Christopher South

The role of texture analysis in imaging as an outcome predictor and potential tool in radiotherapy treatment planning

Predicting a tumours response to radiotherapy prior to the start of treatment could enhance clinical care management by enabling the personalization of treatment plans based on predicted outcome. In recent years, there has been accumulating evidence relating tumour texture to patient survival and response to treatment. Tumour texture could be measured from medical images that provide a non-invasive method of capturing intratumoural heterogeneity and hence could potentially enable a prior assessment of a patients predicted response to treatment. In this article, work presented in the literature regarding texture analysis in radiotherapy in relation to survival and outcome is discussed. Challenges facing integrating texture analysis in radiotherapy planning are highlighted and recommendations for future directions in research are suggested.

Clinical validation and benchmarking of knowledge-based IMRT and VMAT treatment planning in pelvic anatomy

PURPOSE The aim of this work was to determine whether a commercial knowledge-based treatment planning (KBP) module can efficiently produce IMRT and VMAT plans in the pelvic region (prostate & cervical cancer), and to assess sensitivity of plan quality to training data and model parameters. METHODS Initial benchmarking of KBP was performed using prostate cancer cases. Structures and dose distributions from 40 patients previously treated using a 5-field IMRT technique were used for model training. Two types of model were created: one excluded statistical outliers (as identified by RapidPlan guidelines) and the other had no exclusions. A separate model for cervix uteri cancer cases was subsequently developed using 37 clinical patients treated for cervical cancer using RapidArc™ VMAT, with no exclusions. The resulting models were then used to generate plans for ten patients from each patient group who had not been included in the modelling process. Comparisons of generated RapidPlans with the corresponding clinical plans were carried out to indicate the required modifications to the models. Model parameters were then iteratively adjusted until plan quality converged with that obtained by experienced planners without KBP. RESULTS Initial automated model generation settings led to poor conformity, coverage and efficiency compared to clinical plans. Therefore a number of changes to the initial KBP models were required. Before model optimisation, it was found that the PTV coverage was slightly reduced in the superior and inferior directions for RapidPlan compared with clinical plans and therefore PTV parameters were adjusted to improve coverage. OAR doses were similar for both RapidPlan and clinical plans (p>0.05). Excluding outliers had little effect on plan quality (p≫0.05). Manually fixing key optimisation objectives enabled production of clinically acceptable treatment plans without further planner intervention for 9 of 10 prostate test patients and all 10 cervix test patients. CONCLUSIONS The Varian RapidPlan™ system was able to produce IMRT & VMAT treatment plans in the pelvis, in a single optimisation, that had comparable sparing and comparable or better conformity than the original clinically acceptable plans. The system allows for better consistency and efficiency in the treatment planning process and has therefore been adopted clinically within our institute with over 100 patients treated.

Mathematical modelling of tumour volume dynamics in response to stereotactic ablative radiotherapy for non-small cell lung cancer

This paper reports a modelling study of tumour volume dynamics in response to stereotactic ablative radiotherapy (SABR). The main objective was to develop a model that is adequate to describe tumour volume change measured during SABR, and at the same time is not excessively complex as lacking support from clinical data. To this end, various modelling options were explored, and a rigorous statistical method, the Akaike information criterion, was used to help determine a trade-off between model accuracy and complexity. The models were calibrated to the data from 11 non-small cell lung cancer patients treated with SABR. The results showed that it is feasible to model the tumour volume dynamics during SABR, opening up the potential for using such models in a clinical environment in the future.

Magnitude of observer error using cone beam CT for prostate interfraction motion estimation: effect of reducing scan length or increasing exposure.

Objective: Cone beam CT (CBCT) enables soft-tissue registration to planning CT for position verification in radiotherapy. The aim of this study was to determine the interobserver error (IOE) in prostate position verification using a standard CBCT protocol, and the effect of reducing CBCT scan length or increasing exposure, compared with standard imaging protocol. Methods: CBCT images were acquired using a novel 7 cm length image with standard exposure (1644 mAs) at Fraction 1 (7), standard 12 cm length image (1644 mAs) at Fraction 2 (12) and a 7 cm length image with higher exposure (2632 mAs) at Fraction 3 (7H) on 31 patients receiving radiotherapy for prostate cancer. Eight observers (two clinicians and six radiographers) registered the images. Guidelines and training were provided. The means of the IOEs were compared using a Kruzkal–Wallis test. Levenes test was used to test for differences in the variances of the IOEs and the independent prostate position. Results: No significant difference was found between the IOEs of each image protocol in any direction. Mean absolute IOE was the greatest in the anteroposterior direction. Standard deviation (SD) of the IOE was the least in the left–right direction for each of the three image protocols. The SD of the IOE was significantly less than the independent prostate motion in the anterior–posterior (AP) direction only (1.8 and 3.0 mm, respectively: p = 0.017). IOEs were within 1 SD of the independent prostate motion in 95%, 77% and 96% of the images in the RL, SI and AP direction. Conclusion: Reducing CBCT scan length and increasing exposure did not have a significant effect on IOEs. To reduce imaging dose, a reduction in CBCT scan length could be considered without increasing the uncertainty in prostate registration. Precision of CBCT verification of prostate radiotherapy is affected by IOE and should be quantified prior to implementation. Advances in knowledge: This study shows the importance of quantifying the magnitude of IOEs prior to CBCT implementation.

Factors influencing the robustness of P-value measurements in CT texture prognosis studies

Several studies have recently reported on the value of CT texture analysis in predicting survival, although the topic remains controversial, with further validation needed in order to consolidate the evidence base. The aim of this study was to investigate the effect of varying the input parameters in the Kaplan-Meier analysis, to determine whether the resulting P-value can be considered to be a robust indicator of the parameters prognostic potential. A retrospective analysis of the CT-based normalised entropy of 51 patients with lung cancer was performed and overall survival data for these patients were collected. A normalised entropy cut-off was chosen to split the patient cohort into two groups and log-rank testing was performed to assess the survival difference of the two groups. This was repeated for varying normalised entropy cut-offs and varying follow-up periods. Our findings were also compared with previously published results to assess robustness of this parameter in a multi-centre patient cohort. The P-value was found to be highly sensitive to the choice of cut-off value, with small changes in cut-off producing substantial changes in P. The P-value was also sensitive to follow-up period, with particularly noisy results at short follow-up periods. Using matched conditions to previously published results, a P-value of 0.162 was obtained. Survival analysis results can be highly sensitive to the choice in texture cut-off value in dichotomising patients, which should be taken into account when performing such studies to avoid reporting false positive results. Short follow-up periods also produce unstable results and should therefore be avoided to ensure the results produced are reproducible. Previously published findings that indicated the prognostic value of normalised entropy were not replicated here, but further studies with larger patient numbers would be required to determine the cause of the different outcomes.

EP-1432: Experiences in model creation using Varian Rapid PlanTM for 5 field IMRT prostate treatments

lymph nodes and a boost volume including the prostate and seminal vesicles; TPs were generated in simultaneous boost technique. For AP, a progressive engine is used where the user defines prioritized optimization goals for PTV-coverage and dose thresholds and priorities for each organ at risk (OAR). The AP engine automatically creates objectives and required optimization aid structures (OAS), and multiple optimization loops iteratively reformulate and adjust the optimization objectives to meet the goals and further lower dose to OAR with minimal compromise to the target coverage. For manual planning, additional OAS have to be generated by the planner, objectives and priorities have to be adjusted manually for each optimization loop. For plan comparison, various dose and dose volume metrics (Dmed, D98%, D2% V95% for target volumes, D2%, Dmed and Vx% for OARs) as well as homogeneity index (HI = (D2%-D98%)/ D50%) and conformity index (CIPaddick = TV2PI/(PI*TV)) were evaluated. Efficiency of the plan optimization procedure was estimated by means of total time required to create a TP. Results: PTV coverage V95% was 93.5±3.5% and 97.9±1.3% and boost coverage was 95.5±2.0% and 98.3±1.7% for MP and AP, respectively. Homogeneity index for the PTV was 0.14±0.02 and 0.12±0.02 and for the boost it was 0.11±0.02 and 0.07±0.02 for MP and AP, respectively. CI was 13% and 16% higher in manual plans compared to automatic plans for PTV and boost, respectively. Dmed and D2% for bladder and femoral heads showed no particular differences between manual and automatic plans. However, considerable deviations in Dmed were found for the rectum (27.8±4.7Gy vs 33.3±5.8Gy for MP and AP, respectively) and intestine (25.2±7.5Gy and 22.8±8.2Gy for MP and AP, respectively). Further, VTissue30% representing tissue outside the target volumes received 36% more dose in AP compared to MP. The time to create a treatment plan was <1 hour for MP and >2 hours for AP. Conclusions: Automatically generated TPs improve target coverage and homogeneity at the cost of slightly decreased conformity when compared to manual TPs. OAR sparing is mostly comparable, higher dose contribution to normal tissue outside the PTV was found for AP. Since higher low dose volume was detected in normal tissue for AP plans, each TP needs to be evaluated by an experienced planner and adapted when necessary. Prioritized optimization goals in AP need to be carefully established and the overall time required to create a plan remains to be optimized.

PD-0272: Can CBCT image quality be improved to reduce inter observer error in patients receiving radiotherapy to the prostate?

position shifts 1.5 cm the CTV coverage drops rapidly for RA isotropic (V95(CTV) from 100% to 83%) whereas the RA anisotropic were more robust (V95(CTV) from 100% to 98%). The over dosage of ORs, e.g. the spinal cord, were in general of no clinical importance. In general, the most preferable method to increase the robustness was to use arcs with avoiding sectors in combination with arcs without avoiding sectors. If all arcs had avoiding sectors this limited the beam angles geometry at the expense of coverage of targets in the upper parts of the head and neck.