures in liver cancer. METHODS: On-contrast computer tomography image sets of 70 liver cancer patients were used, and four OARs (heart, liver, kidney, and stomach) were manually delineated by three experienced physicians as reference structures. Atlas and deep learning auto-segmentations were respectively performed with MIM Maestro 6.5 (MIM Software Inc., Cleveland, OH) and, with a deep convolution neural network (DCNN). The Hausdorff distance (HD) and, dice similarity coefficient (DSC), volume overlap error (VOE), and relative volume difference (RVD) were used to quantitatively evaluate the four different methods in the case of the reference set of the four OAR structures. RESULTS: The atlas-based method yielded the following average DSC and standard deviation values (SD) for the heart, liver, right kidney, left kidney, and stomach: 0.92 ± 0.04 (DSC ± SD), 0.93 ± 0.02, 0.86 ± 0.07, 0.85 ± 0.11, and 0.60 ± 0.13 respectively. The deep-learning-based method yielded corresponding values for the OARs of 0.94 ± 0.01, 0.93 ± 0.01, 0.88 ± 0.03, 0.86 ± 0.03, and 0.73 ± 0.09. The segmentation results show that the deep learning framework is superior to the atlas-based framwork except in the case of the liver. Specifically, in the case of the stomach, the DSC, VOE, and RVD showed a maximum difference of 21.67, 25.11, 28.80% respectively. CONCLUSIONS: In this study, we demonstrated that a deep learning framework could be used more effectively and efficiently compared to atlas-based auto-segmentation for most OARs in human liver cancer. Extended use of the deep-learning-based framework is anticipated for auto-segmentations of other body sites.

Development of a 3D optical scanner for evaluating patient-specific dose distributions.

PURPOSE: This paper describes the hardware and software characteristics of a 3D optical scanner (P3DS) developed in-house. The P3DS consists of an LED light source, diffuse screen, step motor, CCD camera, and scanner management software with 3D reconstructed software. MATERIALS AND METHOD: We performed optical simulation, 2D and 3D reconstruction image testing, and pre-clinical testing for the P3DS. We developed the optical scanner with three key characteristics in mind. First, we developed a continuous scanning method to expand possible clinical applications. Second, we manufactured a collimator to improve image quality by reducing scattering from the light source. Third, we developed an optical scanner with changeable camera positioning to enable acquisition of optimal images according to the size of the gel dosimeter. RESULTS: We confirmed ray-tracing in P3DS with optic simulation and found that 2D projection and 3D reconstructed images were qualitatively similar to the phantom images. For pre-clinical tests, the dose distribution and profile showed good agreement among RTP, optical CT, and external beam radiotherapy film data for the axial and coronal views. The P3DS has shown that it can scan and reconstruct for evaluation of the gel dosimeter within 1 min. We confirmed that the P3DS system is a useful tool for the measurement of 3D dose distributions for 3D radiation therapy QA. Further experiments are needed to investigate quantitative analysis for 3D dose distribution.

Patient performance-based plan parameter optimization for prostate cancer in tomotherapy.

The purpose of this study is to evaluate the influence of treatment-planning parameters on the quality of treatment plans in tomotherapy and to find the optimized planning parameter combinations when treating patients with prostate cancer under different performances. A total of 3 patients with prostate cancer with Eastern Cooperative Oncology Group (ECOG) performance status of 2 or 3 were included in this study. For each patient, 27 treatment plans were created using a combination of planning parameters (field width of 1, 2.5, and 5cm; pitch of 0.172, 0.287, and 0.43; and modulation factor of 1.8, 3, and 3.5). Then, plans were analyzed using several dosimetrical indices: the prescription isodose to target volume (PITV) ratio, homogeneity index (HI), conformity index (CI), target coverage index (TCI), modified dose HI (MHI), conformity number (CN), and quality factor (QF). Furthermore, dose-volume histogram of critical structures and critical organ scoring index (COSI) were used to analyze organs at risk (OAR) sparing. Interestingly, treatment plans with a field width of 1cm showed more favorable results than others in the planning target volume (PTV) and OAR indices. However, the treatment time of the 1-cm field width was 3 times longer than that of plans with a field width of 5cm. There was no substantial decrease in treatment time when the pitch was increased from 0.172 to 0.43, but the PTV indices were slightly compromised. As expected, field width had the most significant influence on all of the indices including PTV, OAR, and treatment time. For the patients with good performance who can tolerate a longer treatment time, we suggest a field width of 1cm, pitch of 0.172, and modulation factor of 1.8; for the patients with poor performance status, field width of 5cm, pitch of 0.287, and a modulation factor of 3.5 should be considered.

Induction of apoptosis in human acute leukemia Jurkat T cells by Albizzia julibrissin extract is mediated via mitochondria-dependent caspase-3 activation.

To understand antitumor activity of Albizzia julibrissin Durazz (Leguminosae), which has been used as a traditional oriental medicine, the mechanism underlying cytotoxic effect of its extract on human acute leukemia Jurkat T cells were investigated. The methanol extract of the stripped barks (3kg) of Albizzia julibrissin was evaporated, dissolved in water, and then sequentially extracted by chloroform, ethyl acetate, and n-butanol. The substance in the butanol extract containing the most cytotoxic activity was further purified by a series of preparative column chromatography. The active substance obtained (723mg) was designated as HaBC18. When Jurkat T cells were treated with HaBC18 (0.5-2microg/ml), apoptosis along with several biochemical events such as mitochondrial cytochrome c release, activation of caspase-9 and -3, degradation of PARP, and DNA fragmentation was induced in a dose-dependent manner. However, the HaBC18-induced apoptosis was abrogated by an ectopic overexpression of Bcl-xL, which is known to block mitochondrial cytochrome c release. Primary cultures of human PBMC were less sensitive to the cytotoxicity relative to Jurkat T cells. These results demonstrate that the cytotoxicity of HaBC18 toward Jurkat T cells is attributable to apoptosis mediated by mitochondria-dependent death-signaling pathway regulated by Bcl-xL.