Institutions
The presentation includes several public hospitals of 2ndHealthcare Region of Istanbul Province (a.k.a. Anatolia North).
This region is constituted of 13 hospitals and 6 dental hospitals/centers.
In these institutions more than 3 million residents are being served; 11.1 million of outpatient admissions and 250 thousand surgical operations are being performed annually.
Formerly operated,
planned and managed independently,
radiology clinics of these hospitals are currently functioning as units of greater Anatolia North Radiology (ANR).ANR is the largest imaging conglomerate of Turkey,
public and private.
The annual number of radiologic procedures in ANR are more than 4 milion and roughly represents 1/1.000th of world radiological services.
Anatolia North Radiology was specifically designed to provide value-based,
high quality,
effective and efficient imaging services.
On that line,
it employs a comprehensive dose management system powered by dose tracking software.
This system,
mainly includes four hospitals,
all being classified as EuroSafe Imaging Star (*****).
These hospitals were specifically selected to represent a wider sampling of Istanbul Province,
and to make meaningful comparisons possible.
They include a secondary public (Beykoz State Hospital),
a tertiary public (Haydarpasa Numune Training and Research Hospital),
a pediatric tertiary public (Umraniye Children's Hospital),
and a university (Medeniyet University Goztepe Training and Research Hospital) hospital.Figures in the presentation also includes data of Fatih Sultan Mehmet Training and Research Hospital (a.k.a. FSM) ( Fig. 1 ).
FSM is the reference center of ANR,
and was the first institution in Turkey,
where a comprehensive dose management system was implemented and modern dose surveillance IT applications were used. FSM was designed to comply with highest standards and provides a framework to assist other centers develop the required dose management infrastructures according to relevant IAEA guidelines (7).
Dose Management Strategy
Methods used to implement the diagnostic reference levels,
to inform and train the medical staff were initially a part of a larger pan-European project (i.e. Affidea Dose Excellence Campaign) to implement a uniform dose management strategy in different centers with different machines across entirely different countries.
The project’s goals were as follows:
- Transform dose awareness from an act to a habit
- Educate personnel
- Justify high dose examinations
- Standardize protocols and practices
- Optimize protocols and practices
- Communicate message to stakeholders
- Promote best practices and be safe
Standardization
One of the most important steps in dose management is to map local imaging procedures with standard procedures to facilitate data analysis.
So called standardization, this step also includes the asignement to site protocols a specific ID.
The lexion used for this task was the RadLex® Playbook version 1.2.
The process has resulted standartization of 65 protocolsthat would to be used in all participating centers.
The use of nonstandartized protocols are not permitted,
unless clearly justified and for only exceptionally.
Standardization allowed to compare protocols / technologies / radiographers,
and assisted in tracking of examinations and in comparison between same protocols of same CT systems.
Development and implementation of standardized imaging has also shown to lead to better results by means of quality imaging and high patient throughput.
Dose Data Acquisition and Tracking
For current dose levels,
data set was from 01.01.2018 to 17.12.2018.
In that period 369.050 individual standartized sequences for adults and 79.607 individual standartized sequences for children covering wide range of clinical indications were available to analysis.
The majority of the sequences were acquired with 128-slices scanners that had statistical iterative reconstruction algorithms,
and we have calculated DRLs values for these scanners before successful implementation of dose management sytsem and subsequent dose optimization.
Sixteen slices scanners were incorporated into the study at a later phase,
therefore,
we don’t have preoptimization values for them.
For the purpose of dose data acquisition and analysis,
a commercial software (Dosewatch,
GEHC) was used.
This web-based solution captures,tracks and reports radiation dose directly from the medical devices.
It monitors and analyzes high dose alerts and patient cumulative dose,
includes quality metrics to assess the technical factors,
has data consolidation and statistical analysis tools for protocol optimization and dose reduction by optimizing dose levels.
All patients were measured for length and weight before the study.
For pediatric patients Broselow-Luten System for color coded protocols was used where appropriate (8).
This procedure was necessary to calculate size specific dose estimates (SSDE,
see below) and to perform size-based CT dose optimization (9).
Conventional dose data (CTDIvol and Dose Length Product) were recorded or calculated. Basic dose statistics based on protocol (DLP,
minimum,
P25,
mean,
P75,
maximum), were automatically determined using all available patients. Detailed dose information such as SSDE,
effective dose,
and detailed protocol parameters were also recorded or calculated.
Off isocenter shift to identify how the patient was positioned in the bore of the CT and mA modulation to visually observe how dose was optimized along the patient scan length were also presented graphically and observed to determine the optimality of technical parameters.
Justification
If the DLP or CTDIvol for a scan were equal or more than [median x 2],
an alert was triggered.
Alert thresholds were moving targets; so as technicians got better,
dose thresholds became tighter.
Cases for whom overdose alert was triggered had to be justified on the dose traccking software by responsible technician to assist finding root causes of overdoses.
These justifications might point to patient-related (e.g. obesity, incooperative movements),
technician related (e.g.
choosing wrong protocol,
por iso-centering),
or procedure related (e.g.
difficult procedure,
over-length scanning) factors.
Optimization
Optimization calculations took only the mapped protocols into account. We have applied P(75) comparison versus unified pan-European DRLs of Affidea to set the best protocols.
Unified DRLs were determined afer median DRLs published from several European countries (Switzerland,
Belgium,
Finland,
Norway,
France,
Germany,
Poland,
UK) from 2010 onwards (2).
Analyses
The current paper simplifies the study,
and aims to provide the existing CT dose levels for several basic and most used protocols (i.e. Abdomen/pelvis follow up max 1 phase,
RPID 860;; General head helical,
RPID 213; Lumbar spine herniation helical,
RPID 1527; Renal stone,
RPID 343; Chest follow-up,
RPID 16).
Current dose levels of the last calendar year were compared between hospitals and between type of scanners (128 vs 16 slices CTs).
Dose levels before (01.05.2015 to 01.08.2015) and after optimization (01.01.2018 to 17.12.2018 ) were also presented.
Before and after optimization comparisons with unified pan-European tresholds were made.
Dose Optimization Achievements
During preoptimization period,
all doses were significantly higher than pan-European unified DRLs.
After appropriate trainings and technical optimizations by vendors, physicists and radiologists median and standard deviations improved significantly.
For five basic protocols median dose levels decreased on average by 37% and standard deviations improved on average by 44% ( Fig. 2 ).To give an example,
total DLP (mGy.cm) for lumbar spine protocol was 550 ±552.
during preoptimization,
and improved to 417±169 after the optimization.
These values represent an improvement of 24% in median dose level and an improvement of 69% in standart deviation ( Fig. 3 ).
Significant improvement over unified DRLs for several protocols was also reached (up to +94% before and up to -65% after optimization) ( Fig. 4 ).
For the example given abov,e the unified DRL was 460 mGy.cm (vs.
d17 mGy.cm) ( Fig. 3 ).
Convergence Between Hospitals
After dose optimization standard deviations within each protocol have decreased and hospital dose level medians have converged significantly. To give an example,
standard deviations for thorax protocol improved by 75% on average, median dose range droppedby 34%, and protocol dose averages (medians technically) have converged ( Fig. 5 ).
Of notice is the presence of slightly higher doses -but still below the unified DRLs,
in Goztepe Hospital,
which was a university department.
Pediatric Doses
Dedicated pediatric protocols enabled healthcare professionals to decrease dose levels drastically for pediatric patients with regard to adult protocols.
As an example DLP for thorax protocol was 152 ± 83,
wheras it was only 43 ±24 for children.
Dose levels also differ significantly by age group,
positively correlating with the body size. ( Fig. 6 )
Optimization at Wider Scale
Despite of equipment and patient profile differences convergence in DLP levels have been achieved in multiple sites For example,
the thorax protocol in all hospitals and in all scanners falls within a 40 mGy.cm range ( Fig. 7 ).
However,
16 slices scanners had higher doses and lower standart deviations when compared to 128 slices scanners which are installed at same institutions.
The main reason behind that observation is the availability of dose reduction algorithms in 128 slices scanners that allowed to reach optimal image quality with substantially lower doses.