Angel Kennedy
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Angel Kennedy has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Martin Andrew Ebert
Department of Radiation Oncology, Sir Charles Gairdner Hospital, Nedlands, Australia
5 shared publications - Jason Dowling
Commonwealth Scientific and Industrial Research Organisation, Australian E-Health Research Centre, Herston, QLD 4029, Australia
3 shared publications - Annette Haworth
Institute of Medical Physics University of Sydney Sydney New South Wales Australia
2 shared publications - David Joseph
Medical School, University of Western Australia, Nedlands, WA, 6009, Australia
2 shared publications - James William Denham
School of Medicine and Public Health, University of Newcastle, Newcastle, New South Wales, Australia
2 shared publications - Lois Holloway
South Western Sydney Clinical School, Faculty of Medicine, UNSW, Sydney, Australia
2 shared publications
Latest Publications
- Open Access
Biologically Targeted Radiation Therapy: Incorporating Patient-Specific Hypoxia Data Derived from Quantitative Magnetic Resonance Imaging
Purpose: Hypoxia has been linked to radioresistance. Strategies to safely dose escalate dominant intraprostatic lesions have shown promising results, but further dose escalation to overcome the effects of hypoxia require a novel approach to constrain the dose in normal tissue.to safe levels. In this study, we demonstrate a biologically targeted radiotherapy (BiRT) approach that can utilise multiparametric magnetic resonance imaging (mpMRI) to target hypoxia for favourable treatment outcomes. Methods: mpMRI-derived tumour biology maps, developed via a radiogenomics study, were used to generate individualised, hypoxia-targeting prostate IMRT plans using an ultra- hypofractionation schedule. The spatial distribution of mpMRI textural features associated with hypoxia-related genetic profiles was used as a surrogate of tumour hypoxia. The effectiveness of the proposed approach was assessed by quantifying the potential benefit of a general focal boost approach on tumour control probability, and also by comparing the dose to organs at risk (OARs) with hypoxia-guided focal dose escalation (DE) plans generated for five patients. Results: Applying an appropriately guided focal boost can greatly mitigate the impact of hypoxia. Statistically significant reductions in rectal and bladder dose were observed for hypoxia-targeting, biologically optimised plans compared to isoeffective focal DE plans. Conclusion: Results of this study suggest the use of mpMRI for voxel-level targeting of hypoxia, along with biological optimisation, can provide a mechanism for guiding focal DE that is considerably more efficient than application of a general, dose-based optimisation, focal boost.
Open AccessBiologically Targeted Radiation Therapy: Incorporating Patient-Specific Hypoxia Data Derived from Quantitative Magnetic Resonance Imaging
Purpose: Hypoxia has been linked to radioresistance. Strategies to safely dose escalate dominant intraprostatic lesions have shown promising results, but further dose escalation to overcome the effects of hypoxia require a novel approach to constrain the dose in normal tissue.to safe levels. In this study, we demonstrate a biologically targeted radiotherapy (BiRT) approach that can utilise multiparametric magnetic resonance imaging (mpMRI) to target hypoxia for favourable treatment outcomes. Methods: mpMRI-derived tumour biology maps, developed via a radiogenomics study, were used to generate individualised, hypoxia-targeting prostate IMRT plans using an ultra- hypofractionation schedule. The spatial distribution of mpMRI textural features associated with hypoxia-related genetic profiles was used as a surrogate of tumour hypoxia. The effectiveness of the proposed approach was assessed by quantifying the potential benefit of a general focal boost approach on tumour control probability, and also by comparing the dose to organs at risk (OARs) with hypoxia-guided focal dose escalation (DE) plans generated for five patients. Results: Applying an appropriately guided focal boost can greatly mitigate the impact of hypoxia. Statistically significant reductions in rectal and bladder dose were observed for hypoxia-targeting, biologically optimised plans compared to isoeffective focal DE plans. Conclusion: Results of this study suggest the use of mpMRI for voxel-level targeting of hypoxia, along with biological optimisation, can provide a mechanism for guiding focal DE that is considerably more efficient than application of a general, dose-based optimisation, focal boost.
- 18 March 2019
- Published by Wiley in Medical Physics
- Vol. 46 (5) , 2243-2250
Similarity clustering‐based atlas selection for pelvic CT image segmentation
Purpose: To demonstrate selection of a small representative subset of images from a pool of images comprising a potential atlas (PA) pelvic CT set to be used for autosegmentation of a separate target image set. The aim is to balance the need for the atlas set to represent anatomical diversity with the need to minimize resources required to create a high quality atlas set (such as multiobserver delineation), while retaining access to additional information available for the PA image set.Methods: Preprocessing was performed for image standardization, followed by image registration. Clustering was used to select the subset that provided the best coverage of a target dataset as measured by postregistration image intensity similarities. Tests for clustering robustness were performed including repeated clustering runs using different starting seeds and clustering repeatedly using 90% of the target dataset chosen randomly. Comparisons of coverage of a target set (comprising 711 pelvic CT images) were made for atlas sets of five images (chosen from a PA set of 39 pelvic CT and MR images) (a) at random (averaged over 50 random atlas selections), (b) based solely on image similarities within the PA set (representing prospective atlas development), (c) based on similarities within the PA set and between the PA and target dataset (representing retrospective atlas development). Comparisons were also made to coverage provided by the entire PA set of 39 images.Results: Exemplar selection was highly robust with exemplar selection results being unaffected by choice of starting seed with very occasional change to one of the exemplar choices when the target set was reduced. Coverage of the target set, as measured by best normalized cross‐correlation similarity of target images to any exemplar image, provided by five well‐selected atlas images (mean = 0.6497) was more similar to coverage provided by the entire PA set (mean = 0.6658) than randomly chosen atlas subsets (mean = 0.5977). This was true both of the mean values and the shape of the distributions. Retrospective selection of atlases (mean = 0.6497) provided a very small improvement over prospective atlas selection (mean = 0.6431). All differences were significant (P < 1.0E‐10).Conclusions: Selection of a small representative image set from one dataset can be utilized to develop an atlas set for either retrospective or prospective autosegmentation of a different target dataset. The coverage provided by such a judiciously selected subset has the potential to facilitate propagation of numerous retrospectively defined structures, utilizing additional information available with multimodal imaging in the atlas set, without the need to create large atlas image sets.
- 18 March 2019
- Published by Wiley in Medical Physics
- Vol. 46 (5) , 2243-2250
Similarity clustering‐based atlas selection for pelvic CT image segmentation
Purpose: To demonstrate selection of a small representative subset of images from a pool of images comprising a potential atlas (PA) pelvic CT set to be used for autosegmentation of a separate target image set. The aim is to balance the need for the atlas set to represent anatomical diversity with the need to minimize resources required to create a high quality atlas set (such as multiobserver delineation), while retaining access to additional information available for the PA image set.Methods: Preprocessing was performed for image standardization, followed by image registration. Clustering was used to select the subset that provided the best coverage of a target dataset as measured by postregistration image intensity similarities. Tests for clustering robustness were performed including repeated clustering runs using different starting seeds and clustering repeatedly using 90% of the target dataset chosen randomly. Comparisons of coverage of a target set (comprising 711 pelvic CT images) were made for atlas sets of five images (chosen from a PA set of 39 pelvic CT and MR images) (a) at random (averaged over 50 random atlas selections), (b) based solely on image similarities within the PA set (representing prospective atlas development), (c) based on similarities within the PA set and between the PA and target dataset (representing retrospective atlas development). Comparisons were also made to coverage provided by the entire PA set of 39 images.Results: Exemplar selection was highly robust with exemplar selection results being unaffected by choice of starting seed with very occasional change to one of the exemplar choices when the target set was reduced. Coverage of the target set, as measured by best normalized cross‐correlation similarity of target images to any exemplar image, provided by five well‐selected atlas images (mean = 0.6497) was more similar to coverage provided by the entire PA set (mean = 0.6658) than randomly chosen atlas subsets (mean = 0.5977). This was true both of the mean values and the shape of the distributions. Retrospective selection of atlases (mean = 0.6497) provided a very small improvement over prospective atlas selection (mean = 0.6431). All differences were significant (P < 1.0E‐10).Conclusions: Selection of a small representative image set from one dataset can be utilized to develop an atlas set for either retrospective or prospective autosegmentation of a different target dataset. The coverage provided by such a judiciously selected subset has the potential to facilitate propagation of numerous retrospectively defined structures, utilizing additional information available with multimodal imaging in the atlas set, without the need to create large atlas image sets.
- 4 January 2019
- Published by Wiley in Journal of Medical Imaging and Radiation Oncology
- Vol. 63 (2) , 264-271
Multi‐observer contouring of male pelvic anatomy: Highly variable agreement across conventional and emerging structures of interest
Introduction: This study quantified inter‐observer contouring variations for multiple male pelvic structures, many of which are of emerging relevance for prostate cancer radiotherapy progression and toxicity response studies.Methods: Five prostate cancer patient datasets (CT and T2‐weighted MR) were distributed to 13 observers for contouring. CT structures contoured included the clinical target volume (CTV), seminal vesicles, rectum, colon, bowel bag, bladder and peri‐rectal space (PRS). MR contours included CTV, trigone, membranous urethra, penile bulb, neurovascular bundle and multiple pelvic floor muscles. Contouring variations were assessed using the intraclass correlation coefficient (ICC), Dice similarity coefficient (DSC), and multiple additional metrics.Results: Clinical target volume (CT and MR), bladder, rectum and PRS contours showed excellent inter‐observer agreement (median ICC = 0.97; 0.99; 1.00; 0.95; 0.90, DSC = 0.83 ± 0.05; 0.88 ± 0.05; 0.93 ± 0.03; 0.81 ± 0.07; 0.80 ± 0.06, respectively). Seminal vesicle contours were more variable (ICC = 0.75, DSC = 0.73 ± 0.14), while colon and bowel bag contoured volumes were consistent (ICC = 0.97; 0.97), but displayed poor overlap (DSC = 0.58 ± 0.22; 0.67 ± 0.21). Smaller MR structures showed significant inter‐observer variations, with poor overlap for trigone, membranous urethra, penile bulb, and left and right neurovascular bundles (DSC = 0.44 ± 0.22; 0.41 ± 0.21; 0.66 ± 0.21; 0.16 ± 0.17; 0.15 ± 0.15). Pelvic floor muscles recorded moderate to strong inter‐observer agreement (ICC = 0.50–0.97), although large outlier variations were observed.Conclusions: Inter‐observer contouring variation was significant for multiple pelvic structures contoured on MR.
- 4 January 2019
- Published by Wiley in Journal of Medical Imaging and Radiation Oncology
- Vol. 63 (2) , 264-271
Multi‐observer contouring of male pelvic anatomy: Highly variable agreement across conventional and emerging structures of interest
Introduction: This study quantified inter‐observer contouring variations for multiple male pelvic structures, many of which are of emerging relevance for prostate cancer radiotherapy progression and toxicity response studies.Methods: Five prostate cancer patient datasets (CT and T2‐weighted MR) were distributed to 13 observers for contouring. CT structures contoured included the clinical target volume (CTV), seminal vesicles, rectum, colon, bowel bag, bladder and peri‐rectal space (PRS). MR contours included CTV, trigone, membranous urethra, penile bulb, neurovascular bundle and multiple pelvic floor muscles. Contouring variations were assessed using the intraclass correlation coefficient (ICC), Dice similarity coefficient (DSC), and multiple additional metrics.Results: Clinical target volume (CT and MR), bladder, rectum and PRS contours showed excellent inter‐observer agreement (median ICC = 0.97; 0.99; 1.00; 0.95; 0.90, DSC = 0.83 ± 0.05; 0.88 ± 0.05; 0.93 ± 0.03; 0.81 ± 0.07; 0.80 ± 0.06, respectively). Seminal vesicle contours were more variable (ICC = 0.75, DSC = 0.73 ± 0.14), while colon and bowel bag contoured volumes were consistent (ICC = 0.97; 0.97), but displayed poor overlap (DSC = 0.58 ± 0.22; 0.67 ± 0.21). Smaller MR structures showed significant inter‐observer variations, with poor overlap for trigone, membranous urethra, penile bulb, and left and right neurovascular bundles (DSC = 0.44 ± 0.22; 0.41 ± 0.21; 0.66 ± 0.21; 0.16 ± 0.17; 0.15 ± 0.15). Pelvic floor muscles recorded moderate to strong inter‐observer agreement (ICC = 0.50–0.97), although large outlier variations were observed.Conclusions: Inter‐observer contouring variation was significant for multiple pelvic structures contoured on MR.
- 7 June 2018
- Published by Springer Nature in Australasian Physics & Engineering Sciences in Medicine
- Vol. 41 (3) , 739-745
Estimation of Hounsfield unit conversion parameters for pelvic CT images
Linear scaling is used to convert raw computed tomography (CT) pixel values into Hounsfield units corresponding to different tissue values. Analysis of a benchmarking study, presented here, where the same CT scan was imported into and then exported from multiple radiotherapy treatment planning systems, found inconsistencies in HU scaling parameter values exported along with the images, particularly when images were exported using the Radiation Therapy Oncology Group format. Several methods of estimating conversion parameters, based on estimating pixel values corresponding to air and water within the image, for pelvic CT images from a large multi-centre trial were compared against original Digital Imaging and Communications in Medicine export parameters. In general using the mean of a sample region at the centroid of the bladder to estimate the value of water was more accurate than using the minimum or maximum or a single value at the centroid. Accuracy of methods of air estimation tested were dependent in part on features of the CT scanners and treatment planning systems, making it difficult to pick one method as superior that was independent of scanner and treatment planning system type. Based on the above analysis, methods for estimating air and water were selected for use in performing linear scaling of a set of pelvic CT images prior to their use in an interpatient image registration application. The selected methods were validated against a more recent and homogeneous dataset. Estimation error was found to be much lower within the validation set.
- 7 June 2018
- Published by Springer Nature in Australasian Physics & Engineering Sciences in Medicine
- Vol. 41 (3) , 739-745
Estimation of Hounsfield unit conversion parameters for pelvic CT images
Linear scaling is used to convert raw computed tomography (CT) pixel values into Hounsfield units corresponding to different tissue values. Analysis of a benchmarking study, presented here, where the same CT scan was imported into and then exported from multiple radiotherapy treatment planning systems, found inconsistencies in HU scaling parameter values exported along with the images, particularly when images were exported using the Radiation Therapy Oncology Group format. Several methods of estimating conversion parameters, based on estimating pixel values corresponding to air and water within the image, for pelvic CT images from a large multi-centre trial were compared against original Digital Imaging and Communications in Medicine export parameters. In general using the mean of a sample region at the centroid of the bladder to estimate the value of water was more accurate than using the minimum or maximum or a single value at the centroid. Accuracy of methods of air estimation tested were dependent in part on features of the CT scanners and treatment planning systems, making it difficult to pick one method as superior that was independent of scanner and treatment planning system type. Based on the above analysis, methods for estimating air and water were selected for use in performing linear scaling of a set of pelvic CT images prior to their use in an interpatient image registration application. The selected methods were validated against a more recent and homogeneous dataset. Estimation error was found to be much lower within the validation set.
- 10 January 2016
- Published by American Society of Clinical Oncology (ASCO) in Journal of Clinical Oncology
- Vol. 34 (2_suppl) , 82
Modeling severe late rectal bleeding: Results on a large pooled population of prostate cancer patients.
82 Background: To develop a model for grade 3 (G3) late rectal bleeding (LRB) after radical radiotherapy (RT) for prostate cancer, in a pooled population from 2 large prospective trials: Airopros0102 (Fellin RO2014) and TROG 03.04 RADAR (Ebert IJROBP2015). Methods: Both trials included patients (pts) treated with 3DCRT at 66-80Gy, conventional fractionation. Planning data were available for all pts, G3 LRB was prospectively scored using the SOMA/LENT, with a minimum follow-up of 3 years. Rectal dose-volume histograms were reduced to Equivalent Uniform Dose (EUD) calculated with volume parameter n derived by 3 studies: n=0.06 (Rancati RO2004), n=0.05 (Rancati RO2011) and n=0.018 (Defraene IJROBP2011). EUD was inserted into multivariable logistic regression (MVL) together with clinical and treatment features. Irradiation of seminal vesicles (SV), irradiation of pelvic nodes, hormonal therapy, hypertension, previous abdominal surgery (SURG), use of anticoagulants, diabetes, cardiovascular diseases and presence of acute toxicity were considered as potential dose-modifying factors. Goodness of fit was evaluated with Hosmer Lemeshow test (HL), calibration through calibration slope and AUC was used for discrimination power. Results: 1,337 pts were available: 708 RADAR and 669 Airopros. G3 LRB was scored in 95 pts (7.1%): 62 RADAR and 33 Airopros. EUD calculated with n=0.05 was the best dosimetric predictor for G3 LRB. A 4-variable MVL model was fitted including EUD (OR=1.07 p=0.16), SV (OR=4.75 p
- 10 January 2016
- Published by American Society of Clinical Oncology (ASCO) in Journal of Clinical Oncology
- Vol. 34 (2_suppl) , 82
Modeling severe late rectal bleeding: Results on a large pooled population of prostate cancer patients.
82 Background: To develop a model for grade 3 (G3) late rectal bleeding (LRB) after radical radiotherapy (RT) for prostate cancer, in a pooled population from 2 large prospective trials: Airopros0102 (Fellin RO2014) and TROG 03.04 RADAR (Ebert IJROBP2015). Methods: Both trials included patients (pts) treated with 3DCRT at 66-80Gy, conventional fractionation. Planning data were available for all pts, G3 LRB was prospectively scored using the SOMA/LENT, with a minimum follow-up of 3 years. Rectal dose-volume histograms were reduced to Equivalent Uniform Dose (EUD) calculated with volume parameter n derived by 3 studies: n=0.06 (Rancati RO2004), n=0.05 (Rancati RO2011) and n=0.018 (Defraene IJROBP2011). EUD was inserted into multivariable logistic regression (MVL) together with clinical and treatment features. Irradiation of seminal vesicles (SV), irradiation of pelvic nodes, hormonal therapy, hypertension, previous abdominal surgery (SURG), use of anticoagulants, diabetes, cardiovascular diseases and presence of acute toxicity were considered as potential dose-modifying factors. Goodness of fit was evaluated with Hosmer Lemeshow test (HL), calibration through calibration slope and AUC was used for discrimination power. Results: 1,337 pts were available: 708 RADAR and 669 Airopros. G3 LRB was scored in 95 pts (7.1%): 62 RADAR and 33 Airopros. EUD calculated with n=0.05 was the best dosimetric predictor for G3 LRB. A 4-variable MVL model was fitted including EUD (OR=1.07 p=0.16), SV (OR=4.75 p
- Open Access
- 13 December 2014
- Published by Springer Nature in Radiation Oncology
- Vol. 9 (1) , 282
Impact of treatment planning and delivery factors on gastrointestinal toxicity: an analysis of data from the RADAR prostate radiotherapy trial
To assess the impact of incremental modifications of treatment planning and delivery technique, as well as patient anatomical factors, on late gastrointestinal toxicity using data from the TROG 03.04 RADAR prostate radiotherapy trial. The RADAR trial accrued 813 external beam radiotherapy participants during 2003–2008 from 23 centres. Following review and archive to a query-able database, digital treatment plans and data describing treatment technique for 754 patients were available for analysis. Treatment demographics, together with anatomical features, were assessed using uni- and multivariate regression models against late gastrointestinal toxicity at 18-, 36- and 54-month follow-up. Regression analyses were reviewed in the context of dose-volume data for the rectum and anal canal. A multivariate analysis at 36-month follow-up shows that patients planned using a more rigorous dose calculation algorithm (DCA) was associated with a lower risk of stool frequency (OR: 0.435, CI: 0.242–0.783, corrected p = 0.04). Patients using laxative as a method of bowel preparation had higher risk of having increased stool frequency compared to patients with no dietary intervention (OR: 3.639, CI: 1.502–8.818, corrected p = 0.04). Despite higher risks of toxicities, the anorectum, anal canal and rectum dose-volume histograms (DVH) indicate patients using laxative had unremarkably different planned dose distributions. Patients planned with a more rigorous DCA had lower median DVH values between EQD23 = 15 Gy and EQD23 = 35 Gy. Planning target volume (PTV), conformity index, rectal width and prescription dose were not significant when adjusted for false discovery rate. Number of beams, beam energy, treatment beam definition, positioning orientation, rectum-PTV separation, rectal length and mean cross sectional area did not affect the risk of toxicities. The RADAR study dataset has allowed an assessment of technical modifications on gastrointestinal toxicity. A number of interesting associations were subsequently found and some factors, previously hypothesised to influence toxicity, did not demonstrate any significant impact. We recommend trial registries be encouraged to record technical modifications introduced during the trial in order for more powerful evidence to be gathered regarding the impact of the interventions.
- 13 December 2014
- Published by Springer Nature in Radiation Oncology
- Vol. 9 (1) , 282
Open AccessImpact of treatment planning and delivery factors on gastrointestinal toxicity: an analysis of data from the RADAR prostate radiotherapy trial
To assess the impact of incremental modifications of treatment planning and delivery technique, as well as patient anatomical factors, on late gastrointestinal toxicity using data from the TROG 03.04 RADAR prostate radiotherapy trial. The RADAR trial accrued 813 external beam radiotherapy participants during 2003–2008 from 23 centres. Following review and archive to a query-able database, digital treatment plans and data describing treatment technique for 754 patients were available for analysis. Treatment demographics, together with anatomical features, were assessed using uni- and multivariate regression models against late gastrointestinal toxicity at 18-, 36- and 54-month follow-up. Regression analyses were reviewed in the context of dose-volume data for the rectum and anal canal. A multivariate analysis at 36-month follow-up shows that patients planned using a more rigorous dose calculation algorithm (DCA) was associated with a lower risk of stool frequency (OR: 0.435, CI: 0.242–0.783, corrected p = 0.04). Patients using laxative as a method of bowel preparation had higher risk of having increased stool frequency compared to patients with no dietary intervention (OR: 3.639, CI: 1.502–8.818, corrected p = 0.04). Despite higher risks of toxicities, the anorectum, anal canal and rectum dose-volume histograms (DVH) indicate patients using laxative had unremarkably different planned dose distributions. Patients planned with a more rigorous DCA had lower median DVH values between EQD23 = 15 Gy and EQD23 = 35 Gy. Planning target volume (PTV), conformity index, rectal width and prescription dose were not significant when adjusted for false discovery rate. Number of beams, beam energy, treatment beam definition, positioning orientation, rectum-PTV separation, rectal length and mean cross sectional area did not affect the risk of toxicities. The RADAR study dataset has allowed an assessment of technical modifications on gastrointestinal toxicity. A number of interesting associations were subsequently found and some factors, previously hypothesised to influence toxicity, did not demonstrate any significant impact. We recommend trial registries be encouraged to record technical modifications introduced during the trial in order for more powerful evidence to be gathered regarding the impact of the interventions.