A gradient mapping guided explainable deep neural network for extracapsular extension identification in 3D head and neck cancer computed tomography images.

BACKGROUND: Diagnosis and treatment management for head and neck squamous cell carcinoma (HNSCC) is guided by routine diagnostic head and neck computed tomography (CT) scans to identify tumor and lymph node features. The extracapsular extension (ECE) is a strong predictor of patients' survival outcomes with HNSCC. It is essential to detect the occurrence of ECE as it changes staging and treatment planning for patients. Current clinical ECE detection relies on visual identification and pathologic confirmation conducted by clinicians. However, manual annotation of the lymph node region is a required data preprocessing step in most of the current machine learning-based ECE diagnosis studies. PURPOSE: In this paper, we propose a Gradient Mapping Guided Explainable Network (GMGENet) framework to perform ECE identification automatically without requiring annotated lymph node region information. METHODS: The gradient-weighted class activation mapping (Grad-CAM) technique is applied to guide the deep learning algorithm to focus on the regions that are highly related to ECE. The proposed framework includes an extractor and a classifier. In a joint training process, informative volumes of interest (VOIs) are extracted by the extractor without labeled lymph node region information, and the classifier learns the pattern to classify the extracted VOIs into ECE positive and negative. RESULTS: In evaluation, the proposed methods are well-trained and tested using cross-validation. GMGENet achieved test accuracy and area under the curve (AUC) of 92.2% and 89.3%, respectively. GMGENetV2 achieved 90.3% accuracy and 91.7% AUC in the test. The results were compared with different existing models and further confirmed and explained by generating ECE probability heatmaps via a Grad-CAM technique. The presence or absence of ECE has been analyzed and correlated with ground truth histopathological findings. CONCLUSIONS: The proposed deep network can learn meaningful patterns to identify ECE without providing lymph node contours. The introduced ECE heatmaps will contribute to the clinical implementations of the proposed model and reveal unknown features to radiologists. The outcome of this study is expected to promote the implementation of explainable artificial intelligence-assiste ECE detection.

A brain MRI dataset and baseline evaluations for tumor recurrence prediction after Gamma Knife radiotherapy.

Prediction and identification of tumor recurrence are critical for brain cancer treatment design and planning. Stereotactic radiation therapy delivered with Gamma Knife has been developed as one of the common treatment approaches combined with others by delivering radiation that targets accurately on the tumor while not affecting nearby healthy tissues. In this paper, we release a fully publicly available brain cancer MRI dataset and the companion Gamma Knife treatment planning and follow-up data for the purpose of tumor recurrence prediction. The dataset contains original patient MRI images, radiation therapy data, and clinical information. Lesion annotations are provided, and inclusive preprocessing steps have been specified to simplify the usage of this dataset. A baseline framework based on a convolutional neural network is proposed companionably with basic evaluations. The release of this dataset will contribute to the future development of automated brain tumor recurrence prediction algorithms and promote the clinical implementations associated with the computer vision field. The dataset is made publicly available on The Cancer Imaging Archive (TCIA) ( https://doi.org/10.7937/xb6d-py67 ).

Chasing Zero Harm in Radiation Oncology: Using Pre-treatment Peer Review.

Purpose: The Joint Commission has encouraged the healthcare industry to become "High Reliability Organizations" by "Chasing Zero Harm" in patient care. In radiation oncology, the time point of quality checks determines whether errors are prevented or only mitigated. Thus, to "chase zero" in radiation oncology, peer review has to be implemented prior to treatment initiation. A multidisciplinary group consensus peer review (GCPR) model is used pre-treatment at our institution and has been successful in our efforts to "chase zero harm" in patient care. Methods: With the GCPR model, policy-defined complex cases go through a treatment planning conference, which includes physicians, residents, physicists, and dosimetrists. Three major plan aspects are reviewed: target volumes, target and normal tissue dose coverage, and dose distributions. During the review, any team member can ask questions and afterwards a group consensus is taken regarding plan approval. Results: The GCPR model has been implemented through a commitment to peer review and creative conference scheduling. Automated analysis software is used to depict color-coded results for department approved target coverage and dose constraints. About 8% of plans required re-planning while about 23% required minor changes. The mean time for review of each plan was 8 min. Conclusions: Catching errors prior to treatment is the only way to "chase zero" in radiation oncology. Various types of errors may exist in treatment plans and our GCPR model succeeds in preventing many errors of all shapes and sizes in target definition, dose prescriptions, and treatment plans from ever reaching the patients.

Modeling treatment couches in the Pinnacle treatment planning system: Especially important for arc therapy.

This study is to demonstrate the importance and a method of properly modeling the treatment couch for dose calculation in patient treatment using arc therapy. The 2 treatment couch tops-Aktina AK550 and Elekta iBEAM evo-of Elekta LINACs were scanned using Philips Brilliance Big Bore CT Simulator. Various parts of the couch tops were contoured, and their densities were measured and recorded on the Pinnacle treatment planning system (TPS) using the established computed tomography density table. These contours were saved as organ models to be placed beneath the patient during planning. Relative attenuation measurements were performed following procedures outlined by TG-176 as well as absolute dose comparison of static fields of 10 × 10 cm(2) that were delivered through the couch tops with that calculated in the TPS with the couch models. A total of 10 random arc therapy treatment plans (5 volumetric-modulated arc therapy [VMAT] and 5 stereotactic body radiation therapy [SBRT]), using 24 beams, were selected for this study. All selected plans were calculated with and without couch modeling. Each beam was evaluated using the Delta(4) dosimetry system (Delta(4)). The Student t-test was used to determine statistical significance. Independent reviews were exploited as per the Imaging and Radiation Oncology Core head and neck credentialing phantom. The selected plans were calculated on the actual patient anatomies with and without couch modeling to determine potential clinical effects. Large relative beam attenuations were noted dependent on which part of the couch top beams were passing through. Substantial improvements were also noted for static fields both calculated with the TPS and delivered physically when the couch models were included in the calculation. A statistically significant increase in agreement was noted for dose difference, distance to agreement, and γ-analysis with the Delta(4) on VMAT and SBRT plans. A credentialing review showed improvement in treatment delivery after couch modeling with both thermoluminescent dosimeter doses and film analysis. Furthermore, analysis of treatment plans with and without using the couch model showed a statistically significant reduction in planning target volume coverage and increase in skin dose. In conclusion, ignoring the treatment couch, a common practice when generating a patient treatment plan, can overestimate the dose delivered especially for arc therapy. This work shows that explicitly modeling the couch during planning can meaningfully improve the agreement between calculated and measured dose distributions. Because of this project, we have implemented the couch models clinically across all treatment plans.