Classification of the source of treatment deviation in proton therapy using prompt‐gamma imaging information

Khamfongkhruea, Chirasak; Berthold, Jonathan; Janssens, Guillaume; Petzoldt, J.; Smeets, J.; Pausch, G.; Richter, Christian

doi:10.1002/mp.14393

Cited by 5 publications

(23 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, using the existing IBA-camera, Khamfongkhruea et al used training data sets to identify the possible sources of errors (e.g. computed tomography (CT)-based range prediction, patient setup, and anatomical changes between fractions) [69]. Then they built a decision tree in order to classify the error sources.…”

Section: Discussionmentioning

confidence: 99%

Range Verification by Means of Prompt-Gamma Detection in Particle Therapy 1

Wrońska¹,

Dauvergne²

2021

Radiation Detection Systems

View full text Add to dashboard Cite

Ion range verification in particle therapy, possibly in real time, is still an open issue to improve the treatment quality by reducing safety margins. Among the various methods that are under study or already brought to the clinics, prompt gamma offer a unique opportunity of real time control. In this chapter, we present the physical background and conditions specific for the operation environment in the clinics, followed by an overview of the various techniques relying on prompt-gamma detection that have been proposed and developed for more than a decade. Different approaches, based on imaging and non-imaging techniques, providing either integrated or differential information are described, and their maturity, limitations and clinical usefulness are discussed.

show abstract

Section: Discussionmentioning

confidence: 99%

Range Verification by Means of Prompt-Gamma Detection in Particle Therapy 1

Wrońska¹,

Dauvergne²

2021

Radiation Detection Systems

View full text Add to dashboard Cite

show abstract

“…These scenarios correspond to treatment deviations that we aim to detect in a possible application of PGI in a clinical online adaptive workflow. In contrast to previous work, 12 the cases of range prediction errors of ±1% were classified as non‐relevant, as they lead to a maximum proton range shift of about 1 mm for the investigated head‐and‐neck treatment fields. For the cases of anatomical changes, a comparison between the dose on the pCT and cCT images using dose–volume histogram constraints was carried out.…”

Section: Methodsmentioning

confidence: 79%

“…We compared the performance of CNNs with conventional ML models trained on five established handcrafted features previously extracted from optimal PGI data 12 . The creation of these features was based on the used data set and ground‐truth classes (without the addition of the combination error class) and can be seen as the performance benchmark on optimal PGI data.…”

Section: Discussionmentioning

confidence: 99%

“…All cases were reviewed for the demand of treatment adaptation 12 . Each case was thereby classified as a clinically relevant (RE) or non‐relevant deviation (NR), respectively, which is described in detail in the following paragraph.…”

Section: Methodsmentioning

confidence: 99%

“…The basis of this study are CT imaging data of 12 patients and of an anthropomorphic heterogeneous head phantom (n = 13) with tissue equivalent materials (CIRS, Norfolk, USA) as well as PBS treatment plans already used in a previous study. 12 The phantom data set increases the size and variability of the training data and was included as we did not see any notable differences in extracted PGI range shift data between patients and phantom. The imaging data consist of a dual-energy CT (DECT) scan for the phantom as well as the clinical single-energy pCT and 2-5 (median 4) control CT (cCT) scans from patients with loco-regionally advanced head-and-neck squamous cell carcinoma, previously treated with intensity-modulated photon therapy at our institution between January and July 2016.…”

Section: Patient Datamentioning

confidence: 99%

See 2 more Smart Citations

Automatic detection and classification of treatment deviations in proton therapy from realistically simulated prompt gamma imaging data

et al. 2022

Self Cite

View full text Add to dashboard Cite

Background A clinical study regarding the potential of range verification in proton therapy (PT) by prompt gamma imaging (PGI) is carried out at our institution. Manual interpretation of the detected spot‐wise range shift information is time‐consuming, highly complex, and therefore not feasible in a broad routine application. Purpose Here, we present an approach to automatically detect and classify treatment deviations in realistically simulated PGI data for head‐and‐neck cancer (HNC) treatments using convolutional neural networks (CNNs) and conventional machine learning (ML) approaches. Methods For 12 HNC patients and 1 anthropomorphic head phantom (n = 13), pencil beam scanning (PBS) treatment plans were generated, and 1 field per plan was assumed to be monitored with a PGI slit camera system. In total, 386 scenarios resembling different relevant or non‐relevant treatment deviations were simulated on planning and control CTs and manually classified into 7 classes: non‐relevant changes (NR) and relevant changes (RE) triggering treatment intervention due to range prediction errors (±RP), setup errors in beam direction (±SE), anatomical changes (AC), or a combination of such errors (CB). PBS spots with reliable PGI information were considered with their nominal Bragg peak position for the generation of two 3D spatial maps of 16 × 16 × 16 voxels containing PGI‐determined range shift and proton number information. Three complexity levels of simulated PGI data were investigated: (I) optimal PGI data, (II) realistic PGI data with simulated Poisson noise based on the locally delivered proton number, and (III) realistic PGI data with an additional positioning uncertainty of the slit camera following an experimentally determined distribution. For each complexity level, 3D‐CNNs were trained on a data subset (n = 9) using patient‐wise leave‐one‐out cross‐validation and tested on an independent test cohort (n = 4). Both the binary task of detecting RE and the multi‐class task of classifying the underlying error source were investigated. Similarly, four different conventional ML classifiers (logistic regression, multilayer perceptron, random forest, and support vector machine) were trained using five previously established handcrafted features extracted from the PGI data and used for performance comparison. Results On the test data, the CNN ensemble achieved a binary accuracy of 0.95, 0.96, and 0.93 and a multi‐class accuracy of 0.83, 0.81, and 0.76 for the complexity levels (I), (II), and (III), respectively. In the case of binary classification, the CNN ensemble detected treatment deviations in the most realistic scenario with a sensitivity of 0.95 and a specificity of 0.88. The best performing ML classifiers showed a similar test performance. Conclusions This study demonstrates that CNNs can reliably detect relevant changes in realistically simulated PGI data and classify most of the underlying sources of treatment deviations. The CNNs extracted meaningful features from the PGI data with a performance comparable ...

show abstract