- Email: [email protected]
25 The Use of Phase Sequence Image Sets to Reconstruct the Total Volume Occupied by a Mobile Lung Tumor
$8
September 7-10
Discussion: The incidence of patient-reported late GI and GU toxicity following high-dose radiotherapy at PMH is low and compares very favourably with other reported outcomes. We conclude that dose escalated RT using conformal techniques and image guidance is associated with minimal risk of serious late complications, 24 Quantitative Approaches to Patient Safety: Research in Risk Analysis and Resource Management as Applied to Radiotherapy K-L. Kelly 1, R. Lee2, C. Newcomb ~, D. Cooke 3, E. Ekaette 4, P. Craighead ~, P. Dunscombe ~ Tom Baker Cancer Center, University of Calgary, Calgary, Alberta1; University of Calgary Faculty of Medicine, Calgary Health Technology Implementation Unit and Institute of Health Economics, Calgary, Alberta2; University of Calgary Haskayne School of Business, Calgary, Alberta3; University of Calgary Faculty of Medicine, Calgary, Alberta 4
Background: Patient safety is a pressing issue in radiotherapy. There is the potential for inappropriate administration of radiation to patients, with potentially catastrophic consequences. Considerable economic and human resources are expended in quality assurance/quality control activities to prevent such incidents. Presently, the distribution of these resources is based on historical evolution over many years rather than on a systematic and quantitative study of possible incidents in the total radiotherapy process. Objective: To construct a computational risk and decision analysis model to guide the allocation of limited resources within a radiotherapy program so as to minimize the probability and severity of radiation misadministration. Methods: A quantitative probabilistic risk and decision analysis model was developed to evaluate failures in a radiation therapy delivery system. The model represents the radiation oncology process in stages from initial assessment through to follow up. Possible points of failure, whether hardware, software, or human related, were identified at each step. Literature based probabilities were determined for each of the failure points, and the uncertain adverse consequences in terms of impact to patient quality of life were assessed. This information was put into a decision model to allow evaluation and comparison of potential changes to the system. Results: Expenditure of resources to prevent radiation misadministration while providing high quality patient therapy has not been fully optimized. As a result, our department has recently implemented an incident learning system. This system is designed to act as a continuous improvement process, potentially leading to improved patient safety and more reliable health care performance over time. Conclusion: Healthcare providers and organizations need proactive, systematic risk management strategies that incorporate quantitative risk analysis into their decisionmaking. 25 The Use of Phase Sequence l m a g e Sets to Reconstruct the Total Volume Occupied ITy a Mobile Lung Tumor W. Roa1, A. Alexander z, I. Gagne ], D. Robinson 1, R. Halperin2 Cross Cancer Institute, University of Alberta, Edmonton, Albertal ; Kelowna Cancer Centre, University of British Columbia, Kelowna, British Columbia2
Background: Modern radiotherapy relies heavily upon CT information acquired by sampling the continuous and often mobile anatomy of patients. A conventional approach to account for planning and treatment uncertainties in lung radiotherapy is to construct a planning target volume (PTV) by adding 'population-based' safety margins to the gross tumor volume (GTV) defined from a single free-breathing CT scan.
CARO 2005
The inadequacy of this approach in lung radiotherapy has ted many investigators to develop new methodologies to deal with respiratory tumor motion. The use of phase sequence image (PSI) sets to reveal the total volume occupied by a mobile target is presented. Methods: Isocontrast composite clinical target volumes (CC]~) are constructed from PSI sets to reveal the total volume occupied by a mobile target during its course of travel. The ability of the CCTV technique to properly account for target motion is demonstrated by comparison to contours of the true total volume occupied (TVO) for a number of experimental phantom geometries. Finally, using real patient data, the clinical utility of the CCTV technique to property account for internal tumor motion while minimizing the volume of healthy lung tissue irradiated is assessed by comparison to the standard approach of applying safety margins. Results: The phantom study reveals that CCTV cross-sections constructed at 20% isocontrast level yield good agreement wit the total cross-sections (TXO) of mobile targets. These CCTVs conform well to the TVOs of moving targets examined whereby the addition of small uniform margins ensures complete circumscription of the TVO with the inclusion of minimal amounts of surrounding external volumes. Margins required with the CCTV technique are 8-10 times smaller than those required with individual CTVs. Conclusions: The CCTV technique appears superior to the common practice of adding safety margins to individual CTV contours in order to account for internal target motion. 26 Comparison of Voice Recognition Software with Human Transcription in the Radiation Oncology Clinic S. Pate/I, D. G/IF Department of Radiation Oncology, McGill University Health Centre, Montreal, Quebec~; Department of Radiology, McGill University Health Centre, Montreal, Quebec 2
Objective: This prospective study compares the quality of voice recognition software (VRS) and human transcription in an academic centre. Methods: Twenty randomly selected clinical reports were dictated in each of three testing phases by a single investigator. Phase I consisted of dictation with a microcassette recorder and transcription by experienced clerical personnel. Phase II involved dictation with a noise-canceling headset and VRS (Dragon Naturally Speaking Medical 7.3). In phase III, dictations were completed as in phase II; however, errors observed on-screen were simultaneously corrected using voice commands and the computer keyboard at the time of dictation. Results: Reports consisted of a mean of 423 words. The mean dictation time in Phase I was 266s. Transcription accuracy was 99.2% (95% confidence-interval [CI], 98.6-99.7) and the mean time to correct errors was 47s per transcription. The mean dictation time was similar in phase II (251s; p=0.130). Accuracy, however, was reduced to 94.6% (95% CI, 93.296.0%; p=0.002) and the mean correction time increased to 163s (p=0.006). In comparison to phase II, phase III required increased dictation time (mean, 342s; p=0.036) due to simultaneous correction of errors. However, the resulting transcription had 80% fewer errors requiring a mean of 59s to correct. An accuracy rate of 98.9% (95% CI, 98.4-99.4%) was achieved in phase III, which is similar to Phase I (p=0.450), The mean overall physician time required in phases II and III (414 and 400s, respectively) increased by > 2 5 % compared to phase I (313s, p<0.05). Conclusions: Accuracy of VRS is optimized by modifying dictation technique to include correction of observed on-screen errors while dictating, This approach results in accuracy similar to human transcription. However, VRS requires increased overall physician time. Site-specific cost analysis of physician and clerical time is therefore an important factor in the choice of a transcription system.
September 7-10
Discussion: The incidence of patient-reported late GI and GU toxicity following high-dose radiotherapy at PMH is low and compares very favourably with other reported outcomes. We conclude that dose escalated RT using conformal techniques and image guidance is associated with minimal risk of serious late complications, 24 Quantitative Approaches to Patient Safety: Research in Risk Analysis and Resource Management as Applied to Radiotherapy K-L. Kelly 1, R. Lee2, C. Newcomb ~, D. Cooke 3, E. Ekaette 4, P. Craighead ~, P. Dunscombe ~ Tom Baker Cancer Center, University of Calgary, Calgary, Alberta1; University of Calgary Faculty of Medicine, Calgary Health Technology Implementation Unit and Institute of Health Economics, Calgary, Alberta2; University of Calgary Haskayne School of Business, Calgary, Alberta3; University of Calgary Faculty of Medicine, Calgary, Alberta 4
Background: Patient safety is a pressing issue in radiotherapy. There is the potential for inappropriate administration of radiation to patients, with potentially catastrophic consequences. Considerable economic and human resources are expended in quality assurance/quality control activities to prevent such incidents. Presently, the distribution of these resources is based on historical evolution over many years rather than on a systematic and quantitative study of possible incidents in the total radiotherapy process. Objective: To construct a computational risk and decision analysis model to guide the allocation of limited resources within a radiotherapy program so as to minimize the probability and severity of radiation misadministration. Methods: A quantitative probabilistic risk and decision analysis model was developed to evaluate failures in a radiation therapy delivery system. The model represents the radiation oncology process in stages from initial assessment through to follow up. Possible points of failure, whether hardware, software, or human related, were identified at each step. Literature based probabilities were determined for each of the failure points, and the uncertain adverse consequences in terms of impact to patient quality of life were assessed. This information was put into a decision model to allow evaluation and comparison of potential changes to the system. Results: Expenditure of resources to prevent radiation misadministration while providing high quality patient therapy has not been fully optimized. As a result, our department has recently implemented an incident learning system. This system is designed to act as a continuous improvement process, potentially leading to improved patient safety and more reliable health care performance over time. Conclusion: Healthcare providers and organizations need proactive, systematic risk management strategies that incorporate quantitative risk analysis into their decisionmaking. 25 The Use of Phase Sequence l m a g e Sets to Reconstruct the Total Volume Occupied ITy a Mobile Lung Tumor W. Roa1, A. Alexander z, I. Gagne ], D. Robinson 1, R. Halperin2 Cross Cancer Institute, University of Alberta, Edmonton, Albertal ; Kelowna Cancer Centre, University of British Columbia, Kelowna, British Columbia2
Background: Modern radiotherapy relies heavily upon CT information acquired by sampling the continuous and often mobile anatomy of patients. A conventional approach to account for planning and treatment uncertainties in lung radiotherapy is to construct a planning target volume (PTV) by adding 'population-based' safety margins to the gross tumor volume (GTV) defined from a single free-breathing CT scan.
CARO 2005
The inadequacy of this approach in lung radiotherapy has ted many investigators to develop new methodologies to deal with respiratory tumor motion. The use of phase sequence image (PSI) sets to reveal the total volume occupied by a mobile target is presented. Methods: Isocontrast composite clinical target volumes (CC]~) are constructed from PSI sets to reveal the total volume occupied by a mobile target during its course of travel. The ability of the CCTV technique to properly account for target motion is demonstrated by comparison to contours of the true total volume occupied (TVO) for a number of experimental phantom geometries. Finally, using real patient data, the clinical utility of the CCTV technique to property account for internal tumor motion while minimizing the volume of healthy lung tissue irradiated is assessed by comparison to the standard approach of applying safety margins. Results: The phantom study reveals that CCTV cross-sections constructed at 20% isocontrast level yield good agreement wit the total cross-sections (TXO) of mobile targets. These CCTVs conform well to the TVOs of moving targets examined whereby the addition of small uniform margins ensures complete circumscription of the TVO with the inclusion of minimal amounts of surrounding external volumes. Margins required with the CCTV technique are 8-10 times smaller than those required with individual CTVs. Conclusions: The CCTV technique appears superior to the common practice of adding safety margins to individual CTV contours in order to account for internal target motion. 26 Comparison of Voice Recognition Software with Human Transcription in the Radiation Oncology Clinic S. Pate/I, D. G/IF Department of Radiation Oncology, McGill University Health Centre, Montreal, Quebec~; Department of Radiology, McGill University Health Centre, Montreal, Quebec 2
Objective: This prospective study compares the quality of voice recognition software (VRS) and human transcription in an academic centre. Methods: Twenty randomly selected clinical reports were dictated in each of three testing phases by a single investigator. Phase I consisted of dictation with a microcassette recorder and transcription by experienced clerical personnel. Phase II involved dictation with a noise-canceling headset and VRS (Dragon Naturally Speaking Medical 7.3). In phase III, dictations were completed as in phase II; however, errors observed on-screen were simultaneously corrected using voice commands and the computer keyboard at the time of dictation. Results: Reports consisted of a mean of 423 words. The mean dictation time in Phase I was 266s. Transcription accuracy was 99.2% (95% confidence-interval [CI], 98.6-99.7) and the mean time to correct errors was 47s per transcription. The mean dictation time was similar in phase II (251s; p=0.130). Accuracy, however, was reduced to 94.6% (95% CI, 93.296.0%; p=0.002) and the mean correction time increased to 163s (p=0.006). In comparison to phase II, phase III required increased dictation time (mean, 342s; p=0.036) due to simultaneous correction of errors. However, the resulting transcription had 80% fewer errors requiring a mean of 59s to correct. An accuracy rate of 98.9% (95% CI, 98.4-99.4%) was achieved in phase III, which is similar to Phase I (p=0.450), The mean overall physician time required in phases II and III (414 and 400s, respectively) increased by > 2 5 % compared to phase I (313s, p<0.05). Conclusions: Accuracy of VRS is optimized by modifying dictation technique to include correction of observed on-screen errors while dictating, This approach results in accuracy similar to human transcription. However, VRS requires increased overall physician time. Site-specific cost analysis of physician and clerical time is therefore an important factor in the choice of a transcription system.