Analysis of early respiratory-related mortality after radiation therapy of non-small-cell lung cancer: feasibility of automatic data extraction for dose-response studies

Impact of delineation errors on the estimated organ at risk dose and of dose errors on the normal tissue complication probability model

BACKGROUND

Normal tissue complication probability (NTCP) models are often based on doses retrieved from delineated volumes. For retrospective dose-response studies focusing on organs that have not been delineated historically, automatic segmentation might be considered. However, automatic segmentation risks generating considerable delineation errors and knowledge regarding how these errors impact the estimated organ dose is important. Furthermore, organ-at-risk (OAR) dose uncertainties cannot be eliminated and might affect the resulting NTCP model. Therefore, it is also of interest to study how OAR dose errors impact the NTCP modeling results.

PURPOSE

To investigate how random delineation errors of the proximal bronchial tree, heart, and esophagus impact the estimated OAR dose, and to investigate how random errors in the doses used for dose-response modeling affect the estimated NTCPs.

METHODS

We investigated the impact of random delineation errors on the estimated OAR dose using the treatment plans of 39 patients treated with conventionally fractionated radiation therapy of non-small-cell lung cancer. Study-specific reference structures were defined by manually contouring the proximal bronchial tree, heart and esophagus. For each patient and organ, 120 reshaped structures were created by introducing random shifts and margins to the entire reference structure. The mean and near-maximum dose to the reference and reshaped structures were compared. In a separate investigation, the impact of random dose errors on the NTCP model was studied performing dose-response modeling with study sets containing treatment outcomes and OAR doses with and without introduced errors. Universal patient populations with defined population risks, dose-response relationships and distributions of OAR doses were used as ground truth. From such a universal population, we randomly sampled data sets consisting of OAR dose and treatment outcome into reference populations. Study sets of different sizes were created by repeatedly introducing errors to the OAR doses of each reference population. The NTCP models generated with dose errors were compared to the reference NTCP model of the corresponding reference population.

RESULTS

A total of 14 040 reshaped structures with random delineation errors were created. The delineation errors resulted in systematic mean dose errors of less than 1% of the prescribed dose (PD). Mean dose differences above 15% of PD and near-maximum doses differences above 25% of PD were observed for 211 and 457 reshaped structures, respectively. Introducing random errors to OAR doses used for dose-response modeling resulted in systematic underestimations of the median NTCP. For all investigated scenarios, the median differences in NTCP were within 0.1 percentage points (p.p.) when comparing different study sizes.

CONCLUSIONS

Introducing random delineation errors to the proximal bronchial tree, heart and esophagus resulted in mean dose and near-maximum dose differences above 15% and 25% of PD, respectively. We did not observe an association between the dose level and the magnitude of the dose errors. For the scenarios investigated in this study, introducing random errors to OAR doses used for dose-response modeling resulted in systematic underestimations of the median NTCP for reference risks higher than the universal population risk. The median NTCP underestimation was similar for different study sizes, all within 0.1 p.p.

Automated data extraction tool (DET) for external applications in radiotherapy

Oncological Information Systems (OIS) manage information in radiotherapy (RT) departments. Due to database structure limitations, stored information can rarely be directly used except for vendor-specific purposes. Our aim is to enable the use of such data in various external applications by creating a tool for automatic data extraction, cleaning and formatting.

METHODS AND MATERIALS

We used OIS data from a nine-linac RT department in Sweden (70 weeks, 2015-16). Extracted data included patients' referrals and appointments with details for RT sub-tasks. The data extraction tool to prepare the data for external use was built in C# programming language. It used excel-automation queries to remove unassigned/duplicated values, substitute missing data and perform application-specific calculations. Descriptive statistics were used to verify the output with the manually prepared dataset from the corresponding time period.

RESULTS

From the initial raw data, 2030 (51 %)/907 (23 %) patients had known curative and palliative treatment intent for 84 different cancer diagnoses. After removal of incomplete entries, 373 (10 %) patients had unknown treatment intents which were substituted based on the known curative/palliative ratio. Automatically- and manuallyprepared datasets differed < 1 % for Mould, Treatment planning, Quality assurance and ± 5 % for Fractions and Magnetic resonance imaging with overestimations in 80/140 (57 %) entries by the tool.

CONCLUSION

We successfully implemented a software tool to prepare ready-to-use OIS datasets for external applications. Our evaluations showed overall results close to the manually-prepared dataset. The time taken to prepare the dataset using our automated strategy can reduce the time for manual preparation from weeks to seconds.

Harnessing data science to advance radiation oncology

Radiation oncology, a major treatment modality in the care of patients with malignant disease, is a technology‐ and computer‐intensive medical specialty. As such, it should lend itself ideally to data science methods, where computer science, statistics, and clinical knowledge are combined to advance state‐of‐the‐art care. Nevertheless, data science methods in radiation oncology research are still in their infancy and successful applications leading to improved patient care remain scarce. Here, we discuss data interoperability issues within and across organizational boundaries that hamper the introduction of big data and data science techniques in radiation oncology. At the semantic level, creating common underlying models and codification of the data, including the use of data elements with standardized definitions, an ontology, remains a work in progress. Methodological issues in data science and in the use of large population‐based health data registries are identified. We show that data science methods and big data cannot replace randomized clinical trials in comparative effectiveness research by reviewing a series of instances where the outcomes of big data analyses and randomized trials are at odds. We also discuss the modern wave of machine learning and artificial intelligence as represented by deep learning and convolutional neural networks. Finally, we identify promising research avenues and remain optimistic that the data sources in radiation oncology can be linked to yield important insights in the near future. We argue that data science will be a valuable complement to, but not a replacement of, the traditional hypothesis‐driven translational research chain and the randomized clinical trials that form the backbone of evidence‐based medicine.