Search

Heterogeneous graph neural networks (HGNNs) have been blossoming in recent years, but the unique data processing and evaluation setups used by each work obstruct a full understanding of their advancements. In this work, we present a systematical reproduction of 12 recent HGNNs by using their official codes, datasets, settings, and hyperparameters, revealing surprising findings about the progress of HGNNs. We find that the simple homogeneous GNNs, e.g., GCN and GAT, are largely underestimated due to improper settings. GAT with proper inputs can generally match or outperform all existing HGNNs across various scenarios. To facilitate robust and reproducible HGNN research, we construct the Heterogeneous Graph Benchmark (HGB), consisting of 11 diverse datasets with three tasks. HGB standardizes the process of heterogeneous graph data splits, feature processing, and performance evaluation. Finally, we introduce a simple but very strong baseline Simple-HGN--which significantly outperforms all previous models on HGB--to accelerate the advancement of HGNNs in the future., KDD 2021 research track

Mesh brachytherapy is a special type of a permanent brachytherapy implant: it uses low-energy radioactive seeds in an absorbable mesh that is sutured onto the tumor bed immediately after a surgical resection. This treatment offers low additional risk to the patient as the implant procedure is carried out as part of the tumor resection surgery. Mesh brachytherapy utilizes identification of the tumor bed through direct visual evaluation during surgery or medical imaging following surgery through radiographic imaging of radio-opaque markers within the sources located on the tumor bed. Thus, mesh brachytherapy is customizable for individual patients. Mesh brachytherapy is an intraoperative procedure involving mesh implantation and potentially real-time treatment planning while the patient is under general anesthesia. The procedure is multidisciplinary and requires the complex coordination of multiple medical specialties. The preimplant dosimetry calculation can be performed days beforehand or expediently in the operating room with the use of lookup tables. In this report, the guidelines of American Association of Physicists in Medicine (AAPM) are presented on the physics aspects of mesh brachytherapy. It describes the selection of radioactive sources, design and preparation of the mesh, preimplant treatment planning using a Task Group (TG) 43-based lookup table, and postimplant dosimetric evaluation using the TG-43 formalism or advanced algorithms. It introduces quality metrics for the mesh implant and presents an example of a risk analysis based on the AAPM TG-100 report. Recommendations include that the preimplant treatment plan be based upon the TG-43 dose calculation formalism with the point source approximation, and the postimplant dosimetric evaluation be performed by using either the TG-43 approach, or preferably the newer model-based algorithms (viz., TG-186 report) if available to account for effects of material heterogeneities. To comply with the written directive and regulations governing the medical use of radionuclides, this report recommends that the prescription and written directive be based upon the implanted source strength, not target-volume dose coverage. The dose delivered by mesh implants can vary and depends upon multiple factors, such as postsurgery recovery and distortions in the implant shape over time. For the sake of consistency necessary for outcome analysis, prescriptions based on the lookup table (with selection of the intended dose, depth, and treatment area) are recommended, but the use of more advanced techniques that can account for real situations, such as material heterogeneities, implant geometric perturbations, and changes in source orientations, is encouraged in the dosimetric evaluation. The clinical workflow, logistics, and precautions are also presented.

Independent verification of the dose per monitor unit (MU) to deliver the prescribed dose to a patient has been a mainstay of radiation oncology quality assurance. We discuss the role of secondary dose/MU calculation programs as part of a comprehensive Quality Assurance (QA) program. This report provides guidelines on calculation-based dose/MU verification for intensity modulated radiation therapy (IMRT) or volumetric modulated arc therapy (VMAT) provided by various modalities. We provide a review of various algorithms for "independent/second check" of monitor unit calculations for IMRT/VMAT. The report makes recommendations on the clinical implementation of secondary dose/MU calculation programs; on commissioning and acceptance of various commercially available secondary dose/MU calculation programs; on benchmark QA and periodic quality assurance; and on clinically reasonable action levels for agreement of secondary dose/MU calculation programs.

With the advancement of data-intensive technologies, such as image-guided radiation therapy (IGRT) and intensity-modulated radiation therapy (IMRT), the amount and complexity of data to be transferred between clinical subsystems have increased beyond the reach of manual checking. As a result, unintended treatment deviations (e.g., dose errors) may occur if the treatment system is not closely monitored by a comprehensive data transfer quality management program (QM). This report summarizes the findings and recommendations from the task group (TG) on quality assurance (QA) of external beam treatment data transfer (TG-201), with the aim to assist medical physicists in designing their own data transfer QM. As a background, a section of this report describes various models of data flow (distributed data repositories and single data base systems) and general data test characteristics (data integrity, interpretation, and consistency). Recommended tests are suggested based on the collective experience of TG-201 members. These tests are for the acceptance of, commissioning of, and upgrades to subsystems that store and/or modify clinical treatment data. As treatment complexity continues to evolve, we will need to do and know more about ensuring the quality of data transfers. The report concludes with the recommendation to move toward data transfer open standards compatibility and to develop tools that automate data transfer QA.

Massive open online courses are becoming a modish way for education, which provides a large-scale and open-access learning opportunity for students to grasp the knowledge. To attract students' interest, the recommendation system is applied by MOOCs providers to recommend courses to students. However, as a course usually consists of a number of video lectures, with each one covering some specific knowledge concepts, directly recommending courses overlook students'interest to some specific knowledge concepts. To fill this gap, in this paper, we study the problem of knowledge concept recommendation. We propose an end-to-end graph neural network-based approach calledAttentionalHeterogeneous Graph Convolutional Deep Knowledge Recommender(ACKRec) for knowledge concept recommendation in MOOCs. Like other recommendation problems, it suffers from sparsity issues. To address this issue, we leverage both content information and context information to learn the representation of entities via graph convolution network. In addition to students and knowledge concepts, we consider other types of entities (e.g., courses, videos, teachers) and construct a heterogeneous information network to capture the corresponding fruitful semantic relationships among different types of entities and incorporate them into the representation learning process. Specifically, we use meta-path on the HIN to guide the propagation of students' preferences. With the help of these meta-paths, the students' preference distribution with respect to a candidate knowledge concept can be captured. Furthermore, we propose an attention mechanism to adaptively fuse the context information from different meta-paths, in order to capture the different interests of different students. The promising experiment results show that the proposedACKRecis able to effectively recommend knowledge concepts to students pursuing online learning in MOOCs., 10 pages

Purpose: Gamma Knife radiosurgery is a highly precise and accurate treatment technique for treating brain diseases with low risk of serious error that nevertheless could potentially be reduced. We applied the AAPM Task Group 100 recommended failure modes and effects analysis (FMEA) tool to develop a risk-based quality management program for Gamma Knife radiosurgery. Methods: A team consisting of medical physicists, radiation oncologists, neurosurgeons, radiation safety officers, nurses, operating room technologists, and schedulers at our institution and an external physicist expert on Gamma Knife was formed for the FMEA study. A process tree and a failure mode table were created for the Gamma Knife radiosurgery procedures using the Leksell Gamma Knife Perfexion and 4C units. Three scores for the probability of occurrence (O), the severity (S), and the probability of no detection for failure mode (D) were assigned to each failure mode by 8 professionals on a scale from 1 to 10. An overall risk priority number (RPN) for each failure mode was then calculated from the averaged O, S, and D scores. The coefficient of variation for each O, S, or D score was also calculated. The failure modes identified were prioritized in terms of both the RPN scores and the severity scores. Results: The established process tree for Gamma Knife radiosurgery consists of 10 subprocesses and 53 steps, including a subprocess for frame placement and 11 steps that are directly related to the frame-based nature of the Gamma Knife radiosurgery. Out of the 86 failure modes identified, 40 Gamma Knife specific failure modes were caused by the potential for inappropriate use of the radiosurgery head frame, the imaging fiducial boxes, the Gamma Knife helmets and plugs, the skull definition tools as well as other features of the GammaPlan treatment planning system. The other 46 failure modes are associated with the registration, imaging, image transfer, contouring processes that are common for all external beam radiation therapy techniques. The failure modes with the highest hazard scores are related to imperfect frame adaptor attachment, bad fiducial box assembly, unsecured plugs/inserts, overlooked target areas, and undetected machine mechanical failure during the morning QA process. Conclusions: The implementation of the FMEA approach for Gamma Knife radiosurgery enabled deeper understanding of the overall process among all professionals involved in the care of the patient and helped identify potential weaknesses in the overall process. The results of the present study give us a basis for the development of a risk based quality management program for Gamma Knife radiosurgery.

PURPOSE The purpose of this study was to provide guidance on quality management for electronic brachytherapy. MATERIALS AND METHODS The task group used the risk-assessment approach of Task Group 100 of the American Association of Physicists in Medicine. Because the quality management program for a device is intimately tied to the procedure in which it is used, the task group first designed quality interventions for intracavitary brachytherapy for both commercial electronic brachytherapy units in the setting of accelerated partial-breast irradiation. To demonstrate the methodology to extend an existing risk analysis for a different application, the task group modified the analysis for the case of post-hysterectomy, vaginal cuff irradiation for one of the devices. RESULTS The analysis illustrated how the TG-100 methodology can lead to interventions to reduce risks and improve quality for each unit and procedure addressed. CONCLUSION This report provides a model to guide facilities establishing a quality management program for electronic brachytherapy.

Massive Open Online Courses(MOOCs) provide convenient access to knowledge for learners all over the world. Concept Extraction is a basic requirement in MOOCs. However, textual content in MOOCs, such as video subtitles and quizzes, are generally presented as semi-structured or unstructured format. Thus it is hard to extract important concepts with simple methods from MOOCs. In this paper, we design a graph-based propagation method to solve the concept extraction problem. Our method utilize textual and structured data on Wikipedia, to generate implicit and explicit representation for concepts respectively. Experiments show that our method outperforms alternative methods on Chinese dataset(+0.054-0.062 in terms of MAP).

BACKGROUND The energy index (EI) is a measure of dose homogeneity within a target volume calculated by the integral dose divided by the product of prescription dose and tumor volume. OBJECTIVE To assess whether a higher EI is associated with greater local control for brain metastases (BMs) treated by Gamma Knife radiosurgery (GKRS). METHODS We reviewed all patients treated with GKRS for BM at our institution between January 2009 and February 2014. Data on the prescription dose, prescription isodose line, minimum dose, mean dose, integral dose, tumor volume, and EI were collected. Tumor response was assessed by reviewing follow-up brain imaging studies and classified according to the Response Evaluation Criteria in Solid Tumors. Local control per lesion and dosimetric prognostic factors for local control were assessed by univariate and multivariate Cox proportional hazards regression analyses. RESULTS Of 213 patients treated, 126 had follow-up imaging available with a median follow-up of 6 months. Three hundred seventy-three individual tumors were analyzed. Of these, 133 showed a complete response, 157 showed a partial response, 46 remained stable, and 37 developed local failure. Tumors with EI ≥1.6 mJ·mL(-1)·Gy(-1) showed a higher rate of complete response. Local control rates at 6, 11, and 17 months were 95.4%, 86.5%, and 81.5%, respectively. On univariate analysis, the following factors were associated with higher rates of local failure: prescription doses of 16 and 18 Gy compared with a prescription dose of 20 Gy. The following factors were associated with a greater rate of local control: maximum dose and mean dose. On multivariate analysis, the only statistically significant factor associated with a greater rate of local failure was prescription dose of 16 Gy compared with 20 Gy. CONCLUSION GKRS for BM results in a high rate of local control with an 11-month rate of 86.5%. A higher EI was not significantly associated with a higher rate of local control on multivariate analysis. Prescription dose was found to be the only significant predictor of local control on multivariate analysis.

Identifying a reliable method of rectal volume delineation (RVD) for prostate cancer treatment planning using intensity-modulated radiation therapy (IMRT) is invaluable for clinical and investigational purposes. We analyzed rectal dose constraint outcomes utilizing commonly used RVD methods. Sixty-two prostate cancer patients treated with IMRT from August 2008 to March 2010 were reviewed. RVD either included the entire rectal volume (ERV) or limited the volume to a 3 mm rectal wall (3 MM). V50 ,the percentage of rectal volume receiving 50 Gy, V65, V70, and V75.6 were determined and analyzed for differences. We analyzed factors including definitive and postoperative treatment, rectal volume, and length. For definitive treatment, using ERV, the mean rectal V50, V65, V70, and V75.6 was 41.3, 21.6, 15.1, and 7.0 %, respectively. Using 3 MM, the mean V50, V65, V70, and V75.6 was 37.3, 23.4, 18.5, and 10.3 %, respectively. The V65, V70, and V75.6 were significantly lower in ERV than 3 MM (p

Yttrium-90 microsphere brachytherapy of the liver exploits the distinctive features of the liver anatomy to treat liver malignancies with beta radiation and is gaining more wide spread clinical use. This report provides a general overview of microsphere liver brachytherapy and assists the treatment team in creating local treatment practices to provide safe and efficient patient treatment. Suggestions for future improvements are incorporated with the basic rationale for the therapy and currently used procedures. Imaging modalities utilized and their respective quality assurance are discussed. General as well as vendor specific delivery procedures are reviewed. The current dosimetry models are reviewed and suggestions for dosimetry advancement are made. Beta activity standards are reviewed and vendor implementation strategies are discussed. Radioactive material licensing and radiation safety are discussed given the unique requirements of microsphere brachytherapy. A general, team-based quality assurance program is reviewed to provide guidance for the creation of the local procedures. Finally, recommendations are given on how to deliver the current state of the art treatments and directions for future improvements in the therapy.

Purpose: The technical challenges that have prevented routine proton magnetic resonance spectroscopic imaging (1H MRSI) examinations of the breast include insufficient spatial resolution, increased difficulties in shimming compared to the brain, and strong lipid contamination at short echo time (TE) at 1.5 T . The authors investigated the feasibility of high spatial resolution 1H MRSI of human breast cancer in a clinical setting at 3 T . Methods: Ten patient studies (eight cancers and two benign lesions) were performed in a 3 T whole-body clinical imager using a pulse sequence consisting of optional outer volume presaturation, optional CHESS pulse for lipid suppression, CHESS pulse for water suppression, and standard 2D/3D PRESS pulse sequence with an elliptical weighted k -space sampling scheme. Results: All ten studies were technically successful. The spectral quality was acceptable for all cases even the one with a 65 Hz width of water peak at half height. Choline (Cho) signals were clearly visible in malignant lesion areas, while there was no detectable Cho in normal appearing breast or in benign lesions. It was also observed that the distribution of Cho signal can be nonuniform across MRI demonstrated lesions. Conclusions: To the author’s knowledge, this is the first 2D/3D MRSI study of human breast cancer with short TE (less than 135 ms ) at 3 T and the highest spatial resolution (up to 0.25 cm 3 ) to date. In conclusion, the authors have presented a robust technique for high spatial resolutionin vivo 1H MRSI of human breast cancer that uses the combined advantages of high field, short TE, multivoxel, and high spatial resolution itself to overcome the major technical challenges and illustrated its potential for routine clinical examination as well as advantages over single-voxel techniques in studying metabolite heterogeneity.

The ever‐increasing data demands in a radiation oncology (RO) clinic require medical physicists to have a clearer understanding of information technology (IT) resource management issues. Clear lines of collaboration and communication among administrators, medical physicists, IT staff, equipment service engineers, and vendors need to be established. In order to develop a better understanding of the clinical needs and responsibilities of these various groups, an overview of the role of IT in RO is provided. This is followed by a list of IT‐related tasks and a resource map. The skill set and knowledge required to implement these tasks are described for the various RO professionals. Finally, various models for assessing one's IT resource needs are described. The exposition of ideas in this white paper is intended to be broad, in order to raise the level of awareness of the RO community; the details behind these concepts will not be given here and are best left to future task group reports. PACS number: 87.52.Tr, 87.53.St, 87.53.Xd, 87.90.+y

We examined the fetal dose from irradiation of glioblastoma during pregnancy using intensity modulated radiation therapy (IMRT), and describe fetal dose minimization using mobile shielding devices. A case report is described of a pregnant woman with glioblastoma who was treated during the third trimester of gestation with 60 Gy of radiation delivered via a 6 MV photon IMRT plan. Fetal dose without shielding was estimated using an anthropomorphic phantom with ion chamber and diode measurements. Clinical fetal dose with shielding was determined with optically stimulated luminescent dosimeters and ion chamber. Clinical target volume (CTV) and planning target volume (PTV) coverage was 100 and 98 % receiving 95 % of the prescription dose, respectively. Normal tissue tolerances were kept below quantitative analysis of normal tissue effects in the clinic (QUANTEC) recommendations. Without shielding, anthropomorphic phantom measurements showed a cumulative fetal dose of 0.024 Gy. In vivo measurements with shielding in place demonstrated a cumulative fetal dose of 0.016 Gy. The fetal dose estimated without shielding was 0.04 % and with shielding was 0.026 % of the target dose. In vivo estimation of dose equivalent received by the fetus was 24.21 mSv. Using modern techniques, brain irradiation can be delivered to pregnant patients in the third trimester with very low measured doses to the fetus, without compromising target coverage or normal tissue dose constraints. Fetal dose can further be reduced with the use of shielding devices, in keeping with the principle of as low as reasonably achievable.

The radiotherapy treatment planning process has evolved over the years with innovations in treatment planning, treatment delivery and imaging systems. Treatment modality and simulation technologies are also rapidly improving and affecting the planning process. For example, Image-guided-radiation-therapy has been widely adopted for patient setup, leading to margin reduction and isocenter repositioning after simulation. Stereotactic Body radiation therapy (SBRT) and Radiosurgery (SRS) have gradually become the standard of care for many treatment sites, which demand a higher throughput for the treatment plans even if the number of treatments per day remains the same. Finally, simulation, planning and treatment are traditionally sequential events. However, with emerging adaptive radiotherapy, they are becoming more tightly intertwined, leading to iterative processes. Enhanced efficiency of planning is therefore becoming more critical and poses serious challenge to the treatment planning process; Lean Six Sigma approaches are being utilized increasingly to balance the competing needs for speed and quality. In this symposium we will discuss the treatment planning process and illustrate effective techniques for managing workflow. Topics will include: 1. Planning techniques: (a) beam placement, (b) dose optimization, (c) plan evaluation (d) export to RVS. 2. Planning workflow: (a) import images, (b) Image fusion, (c) contouring, (d) plan approval (e) plan check (f) chart check, (g) sequential and iterative process 3. Influence of upstream and downstream operations: (a) simulation, (b) immobilization, (c) motion management, (d) QA, (e) IGRT, (f) Treatment delivery, (g) SBRT/SRS (h) adaptive planning 4. Reduction of delay between planning steps with Lean systems due to (a) communication, (b) limited resource, (b) contour, (c) plan approval, (d) treatment. 5. Optimizing planning processes: (a) contour validation (b) consistent planning protocol, (c) protocol/template sharing, (d) semi-automatic plan evaluation, (e) quality checklist for error prevention, (f) iterative process, (g) balance of speed and quality Learning Objectives: 1. Gain familiarity with the workflow of modern treatment planning process. 2. Understand the scope and challenges of managing modern treatment planning processes. 3. Gain familiarity with Lean Six Sigma approaches and their implementation in the treatment planning workflow.

Yttrium-90 microsphere brachytherapy of the liver exploits the distinctive features of the liver anatomy to treat liver malignancies with beta radiation and is gaining more wide spread clinical use. This report provides a general overview of microsphere liver brachytherapy and assists the treatment team in creating local treatment practices to provide safe and efficient patient treatment. Suggestions for future improvements are incorporated with the basic rationale for the therapy and currently used procedures. Imaging modalities utilized and their respective quality assurance are discussed. General as well as vendor specific delivery procedures are reviewed. The current dosimetry models are reviewed and suggestions for dosimetry advancement are made. Beta activity standards are reviewed and vendor implementation strategies are discussed. Radioactive material licensing and radiation safety are discussed given the unique requirements of microsphere brachytherapy. A general, team-based quality assurance program is reviewed to provide guidance for the creation of the local procedures. Finally, recommendations are given on how to deliver the current state of the art treatments and directions for future improvements in the therapy.

While ultrasound guided prostate brachytherapy has gained wide acceptance as a primary treatment tool for prostate cancer, quality assurance of the ultrasound guidance system has received very little attention. Task Group 128 of the American Association of Physicists in Medicine was created to address quality assurance requirements specific to transrectal ultrasound used for guidance of prostate brachytherapy. Accurate imaging guidance and dosimetry calculation depend upon the quality and accuracy of the ultrasound image. Therefore, a robust quality assurance program for the ultrasound system is essential. A brief review of prostate brachytherapy and ultrasound physics is provided, followed by a recommendation for elements to be included in a comprehensive test phantom. Specific test recommendations are presented, covering grayscale visibility, depth of penetration, axial and lateral resolution, distance measurement, area measurement, volume measurement, needle template/electronic grid alignment, and geometric consistency with the treatment planning computer.

Purpose: To quantify the spatial distortion of the MRIimages in the Elekta Leksell GammaPlan version 9 planning system using different registration methods. Methods: The RPC SRS phantom was imaged using a GE Signa HDxt 1.5Tesla MR scanner and a GE LightSpeed VCT CTscanner. Both the T1 weighted MRIimages and the fast imaging employing steady state acquisition (FIESTA) MRIimages were acquired, along with the axial and helical CTimages. The skull shell and the target in the phantom were contoured in all the four series of images acquired. The MRIimages were registered in the planning system in three different ways: 1) using the fiduciary marks in the imaging box; 2) using a global co‐registration with FOV just covering skull shell; 3) using a local co‐registration to max 8.5cm FOV at image center. The target positions in all the four series of images were compared using the software tool in the planning system. Results: The target positions as obtained from the helical and the axial CTimages agree within 0.1mm. For the registration method the fiduciary marks, the T1 weighted MRIimages are shifted from the CTimages 0.9mm in the anterior‐posterior direction, and 0.5mm in both the superior‐inferior and the left‐right directions. The corresponding displacements for the FIESTA images are 0.6mm, 0.5mm and 0.5mm respectively. The shifts of the target positions are significantly reduced in the global co‐registration approach, and all less than 0.2mm in the local co‐registration approach. Conclusions: MRIimage distortion in GammaPlan version 9 is in the sub‐millimeter range for the central region of the MRIimage. The spatial distortion in the FIESTA images is smaller than that in the T1 weighted images. Co‐registration to the central region of images can reduce MRIimaging distortion effect significantly.

In this presentation, we will focus on two types of scintillators with trapping of excited electrons for dosimetric purposes. The conventional type is thermoluminescent dosimeter(TLDs), in which the electrons are trapped in excited status when exposed to ionizing radiation, and when later heated, the trapped electrons then fall to their ground state with light emission. By measuring the light emitted one can determine the radiationdose previously received. In recent years, optically stimulated luminescent (OSL)dosimeter has grown to be an alternative to TLDs due to the convenience of its usage. The OSLdosimeters also trap excited electrons when exposed to ionizing radiation. Instead of heat, stimulation light is used to release the stored energy in OSLdosimeters. The intensity of the emitted light depends on the dose absorbed by the OSLdosimeters and the intensity of the stimulation light. Its sensitivity, reproducibility, precision in responding to accumulative dose and its energy dependence have also been studied in order to demonstrate its potential. We will discuss both TLDs and OSLDs with regards to their dosimetric properties and applications in radiation therapydosimetry as well as in diagnostic imaging dose monitoring. Learning Objectives: 1. To understand the basic principles of optically stimulated luminescent (OSL) and thermoluminescence(TLD)dosimeters. 2. To learn their physical properties and characteristics as dosimeters. 3. To compare the applications of OSL and TLDdosimeters in radiation therapydosimetery as well as in diagnostic imaging dose monitoring.

Purpose:IGRT has become an essential modality to accurately setup patients for radiation treatment. 4D/CT/PET systems generating multiple phases of respirationimage‐sets have greatly helped clinicians contour integrated target volumes. However, IGRT creates 30–40‐times more image data. PACS become important in modern Radiotherapy for massive storage needs. In this study, we defined a new system by expanding PACS into RT extension to cover additional non‐DICOM, RT‐specified data and intra‐departmental communications. Method and Materials: PACS‐RT was designed with two essential core architectures: data storage (archiving) and quality tracking (communications). The RT data stored includes DICOM/RT and non‐DICOM‐objects: TPS, RV it safeguards patient information and tracks quality in each step we perform. In a high‐volume cancer center with advanced treatments, this type of PACS‐RT system is proving essential. Conflict of Interest: N/A.