An Automated, Dynamic Radiation Oncology Prescription Checking System

PURPOSE

Despite serving as a critical communication tool, radiation oncology prescriptions are entered manually and prone to error. An automated prescription checking system was developed and implemented to help address this problem.

METHODS AND MATERIALS

Rules defining clinically appropriate prescriptions were generated, examining specific types of errors: (1) unapproved dose per fraction for a given disease site; (2) dose per fraction too large for nonstereotactic treatment technique; and (3) dose per fraction too low. With a goal of catching errors as upstream as possible to minimize their propagation, a report was created and ran every 30 minutes to check all newly written or approved prescriptions against the 3 rules. When a prescription violated these rules, an automated email was immediately sent to the prescriber alerting them of the potential error. System performance was continuously monitored and the criteria triggering an alert adjusted to balance error detection against false positives. Alerts leading to prescription amendment were considered true errors.

RESULTS

From June 2021 to November 2022, the system checked 24,047 prescriptions. A total of 241 email alerts were triggered, for an average alert rate of 1%. Of the 241 alerts, 198 (82.2%) were unapproved doses per fraction for the disease site, 14 (5.8%) were doses per fraction that were too low, and 29 (12%) were doses too large for nonstereotactic treatment technique. Thirty-one percent of alerts led to a change of prescription, suggesting they were true errors. The baseline rate of erroneous prescription entry was 0.3%. A regression model showed that trainee prescription entry and dose per fraction <150 cGy were significantly associated with true errors.

CONCLUSIONS

Given the significant consequences of erroneous prescription entry, which ranged from wasted resources and treatment delays to potentially serious misadministration, there is significant value in implementing automated prescription checking systems in radiation oncology clinics.

Prospective Evaluation of Implementation of a Tattoo-less Workflow for Non-Spine Bone SBRT

PURPOSE/OBJECTIVE(S)

Oligometastatic disease has expanded the clinical indications for non-spine bone (NSB) SBRT. Optical surface monitoring systems (OSMS) may reduce treatment time if it represents an effective surrogate for bone intrafraction motion monitoring. We aimed to identify whether OSMS could substitute for 2D-3D mid-imaging and enable a tattoo-less set up in NSB SBRT.

MATERIALS/METHODS

Beginning 11/2019, OSMS was incorporated in parallel with an existing workflow using CBCT and mid 2D-3D kV/kV imaging for pre- and mid-imaging for NSB SBRT. The ability of OSMS to detect both observed out-of-tolerance (>2mm/>2deg) shifts and manual couch shifts was analyzed. A workflow incorporating OSMS reference captures, CBCT for pre-treatment verification and OSMS/triggered imaging (TI) for intrafraction monitoring was developed and deployed for rib/sternum SBRT beginning 11/2021 and all NSB SBRT beginning 2/2022. All NSB SBRT treatment appointments were analyzed through statistical process control (SPC) with use of an XmR chart of average set up and total treatment time per quarter from 2/2019 to 2/2023. Special cause rules were based on IHI rules and conduced with spreadsheet software.

RESULTS

From 2/2019 to 2/2023, 1962 NSB SBRT fractions were delivered, including 337 rib, 150 sternum, 197 femur, 266 ilium, 222 multi-site. Over 104 femur and 87 ilium images, there were no over tolerance intra-fraction events detected with 2D-3D or OSMS. Over 20 manual shifts, OSMS could detect 2mm shifts to within 0.4mm 67% of the time and 0.8mm 95% of time. There was no difference in treatment set up time following adoption of an OSMS/TI workflow as a replacement for 2D-3D mid-imaging. A reduction in rib SBRT delivery and multi-site treatment set-up times was significantly associated with the adoption of OSMS/TI and OSMS, respectively, as assessed based on special cause variation with 8 consecutive points below the mean.

CONCLUSION

Integration of OSMS and triggered imaging has enabled the transition to a completely tattoo-less workflow, thus sparing patients the need for permanent tattoos whilst also allowing more continuous motion monitoring and reduced radiation exposure related to unnecessary 2D-3D or CBCT mid-imaging. Treatment times were significantly reduced for patients receiving rib SBRT or multi-site NSB SBRT with this workflow.

Deep Learning-Based Automated Quality Assurance for Palliative Spinal Treatment Planning in Radiotherapy

PURPOSE/OBJECTIVE(S)

Robustquality assurance (QA) for palliative spine radiation therapy (RT) remains critical due to the risk of wrong anatomic level treatment on account of human error in enumerating vertebral bodies accurately based on morphology with incomplete imaging of the spine and prevalence of anatomic variants (10%). We propose a rapid, fully automated deep learning-based QA (DL-QA) tool for segmenting and enumerating vertebral structures from image data, capable of identifying misalignment based on discrepancies in calculated dose coverage.

MATERIALS/METHODS

Ina retrospective cohort of 514 patients who received palliative spine radiation treatment at a single institution for spinal metastases, vertebral volumes for each individual spine level were automatically segmented on RT planning computed tomography scans using a publicly available deep learning algorithm, Total Segmentator (TS) deployed in the treatment planning system (Wasserthal et al, 2022). Departmental policy requires that the prescription/plan name include all spinal levels that receive a prescribed dosimetric threshold of V50% > 50%. By comparing the intended spine level target in the prescription and plan name against the TS volumes, the DL-QA flagged all cases for which any target vertebrae did not receive this threshold dose and/or any non-targeted vertebrae that received V50% > 50%. To detect spine anomalies, cases were also flagged if any vertebrae volume was not within ±1σ of the entire population of vertebrae volumes. Flagged cases were either categorized as: (1) wrong spine level RT error; (2) documentation error, in which treatment was correct but the prescription/plan name did not follow Departmental policy; or (3) potential spinal geometric error. All flagged cases were verified manually by checking the original images and treatment planning data.

RESULTS

Outof 514 patients, 29 cases were flagged as potential errors. Manual review revealed that one of these was a previously discovered true treatment error (due to anatomic variant with 4 lumbar bodies) while 10 were treated as intended but showed documentation errors due to variants in the number of vertebral bodies, kyphosis of the spine causing non-targeted vertebrae to appear in the treatment field, or improper observation of the Departmental plan naming policy. The remaining 18 cases were associated with flagged vertebrae volumes. Reviewing those patients, we identified spinal anomalies where TS attempted to account for extra or missing vertebrae (N = 9) and cases where TS made segmentation errors (N = 9).

CONCLUSION

Theproposed automated DL-QA system successfully identified patients with spine anomalies, flagged documentation errors, verified the correct target levels of spine RT treatments, and detected a known misadministration. The next phase will involve prospective testing of the system in a clinical setting upstream of treatment delivery.

Abstract P1-06-06: Alteration of stromal collagen fiber orientation in DCIS

Approximately 20% of new diagnoses of breast cancer are ductal carcinoma in situ (DCIS), a non-invasive form of breast cancer. Treatment decision-making for DCIS is challenging since current predictors of disease-free survival are limited, so that most women are presented with options for surgery, radiation and tamoxifen - all options with consequences for quality of life. Prior studies of prognostic factors for DCIS have focused on morphologic, genetic, and protein expression patterns of the DCIS cells. However, laboratory evidence suggests that the tumor microenvironment may play a key role in tumor invasion and progression. Collagen is the most abundant component of the stroma surrounding the breast ducts in which cancers develop. We previously observed that, in invasive breast cancer, tumors with greater numbers of collagen fibers aligned perpendicularly from the tumor were more likely to predict poor survival than tumors with collagen fibers in primarily parallel patterns near the tumor boundary (Conklin Am J Pathol 2011). To improve our ability to predict breast cancer outcomes in women with DCIS, we examined the alignment of collagen adjacent to ducts affected by DCIS to test whether alignment patterns were similar to patterns observed in tissue labeled as “normal” from biopsy and surgical sections. We evaluated collagen alignment in 255 Wisconsin women diagnosed with DCIS in 1997-2000 and followed for a median of 11.2 years (range 1-15). Stromal collagen alignment was evaluated from routine H&E tissue slides prepared at the time of diagnosis using second harmonic generation (SHG) microscopy, a label-free multiphoton laser scanning technique that selectively images collagen. SHG images were acquired and evaluated for 3-5 regions on each DCIS and normal slide for each patient; the angles of collagen fibers with respect to the DCIS lesion/stroma boundary were calculated using customized imaging software. Data for the distribution of angles were compared for normal ducts and DCIS lesions using compositional data analysis with the number of fibers totaled according to 5-angle bins (1-5, 6-10, 11-15, …, 86-90 degrees). Repeated measures linear regression models were fit to log-transformed ratios of binned counts as a function of tissue type. Dependence among repeated counts within a single region was modeled using an unstructured variance-covariance matrix. Dependence among measurements within a single subject was modeled using a compound symmetry correlation structure. Overall, the distribution of collagen fiber angles from DCIS lesions differed significantly (P = 0.0002) from the distribution of collagen fibers surrounding normal ducts. Collagen fibers surrounding DCIS lesions were 11-18% more likely to orient at 75-90 degrees relative to the lesion boundary than fibers surrounding normal ducts; fibers were more similarly aligned in both DCIS lesions and normal ducts at other smaller angles. These results underscore the relevance of the tumor microenvironment, in particular the arrangement of the collagen fiber matrix. Planned data analysis will next examine whether collagen fiber alignment patterns differ between DCIS patients who did and did not experience a second breast cancer diagnosis over the course of follow-up.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P1-06-06.