Landmark-based auto-contouring of clinical target volumes for radiotherapy of nasopharyngeal cancer.

BACKGROUND: The delineation of clinical target volumes (CTVs) for radiotherapy for nasopharyngeal cancer is complex and varies based on the location and extent of disease. PURPOSE: The current study aimed to develop an auto-contouring solution following one protocol guidelines (NRG-HN001) that can be adjusted to meet other guidelines, such as RTOG-0225 and the 2018 International guidelines. METHODS: The study used 2-channel 3-dimensional U-Net and nnU-Net framework to auto-contour 27 normal structures in the head and neck (H&N) region that are used to define CTVs in the protocol. To define the CTV-Expansion (CTV1 and CTV2) and CTV-Overall (the outer envelope of all the CTV contours), we used adjustable morphological geometric landmarks and mimicked physician interpretation of the protocol rules by partially or fully including select anatomic structures. The results were evaluated quantitatively using the dice similarity coefficient (DSC) and mean surface distance (MSD) and qualitatively by independent reviews by two H&N radiation oncologists. RESULTS: The auto-contouring tool showed high accuracy for nasopharyngeal CTVs. Comparison between auto-contours and clinical contours for 19 patients with cancers of various stages showed a DSC of 0.94 ± 0.02 and MSD of 0.4 ± 0.4 mm for CTV-Expansion and a DSC of 0.83 ± 0.02 and MSD of 2.4 ± 0.5 mm for CTV-Overall. Upon independent review, two H&N physicians found the auto-contours to be usable without edits in 85% and 75% of cases. In 15% of cases, minor edits were required by both physicians. Thus, one physician rated 100% of the auto-contours as usable (use as is, or after minor edits), while the other physician rated 90% as usable. The second physician required major edits in 10% of cases. CONCLUSIONS: The study demonstrates the ability of an auto-contouring tool to reliably delineate nasopharyngeal CTVs based on protocol guidelines. The tool was found to be clinically acceptable by two H&N radiation oncology physicians in at least 90% of the cases.

Outcomes of Complex Abdominal Wall Reconstruction After Oncologic Resection: 14-Year Experience at an NCI-Designated Cancer Center.

BACKGROUND: Outcomes studies for abdominal wall reconstruction (AWR) in the setting of previous oncologic extirpation are lacking. We sought to evaluate long-term outcomes of AWR using acellular dermal matrix (ADM) after extirpative resection, compare them to primary herniorrhaphy, and report the rates and predictors of postoperative complications. METHODS: We conducted a retrospective cohort study of patients who underwent AWR after oncologic resection from March 2005 to June 2019 at a tertiary cancer center. The primary outcome was hernia recurrence (HR). Secondary outcomes included surgical site occurrences (SSOs), surgical site infection (SSIs), length of hospital stay (LOS), reoperation, and 30-day readmission. RESULTS: Of 720 consecutive patients who underwent AWR during the study period, 194 (26.9%) underwent AWR following resection of abdominal wall tumors. In adjusted analyses, patients who had AWR after extirpative resection were more likely to have longer LOS (ß, 2.57; 95%CI, 1.27 to 3.86, p < 0.001) than those with primary herniorrhaphy, but the risk of HR, SSO, SSI, 30-day readmission, and reoperation did not differ significantly. In the extirpative cohort, obesity (Hazard ratio, 6.48; p = 0.003), and bridged repair (Hazard ratio, 3.50; p = 0.004) were predictors of HR. Radiotherapy (OR, 2.23; p = 0.017) and diabetes mellites (OR, 3.70; p = 0.005) were predictors of SSOs. Defect width (OR, 2.30; p < 0.001) and mesh length (OR, 3.32; p = 0.046) were predictors of SSIs. Concomitant intra-abdominal surgery for active disease was not associated with worse outcomes. CONCLUSIONS: AWR with ADM following extirpative resection demonstrated outcomes comparable with primary herniorrhaphy. Preoperative risk assessment and optimization are imperative for improving outcomes.

Automating the treatment planning process for 3D-conformal pediatric craniospinal irradiation therapy.

PURPOSE: Pediatric patients with medulloblastoma in low- and middle-income countries (LMICs) are most treated with 3D-conformal photon craniospinal irradiation (CSI), a time-consuming, complex treatment to plan, especially in resource-constrained settings. Therefore, we developed and tested a 3D-conformal CSI autoplanning tool for varying patient lengths. METHODS AND MATERIALS: Autocontours were generated with a deep learning model trained:tested (80:20 ratio) on 143 pediatric medulloblastoma CT scans (patient ages: 2-19 years, median = 7 years). Using the verified autocontours, the autoplanning tool generated two lateral brain fields matched to a single spine field, an extended single spine field, or two matched spine fields. Additional spine subfields were added to optimize the corresponding dose distribution. Feathering was implemented (yielding nine to 12 fields) to give a composite plan. Each planning approach was tested on six patients (ages 3-10 years). A pediatric radiation oncologist assessed clinical acceptability of each autoplan. RESULTS: The autocontoured structures' average Dice similarity coefficient ranged from .65 to .98. The average V95 for the brain/spinal canal for single, extended, and multi-field spine configurations was 99.9% ± 0.06%/99.9% ± 0.10%, 99.9% ± 0.07%/99.4% ± 0.30%, and 99.9% ± 0.06%/99.4% ± 0.40%, respectively. The average maximum dose across all field configurations to the brainstem, eyes (L/R), lenses (L/R), and spinal cord were 23.7 ± 0.08, 24.1 ± 0.28, 13.3 ± 5.27, and 25.5 ± 0.34 Gy, respectively (prescription = 23.4 Gy/13 fractions). Of the 18 plans tested, all were scored as clinically acceptable as-is or clinically acceptable with minor, time-efficient edits preferred or required. No plans were scored as clinically unacceptable. CONCLUSION: The autoplanning tool successfully generated pediatric CSI plans for varying patient lengths in 3.50 ± 0.4 minutes on average, indicating potential for an efficient planning aid in a resource-constrained settings.

Customizable landmark-based field aperture design for automated whole-brain radiotherapy treatment planning.

PURPOSE: To develop and evaluate an automated whole-brain radiotherapy (WBRT) treatment planning pipeline with a deep learning-based auto-contouring and customizable landmark-based field aperture design. METHODS: The pipeline consisted of the following steps: (1) Auto-contour normal structures on computed tomography scans and digitally reconstructed radiographs using deep learning techniques, (2) locate the landmark structures using the beam's-eye-view, (3) generate field apertures based on eight different landmark rules addressing different clinical purposes and physician preferences. Two parallel approaches for generating field apertures were developed for quality control. The performance of the generated field shapes and dose distributions were compared with the original clinical plans. The clinical acceptability of the plans was assessed by five radiation oncologists from four hospitals. RESULTS: The performance of the generated field apertures was evaluated by the Hausdorff distance (HD) and mean surface distance (MSD) from 182 patients' field apertures used in the clinic. The average HD and MSD for the generated field apertures were 16 ± 7 and 7 ± 3 mm for the first approach, respectively, and 17 ± 7 and 7 ± 3 mm, respectively, for the second approach. The differences regarding HD and MSD between the first and the second approaches were 1 ± 2 and 1 ± 3 mm, respectively. A clinical review of the field aperture design, conducted using 30 patients, achieved a 100% acceptance rate for both the first and second approaches, and the plan review achieved a 100% acceptance rate for the first approach and a 93% acceptance rate for the second approach. The average acceptance rate for meeting lens dosimetric recommendations was 80% (left lens) and 77% (right lens) for the first approach, and 70% (both left and right lenses) for the second approach, compared with 50% (left lens) and 53% (right lens) for the clinical plans. CONCLUSION: This study provided an automated pipeline with two field aperture generation approaches to automatically generate WBRT treatment plans. Both quantitative and qualitative evaluations demonstrated that our novel pipeline was comparable with the original clinical plans.

Identifying the optimal deep learning architecture and parameters for automatic beam aperture definition in 3D radiotherapy.

PURPOSE: Two-dimensional radiotherapy is often used to treat cervical cancer in low- and middle-income countries, but treatment planning can be challenging and time-consuming. Neural networks offer the potential to greatly decrease planning time through automation, but the impact of the wide range of hyperparameters to be set during training on model accuracy has not been exhaustively investigated. In the current study, we evaluated the effect of several convolutional neural network architectures and hyperparameters on 2D radiotherapy treatment field delineation. METHODS: Six commonly used deep learning architectures were trained to delineate four-field box apertures on digitally reconstructed radiographs for cervical cancer radiotherapy. A comprehensive search of optimal hyperparameters for all models was conducted by varying the initial learning rate, image normalization methods, and (when appropriate) convolutional kernel size, the number of learnable parameters via network depth and the number of feature maps per convolution, and nonlinear activation functions. This yielded over 1700 unique models, which were all trained until performance converged and then tested on a separate dataset. RESULTS: Of all hyperparameters, the choice of initial learning rate was most consistently significant for improved performance on the test set, with all top-performing models using learning rates of 0.0001. The optimal image normalization was not consistent across architectures. High overlap (mean Dice similarity coefficient = 0.98) and surface distance agreement (mean surface distance < 2 mm) were achieved between the treatment field apertures for all architectures using the identified best hyperparameters. Overlap Dice similarity coefficient (DSC) and distance metrics (mean surface distance and Hausdorff distance) indicated that DeepLabv3+ and D-LinkNet architectures were least sensitive to initial hyperparameter selection. CONCLUSION: DeepLabv3+ and D-LinkNet are most robust to initial hyperparameter selection. Learning rate, nonlinear activation function, and kernel size are also important hyperparameters for improving performance.

The Emergence of Artificial Intelligence within Radiation Oncology Treatment Planning.

BACKGROUND: The future of artificial intelligence (AI) heralds unprecedented change for the field of radiation oncology. Commercial vendors and academic institutions have created AI tools for radiation oncology, but such tools have not yet been widely adopted into clinical practice. In addition, numerous discussions have prompted careful thoughts about AI's impact upon the future landscape of radiation oncology: How can we preserve innovation, creativity, and patient safety? When will AI-based tools be widely adopted into the clinic? Will the need for clinical staff be reduced? How will these devices and tools be developed and regulated? SUMMARY: In this work, we examine how deep learning, a rapidly emerging subset of AI, fits into the broader historical context of advancements made in radiation oncology and medical physics. In addition, we examine a representative set of deep learning-based tools that are being made available for use in external beam radiotherapy treatment planning and how these deep learning-based tools and other AI-based tools will impact members of the radiation treatment planning team. Key Messages: Compared to past transformative innovations explored in this article, such as the Monte Carlo method or intensity-modulated radiotherapy, the development and adoption of deep learning-based tools is occurring at faster rates and promises to transform practices of the radiation treatment planning team. However, accessibility to these tools will be determined by each clinic's access to the internet, web-based solutions, or high-performance computing hardware. As seen by the trends exhibited by many technologies, high dependence on new technology can result in harm should the product fail in an unexpected manner, be misused by the operator, or if the mitigation to an expected failure is not adequate. Thus, the need for developers and researchers to rigorously validate deep learning-based tools, for users to understand how to operate tools appropriately, and for professional bodies to develop guidelines for their use and maintenance is essential. Given that members of the radiation treatment planning team perform many tasks that are automatable, the use of deep learning-based tools, in combination with other automated treatment planning tools, may refocus tasks performed by the treatment planning team and may potentially reduce resource-related burdens for clinics with limited resources.

Beam energy metrics for the acceptance and quality assurance of Halcyon linear accelerator.

PURPOSE: Establish and compare two metrics for monitoring beam energy changes in the Halcyon platform and evaluate the accuracy of these metrics across multiple Halcyon linacs. METHOD: The first energy metric is derived from the diagonal normalized flatness (FDN ), which is defined as the ratio of the average measurements at a fixed off-axis equal distance along the open profiles in two diagonals to the measurement at the central axis with an ionization chamber array (ICA). The second energy metric comes from the area ratio (AR) of the quad wedge (QW) profiles measured with the QW on the top of the ICA. Beam energy is changed by adjusting the magnetron current in a non-clinical Halcyon. With D10cm measured in water at each beam energy, the relationships between FDN or AR energy metrics to D10cm in water is established with linear regression across six energy settings. The coefficients from these regressions allow D10cm (FDN ) calculation from FDN using open profiles and D10cm (QW) calculation from AR using QW profiles. RESULTS: Five Halcyon linacs from five institutions were used to evaluate the accuracy of the D10cm (FDN ) and the D10cm (QW) energy metrics by comparing to the D10cm values computed from the treatment planning system (TPS) and D10cm measured in water. For the five linacs, the D10cm (FDN ) reported by the ICA based on FDN from open profiles agreed with that calculated by TPS within -0.29 ± 0.23% and 0.61% maximum discrepancy; the D10cm (QW) reported by the QW profiles agreed with that calculated by TPS within -0.82 ± 1.27% and -2.43% maximum discrepancy. CONCLUSION: The FDN -based energy metric D10cm (FDN ) can be used for acceptance testing of beam energy, and also for the verification of energy in periodic quality assurance (QA) processes.

Clinical implementation of automated treatment planning for whole-brain radiotherapy.

The purpose of the study was to develop and clinically deploy an automated, deep learning-based approach to treatment planning for whole-brain radiotherapy (WBRT). We collected CT images and radiotherapy treatment plans to automate a beam aperture definition from 520 patients who received WBRT. These patients were split into training (n = 312), cross-validation (n = 104), and test (n = 104) sets which were used to train and evaluate a deep learning model. The DeepLabV3+ architecture was trained to automatically define the beam apertures on lateral-opposed fields using digitally reconstructed radiographs (DRRs). For the beam aperture evaluation, 1st quantitative analysis was completed using a test set before clinical deployment and 2nd quantitative analysis was conducted 90 days after clinical deployment. The mean surface distance and the Hausdorff distances were compared in the anterior-inferior edge between the clinically used and the predicted fields. Clinically used plans and deep-learning generated plans were evaluated by various dose-volume histogram metrics of brain, cribriform plate, and lens. The 1st quantitative analysis showed that the average mean surface distance and Hausdorff distance were 7.1 mm (±3.8 mm) and 11.2 mm (±5.2 mm), respectively, in the anterior-inferior edge of the field. The retrospective dosimetric comparison showed that brain dose coverage (D99%, D95%, D1%) of the automatically generated plans was 29.7, 30.3, and 32.5 Gy, respectively, and the average dose of both lenses was up to 19.0% lower when compared to the clinically used plans. Following the clinical deployment, the 2nd quantitative analysis showed that the average mean surface distance and Hausdorff distance between the predicted and clinically used fields were 2.6 mm (±3.2 mm) and 4.5 mm (±5.6 mm), respectively. In conclusion, the automated patient-specific treatment planning solution for WBRT was implemented in our clinic. The predicted fields appeared consistent with clinically used fields and the predicted plans were dosimetrically comparable.

Acceptance and verification of the Halcyon-Eclipse linear accelerator-treatment planning system without 3D water scanning system.

We tested whether an ionization chamber array (ICA) and a one-dimensional water scanner (1DS) could be used instead of a three-dimensional water scanning system (3DWS) for acceptance testing and commissioning verification of the Varian Halcyon-Eclipse Treatment Planning System (TPS). The Halcyon linear accelerator has a single 6-MV flattening-filter-free beam and a nonadjustable beam model for the TPS. Beam data were measured with a 1DS, ICA, ionization chambers, and electrometer. Acceptance testing and commissioning were done simultaneously by comparing the measured data with TPS-calculated percent-depth-dose (PDD) and profiles. The ICA was used to measure profiles of various field sizes (10-, 20-, and 28 cm2 ) at depths of dmax (1.3 cm), 5-, 10-, and 20 cm. The 1DS was used for output factors (OFs) and PDDs. OFs were measured with 1DS for various fields (2-28 cm2 ) at a source-to-surface distance of 90 cm. All measured data were compared with TPS-calculations. Profiles, off-axis ratios (OAR), PDDs and OFs were also measured with a 3DWS as a secondary check. Profiles between the ICA and TPS (ICA and 3DWS) at various depths across the fields indicated that the maximum discrepancies in high-dose and low-dose tail were within 2% and 3%, respectively, and the maximum distance-to-agreement in the penumbra region was <3 mm. The largest OAR differences between ICA and TPS (ICA and 3DWS) values were 0.23% (-0.25%) for a 28 × 28 cm2 field, and the largest point-by-point PDD differences between 1DS and TPS (1DS and 3DWS) were -0.41% ± 0.12% (-0.32% ± 0.17%) across the fields. Both OAR and PDD showed the beam energy is well matched to the TPS model. The average ratios of 1DS-measured OFs to the TPS (1DS to 3DWS) values were 1.000 ± 0.002 (0.999 ± 0.003). The Halcyon-Eclipse system can be accepted and commissioned without the need for a 3DWS.

Dosimetric impact and detectability of multi-leaf collimator positioning errors on Varian Halcyon.

The purpose of this study is to investigate the dosimetric impact of multi-leaf collimator (MLC) positioning errors on a Varian Halcyon for both random and systematic errors, and to evaluate the effectiveness of portal dosimetry quality assurance in catching clinically significant changes caused by these errors. Both random and systematic errors were purposely added to 11 physician-approved head and neck volumetric modulated arc therapy (VMAT) treatment plans, yielding a total of 99 unique plans. Plans were then delivered on a preclinical Varian Halcyon linear accelerator and the fluence was captured by an opposed portal dosimeter. When comparing dose-volume histogram (DVH) values of plans with introduced MLC errors to known good plans, clinically significant changes to target structures quickly emerged for plans with systematic errors, while random errors caused less change. For both error types, the magnitude of clinically significant changes increased as error size increased. Portal dosimetry was able to detect all systematic errors, while random errors of ±5 mm or less were unlikely to be detected. Best detection of clinically significant errors, while minimizing false positives, was achieved by following the recommendations of AAPM TG-218. Furthermore, high- to moderate correlation was found between dose DVH metrics for normal tissues surrounding the target and portal dosimetry pass rates. Therefore, it may be concluded that portal dosimetry on the Halcyon is robust enough to detect errors in MLC positioning before they introduce clinically significant changes to VMAT treatment plans.

Independent validation of machine performance check for the Halcyon and TrueBeam linacs for daily quality assurance.

PURPOSE: To evaluate the ability of the machine performance check (MPC) on the Halcyon to detect errors, with comparison with the TrueBeam. METHODS: MPC is an automated set of quality assurance (QA) tests that use a phantom placed on the couch and the linac's imaging system(s) to verify the beam constancy and mechanical performance of the Halcyon and TrueBeam linacs. In order to evaluate the beam constancy tests, we inserted solid water slabs between the beam source and the megavoltage imager to simulate changes in beam output, flatness, and symmetry. The MPC results were compared with measurements, using two-dimensional array under the same conditions. We then studied the accuracy of MPC geometric tests. The accuracies of the relative gantry offset and couch shift tests were evaluated by intentionally inserting phantom shifts, using a rotating or linear motion stage. The MLC offset and absolute gantry offset tests were assessed by miscalibrating these motions on a Halcyon linac. RESULTS: For the Halcyon system, the average difference in the measured beam output between the IC Profiler and MPC, after intentional changes, was 1.3 ± 0.5% (for changes ≤5%). For Halcyon, the MPC test failed (i.e., prevented treatment) when the beam symmetry change was over 1.9%. The accuracy of the MLC offset test was within 0.05 mm. The absolute gantry offset test was able to detect an offset as small as 0.02°. The accuracy of the absolute couch shift test was 0.03 mm. The accuracy of relative couch shift test of Halcyon was measured as 0.16 mm. CONCLUSION: We intentionally inserted errors to evaluate the ability of the MPC to identify errors in dosimetric and geometric parameters. These results showed that the MPC is sufficiently accurate to be effectively used for daily QA of the Halcyon and TrueBeam treatment devices.

Normal tissue doses from MV image-guided radiation therapy (IGRT) using orthogonal MV and MV-CBCT.

PURPOSE: The aim of this study was to measure and compare the mega-voltage imaging dose from the Halcyon medical linear accelerator (Varian Medical Systems) with measured imaging doses with the dose calculated by Eclipse treatment planning system. METHODS: An anthropomorphic thorax phantom was imaged using all imaging techniques available with the Halcyon linac - MV cone-beam computed tomography (MV-CBCT) and orthogonal anterior-posterior/lateral pairs (MV-MV), both with high-quality and low-dose modes. In total, 54 imaging technique, isocenter position, and field size combinations were evaluated. The imaging doses delivered to 11 points in the phantom (in-target and extra-target) were measured using an ion chamber, and compared with the imaging doses calculated using Eclipse. RESULTS: For high-quality MV-MV mode, the mean extra-target doses delivered to the heart, left lung, right lung and spine were 1.18, 1.64, 0.80, and 1.11 cGy per fraction, respectively. The corresponding mean in-target doses were 3.36, 3.72, 2.61, and 2.69 cGy per fraction, respectively. For MV-MV technique, the extra-target imaging dose had greater variation and dependency on imaging field size than did the in-target dose. Compared to MV-MV technique, the imaging dose from MV-CBCT was less sensitive to the location of the organ relative to the treatment field. For high-quality MV-CBCT mode, the mean imaging doses to the heart, left lung, right lung, and spine were 8.45, 7.16, 7.19, and 6.51 cGy per fraction, respectively. For both MV-MV and MV-CBCT techniques, the low-dose mode resulted in an imaging dose about half of that in high-quality mode. CONCLUSION: The in-target doses due to MV imaging using the Halcyon ranged from 0.59 to 9.75 cGy, depending on the choice of imaging technique. Extra-target doses from MV-MV technique ranged from 0 to 2.54 cGy. The MV imaging dose was accurately calculated by Eclipse, with maximum differences less than 0.5% of a typical treatment dose (assuming a 60 Gy prescription). Therefore, the cumulative imaging and treatment plan dose distribution can be expected to accurately reflect the actual dose.

A snapshot of medical physics practice patterns.

A large number of surveys have been sent to the medical physics community addressing many clinical topics for which the medical physicist is, or may be, responsible. Each survey provides an insight into clinical practice relevant to the medical physics community. The goal of this study was to create a summary of these surveys giving a snapshot of clinical practice patterns. Surveys used in this study were created using SurveyMonkey and distributed between February 6, 2013 and January 2, 2018 via the MEDPHYS and MEDDOS listserv groups. The format of the surveys included questions that were multiple choice and free response. Surveys were included in this analysis if they met the following criteria: more than 20 responses, relevant to radiation therapy physics practice, not single-vendor specific, and formatted as multiple-choice questions (i.e., not exclusively free-text responses). Although the results of free response questions were not explicitly reported, they were carefully reviewed, and the responses were considered in the discussion of each topic. Two-hundred and fifty-two surveys were available, of which 139 passed the inclusion criteria. The mean number of questions per survey was 4. The mean number of respondents per survey was 63. Summaries were made for the following topics: simulation, treatment planning, electron treatments, linac commissioning and quality assurance, setup and treatment verification, IMRT and VMAT treatments, SRS/SBRT, breast treatments, prostate treatments, brachytherapy, TBI, facial lesion treatments, clinical workflow, and after-hours/emergent treatments. We have provided a coherent overview of medical physics practice according to surveys conducted over the last 5 yr, which will be instructive for medical physicists.

Cost-effective immobilization for whole brain radiation therapy.

To investigate the inter- and intra-fraction motion associated with the use of a low-cost tape immobilization technique as an alternative to thermoplastic immobilization masks for whole-brain treatments. The results of this study may be of interest to clinical staff with severely limited resources (e.g., in low-income countries) and also when treating patients who cannot tolerate standard immobilization masks. Setup reproducibility of eight healthy volunteers was assessed for two different immobilization techniques. (a) One strip of tape was placed across the volunteer's forehead and attached to the sides of the treatment table. (b) A second strip was added to the first, under the chin, and secured to the table above the volunteer's head. After initial positioning, anterior and lateral photographs were acquired. Volunteers were positioned five times with each technique to allow calculation of inter-fraction reproducibility measurements. To estimate intra-fraction reproducibility, 5-minute anterior and lateral videos were taken for each technique per volunteer. An in-house software was used to analyze the photos and videos to assess setup reproducibility. The maximum intra-fraction displacement for all volunteers was 2.8 mm. Intra-fraction motion increased with time on table. The maximum inter-fraction range of positions for all volunteers was 5.4 mm. The magnitude of inter-fraction and intra-fraction motion found using the "1-strip" and "2-strip" tape immobilization techniques was comparable to motion restrictions provided by a thermoplastic mask for whole-brain radiotherapy. The results suggest that tape-based immobilization techniques represent an economical and useful alternative to the thermoplastic mask.

METRIC-GUIDED IMAGE RECONSTRUCTION BOUNDS VIA CONFORMAL PREDICTION.

Recent advancements in machine learning have led to the development of novel medical imaging systems and algorithms that address ill-posed problems. Assessing their trustworthiness and understanding how to deploy them safely at test time remains an important and open problem. In this work, we propose using conformal prediction to compute valid and distribution-free bounds on downstream metrics given reconstructions generated by one algorithm, and retrieve upper/lower bounds and inlier/outlier reconstructions according to the adjusted bounds. Our work offers 1) test time image reconstruction evaluation without ground truth, 2) downstream performance guarantees, 3) meaningful upper/lower bound reconstructions, and 4) meaningful statistical inliers/outlier reconstructions. We demonstrate our method on post-mastectomy radiotherapy planning using 3D breast CT reconstructions, and show 1) that metric-guided bounds have valid coverage for downstream metrics while conventional pixel-wise bounds do not and 2) anatomical differences of upper/lower bounds between metric-guided and pixel-wise methods. Our work paves way for more meaningful and trustworthy test-time evaluation of medical image reconstructions. Code available at https://github.com/matthewyccheung/conformal-metric.

Severe Lymphopenia During Chemoradiation Therapy for Esophageal Cancer: Comprehensive Analysis of Randomized Phase 2B Trial of Proton Beam Therapy Versus Intensity Modulated Radiation Therapy.

PURPOSE: Lymphocytes play an important role in antitumor immunity; however, they are also especially vulnerable to depletion during chemoradiation therapy (CRT). The purpose of this study was to compare the incidence of grade 4 lymphopenia (G4L) between proton beam therapy (PBT) and intensity modulated photon radiation therapy (IMRT) in patients with esophageal cancer treated with CRT in a completed randomized trial and to ascertain patient heterogeneity to G4L risk based on treatment and established prognostic factors. METHODS AND MATERIALS: Between April 2012 and March 2019, a single-institution, open-label, nonblinded, phase 2 randomized trial (NCT01512589) was conducted at the University of Texas MD Anderson Cancer Center. Patients were randomly assigned to IMRT or PBT, either definitively or preoperatively. This secondary analysis of the randomized trial was G4L during concurrent CRT according to Common Terminology Criteria for Adverse Events version 5.0. RESULTS: Among 105 patients evaluable for analysis, 44 patients (42%) experienced G4L at a median of 28 days after the start date of concurrent CRT. Induction chemotherapy (P = .003), baseline absolute lymphocyte count (P < .001), radiation therapy modality (P = .002), and planning treatment volume (P = .033) were found to be significantly associated with G4L. Multivariate classification analysis partitioned patients into 5 subgroups for whom the incidence of G4L was observed in 0%, 14%, 35%, 70%, and 100% of patients. The benefit of PBT over IMRT was most pronounced in patients with an intermediate baseline absolute lymphocyte count and large planning treatment volume (P = .011). CONCLUSIONS: This is the first prospective evidence that limiting dose scatter by PBT significantly reduced the incidence of G4L, especially in the intermediate-risk patients. The implication of this immune-sparing effect of PBT, especially in the context of standard adjuvant immunotherapy, needs further examination in the current phase 3 randomized trials.

Analyzing the Relationship between Dose and Geometric Agreement Metrics for Auto-Contouring in Head and Neck Normal Tissues.

This study aimed to determine the relationship between geometric and dosimetric agreement metrics in head and neck (H&N) cancer radiotherapy plans. A total 287 plans were retrospectively analyzed, comparing auto-contoured and clinically used contours using a Dice similarity coefficient (DSC), surface DSC (sDSC), and Hausdorff distance (HD). Organs-at-risk (OARs) with ≥200 cGy dose differences from the clinical contour in terms of Dmax (D0.01cc) and Dmean were further examined against proximity to the planning target volume (PTV). A secondary set of 91 plans from multiple institutions validated these findings. For 4995 contour pairs across 19 OARs, 90% had a DSC, sDSC, and HD of at least 0.75, 0.86, and less than 7.65 mm, respectively. Dosimetrically, the absolute difference between the two contour sets was <200 cGy for 95% of OARs in terms of Dmax and 96% in terms of Dmean. In total, 97% of OARs exhibiting significant dose differences between the clinically edited contour and auto-contour were within 2.5 cm PTV regardless of geometric agreement. There was an approximately linear trend between geometric agreement and identifying at least 200 cGy dose differences, with higher geometric agreement corresponding to a lower fraction of cases being identified. Analysis of the secondary dataset validated these findings. Geometric indices are approximate indicators of contour quality and identify contours exhibiting significant dosimetric discordance. For a small subset of OARs within 2.5 cm of the PTV, geometric agreement metrics can be misleading in terms of contour quality.

Artificial Intelligence Uncertainty Quantification in Radiotherapy Applications - A Scoping Review.

Background/purpose: The use of artificial intelligence (AI) in radiotherapy (RT) is expanding rapidly. However, there exists a notable lack of clinician trust in AI models, underscoring the need for effective uncertainty quantification (UQ) methods. The purpose of this study was to scope existing literature related to UQ in RT, identify areas of improvement, and determine future directions. Methods: We followed the PRISMA-ScR scoping review reporting guidelines. We utilized the population (human cancer patients), concept (utilization of AI UQ), context (radiotherapy applications) framework to structure our search and screening process. We conducted a systematic search spanning seven databases, supplemented by manual curation, up to January 2024. Our search yielded a total of 8980 articles for initial review. Manuscript screening and data extraction was performed in Covidence. Data extraction categories included general study characteristics, RT characteristics, AI characteristics, and UQ characteristics. Results: We identified 56 articles published from 2015-2024. 10 domains of RT applications were represented; most studies evaluated auto-contouring (50%), followed by image-synthesis (13%), and multiple applications simultaneously (11%). 12 disease sites were represented, with head and neck cancer being the most common disease site independent of application space (32%). Imaging data was used in 91% of studies, while only 13% incorporated RT dose information. Most studies focused on failure detection as the main application of UQ (60%), with Monte Carlo dropout being the most commonly implemented UQ method (32%) followed by ensembling (16%). 55% of studies did not share code or datasets. Conclusion: Our review revealed a lack of diversity in UQ for RT applications beyond auto-contouring. Moreover, there was a clear need to study additional UQ methods, such as conformal prediction. Our results may incentivize the development of guidelines for reporting and implementation of UQ in RT.

Training Robust T1-Weighted Magnetic Resonance Imaging Liver Segmentation Models Using Ensembles of Datasets with Different Contrast Protocols and Liver Disease Etiologies.

Image segmentation of the liver is an important step in several treatments for liver cancer. However, manual segmentation at a large scale is not practical, leading to increasing reliance on deep learning models to automatically segment the liver. This manuscript develops a deep learning model to segment the liver on T1w MR images. We sought to determine the best architecture by training, validating, and testing three different deep learning architectures using a total of 819 T1w MR images gathered from six different datasets, both publicly and internally available. Our experiments compared each architecture's testing performance when trained on data from the same dataset via 5-fold cross validation to its testing performance when trained on all other datasets. Models trained using nnUNet achieved mean Dice-Sorensen similarity coefficients > 90% when tested on each of the six datasets individually. The performance of these models suggests that an nnUNet liver segmentation model trained on a large and diverse collection of T1w MR images would be robust to potential changes in contrast protocol and disease etiology.