Paul Doolan

Dr. Paul Doolan

Medical Physics, German Oncology Center, Limassol, Lemesós, Cyprus
orcid

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 August 2023 in Frontiers in Oncology
Verified
Journal Article
Open access
Paul DoolanPaul Doolan
Paul Doolan
SCStefanie Charalambous
Stefanie Charalambous
YRYiannis Roussakis
Yiannis Roussakis
ALAgnes Leczynski
Agnes Leczynski
MPMary Peratikou
Mary Peratikou
EKEfstratios Karagiannis
Efstratios Karagiannis
Purpose/objective(s): Auto-segmentation with artificial intelligence (AI) offers an opportunity to reduce inter- and intra-observer variability in contouring, to improve the quality of contours, as well as to reduce the time taken to conduct this manual task. In this work we benchmark the AI auto-segmentation contours produced by five commercial vendors against a common dataset.Methods and materials: The organ at risk (OAR) contours generated by five commercial AI auto-segmentation solutions (Mirada (Mir), MVision (MV), Radformation (Rad), RayStation (Ray) and TheraPanacea (Ther)) were compared to manually-drawn expert contours from 20 breast, 20 head and neck, 20 lung and 20 prostate patients. Comparisons were made using geometric similarity metrics including volumetric and surface Dice similarity coefficient (vDSC and sDSC), Hausdorff distance (HD) and Added Path Length (APL). To assess the time saved, the time taken to manually draw the expert contours, as well as the time to correct the AI contours, were recorded.Results: There are differences in the number of CT contours offered by each AI auto-segmentation solution at the time of the study (Mir 99; MV 143; Rad 83; Ray 67; Ther 86), with all offering contours of some lymph node levels as well as OARs. Averaged across all structures, the median vDSCs were good for all systems and compared favorably with existing literature: Mir 0.82; MV 0.88; Rad 0.86; Ray 0.87; Ther 0.88. All systems offer substantial time savings, ranging between: breast 14-20 mins; head and neck 74-93 mins; lung 20-26 mins; prostate 35-42 mins. The time saved, averaged across all structures, was similar for all systems: Mir 39.8 mins; MV 43.6 mins; Rad 36.6 min; Ray 43.2 mins; Ther 45.2 mins.Conclusions: All five commercial AI auto-segmentation solutions evaluated in this work offer high quality contours in significantly reduced time compared to manual contouring, and could be used to render the radiotherapy workflow more efficient and standardized.

10 February 2015 in Physics in Medicine & Biology
Verified
Journal Article
Open access
Paul DoolanPaul Doolan
Paul Doolan
MTM Testa
M Testa
GSG Sharp
G Sharp
EBE H Bentefour
E H Bentefour
GRGary Royle
Gary Royle
HLH-M Lu
H-M Lu
A simple robust optimizer has been developed that can produce patient-specific calibration curves to convert x-ray computed tomography (CT) numbers to relative stopping powers (HU-RSPs) for proton therapy treatment planning. The difference between a digitally reconstructed radiograph water-equivalent path length (DRRWEPL) map through the x-ray CT dataset and a proton radiograph (set as the ground truth) is minimized by optimizing the HU-RSP calibration curve. The function of the optimizer is validated with synthetic datasets that contain no noise and its robustness is shown against CT noise. Application of the procedure is then demonstrated on a plastic and a real tissue phantom, with proton radiographs produced using a single detector. The mean errors using generic/optimized calibration curves between the DRRWEPL map and the proton radiograph were 1.8/0.4% for a plastic phantom and -2.1/ - 0.2% for a real tissue phantom. It was then demonstrated that these optimized calibration curves offer a better prediction of the water equivalent path length at a therapeutic depth. We believe that these promising results are suggestive that a single proton radiograph could be used to generate a patient-specific calibration curve as part of the current proton treatment planning workflow.

10 June 2015 in Physics in Medicine & Biology
Verified
Journal Article
Closed access
PDF file
CFCharles-Antoine Collins Fekete
Charles-Antoine Collins Fekete
Paul DoolanPaul Doolan
Paul Doolan
MDMarta F Dias
Marta F Dias
LBLuc Beaulieu
Luc Beaulieu
Joao SecoJoao Seco
Joao Seco
To develop an accurate phenomenological model of the cubic spline path estimate of the proton path, accounting for the initial proton energy and water equivalent thickness (WET) traversed.Monte Carlo (MC) simulations were used to calculate the path of protons crossing various WET (10–30 cm) of different material (LN300, water and CB2-50% CaCO3) for a range of initial energies (180–330 MeV). For each MC trajectory, cubic spline trajectories (CST) were constructed based on the entrance and exit information of the protons and compared with the MC using the root mean square (RMS) metric. The CST path is dependent on the direction vector magnitudes (|P0,1|). First, |P0,1| is set to the proton path length (with factor = 1.0). Then, two optimal factor are introduced in |P0,1|. The factors are varied to minimize the RMS difference with MC paths for every configuration. A set of factors, function of WET/water equivalent path length (WEPL), that minimizes the RMS are presented. MTF analysis is then performed on proton radiographs of a line-pair phantom reconstructed using the CST trajectories. was fitted to the WET/WEPL ratio using a quadratic function (Y = A + BX2 where A = 1.01,0.99, B = 0.43,− 0.46 respectively for , ). The RMS deviation calculated along the path, between the CST and the MC, increases with the WET. The increase is larger when using than (difference of 5.0% with WET/WEPL = 0.66). For 230/330 MeV protons, the MTF10% was found to increase by 40/16% respectively for a thin phantom (15 cm) when using the model compared to the model. Calculation times for are scaled down compared to MLP and RMS deviation are similar within standard deviation.Based on the results of this study, using CST with the factors reduces the RMS deviation and increases the spatial resolution when reconstructing proton trajectories.

9 February 2017 in Physics in Medicine & Biology
Verified
Journal Article
Closed access
KJK-W Jee
K-W Jee
RZR Zhang
R Zhang
EBE H Bentefour
E H Bentefour
Paul DoolanPaul Doolan
Paul Doolan
ECE Cascio
E Cascio
HLH-M Lu
H-M Lu
Proton beam therapy benefits from the Bragg peak and delivers highly conformal dose distributions. However, the location of the end-of-range is subject to uncertainties related to the accuracy of the relative proton stopping power estimates and thereby the water-equivalent path length (WEPL) along the beam. To remedy the range uncertainty, an in vivo measurement of the WEPL through the patient, i.e. a proton-range radiograph, is highly desirable. Towards that goal, we have explored a novel method of proton radiography based on the time-resolved dose measured by a flat panel imager (FPI). A 226 MeV pencil beam and a custom-designed range modulator wheel (MW) were used to create a time-varying broad beam. The proton imaging technique used exploits this time dependency by looking at the dose rate at the imager as a function of time. This dose rate function (DRF) has a unique time-varying dose pattern at each depth of penetration. A relatively slow rotation of the MW (0.2 revolutions per second) and a fast image acquisition (30 frames per second, ~33 ms sampling) provided a sufficient temporal resolution for each DRF. Along with the high output of the CsI:Tl scintillator, imaging with pixel binning (2 × 2) generated high signal-to-noise data at a very low radiation dose (~0.1 cGy). Proton radiographs of a head phantom and a Gammex CT calibration phantom were taken with various configurations. The results of the phantom measurements show that the FPI can generate low noise and high spatial resolution proton radiographs. The WEPL values of the CT tissue surrogate inserts show that the measured relative stopping powers are accurate to ~2%. The panel did not show any noticeable radiation damage after the accumulative dose of approximately 3831 cGy. In summary, we have successfully demonstrated a highly practical method of generating proton radiography using an x-ray flat panel imager.

30 June 2023 in Radiotherapy and Oncology
Verified
Journal Article
Open access
JHJulius C. Holzschuh
Julius C. Holzschuh
MMMichael Mix
Michael Mix
JRJuri Ruf
Juri Ruf
THTobias Hölscher
Tobias Hölscher
JKJörg Kotzerke
Jörg Kotzerke
CZConstantinos Zamboglou
Constantinos Zamboglou
With the increased use of focal radiation dose escalation for primary prostate cancer (PCa), accurate delineation of gross tumor volume (GTV) in prostate-specific membrane antigen PET (PSMA-PET) becomes crucial. Manual approaches are time-consuming and observer dependent. The purpose of this study was to create a deep learning model for the accurate delineation of the intraprostatic GTV in PSMA-PET. A 3D U-Net was trained on 128 different 18F-PSMA-1007 PET images from three different institutions. Testing was done on 52 patients including one independent internal cohort (Freiburg: n = 19) and three independent external cohorts (Dresden: n = 14 18F-PSMA-1007, Boston: Massachusetts General Hospital (MGH): n = 9 18F-DCFPyL-PSMA and Dana-Farber Cancer Institute (DFCI): n = 10 68Ga-PSMA-11). Expert contours were generated in consensus using a validated technique. CNN predictions were compared to expert contours using Dice similarity coefficient (DSC). Co-registered whole-mount histology was used for the internal testing cohort to assess sensitivity/specificity. Median DSCs were Freiburg: 0.82 (IQR: 0.73-0.88), Dresden: 0.71 (IQR: 0.53-0.75), MGH: 0.80 (IQR: 0.64-0.83) and DFCI: 0.80 (IQR: 0.67-0.84), respectively. Median sensitivity for CNN and expert contours were 0.88 (IQR: 0.68-0.97) and 0.85 (IQR: 0.75-0.88) (p = 0.40), respectively. GTV volumes did not differ significantly (p > 0.1 for all comparisons). Median specificity of 0.83 (IQR: 0.57-0.97) and 0.88 (IQR: 0.69-0.98) were observed for CNN and expert contours (p = 0.014), respectively. CNN prediction took 3.81 seconds on average per patient. The CNN was trained and tested on internal and external datasets as well as histopathology reference, achieving a fast GTV segmentation for three PSMA-PET tracers with high diagnostic accuracy comparable to manual experts.

1 October 2020 in Cancer Management and Research
Verified
Journal Article
Open access
PDF file
AHAndria Hadjipanteli
Andria Hadjipanteli
Paul DoolanPaul Doolan
Paul Doolan
EKEfthyvoulos Kyriacou
Efthyvoulos Kyriacou
ACAnastasia Constantinidou
Anastasia Constantinidou
Purpose: Breast cancer brain metastasis (BCBM) represents a major clinical challenge. Can MRI help in advancements in the management of BCBM? This review discusses MRI developments and the corresponding potential advancements in BCBM management. Methods: An exhaustive literature search was undertaken to identify studies which look into the potential of MRI in BCBM management. Seven hundred and eighty-four studies published from September 1984 to May 2020 were identified. Three topics are covered where MRI is not clinically established yet: 1) the prognosis of BCBM; 2) the screening of BC patients for BCBM development, and 3) the assessment of imaging features correlated to BC subtype. Results: Thirty-six studies were considered eligible for the purposes of this review. On-going progress is made with the identification of different BCBM characteristics and MRI metrics that might be related to prognosis. Progress has been made with the identification of different BCBM characteristics, including BCBM location, degree of edema, white matter disruption, tumor edge sharpness, and temporal muscle thickness. A more accurate prediction of prognosis could lead to more suitable patient management and treatment. The use of MRI in BCBM screening of the high-risk breast cancer population remains a controversial subject. To date, there are no results from clinical trials; however, there is a rising number of relatively small studies that show concern on this subject and support BCBM screening. It is important to oncologists to be able to assess the tumor subtype non-invasively. MRI features, which have shown some correlation with subtype, include the number of tumors, location, and their distribution in the brain. Advanced tools and metrics have been produced to carry out radiological characteristics analysis on MRI images. Assessing MRI features in more detail could provide a more personalized management of patients. Conclusion: Developments in the use of MRI have the potential to improve BCBM management.