Here we will outline some of the previous and ongoing work carried out within University College Cork,
Cork University Hospital and the Mercy University Hospital which has helped to begin to clarify the role that iterative reconstruction can play within the clinical sphere.
A prospective feasibility study of sub-millisievert abdominopelvic CT using iterative reconstruction in Crohn’s disease (12)
In one of our earliest studies assessing the feasibility of the clinical application of a low dose protocol with ASIR reconstruction we recruited 40 patients to undergo two synchronous CT protocols using a conventional dose and a low dose protocol.
The studies acquired using a low dose approach were reconstructed using an ASIR algorithm.
Fig. 1: Conventional (2.5 mSv) and b low-dose (0.48 mSv) axial CT images reconstructed with ASIR40 demonstrating sigmoid mucosal hyperenhancement (arrow) in a 17-year-old male patient
The conventional dose CT abdomen and pelvis studies (these were reconstructed with filtered back projection and 40% ASIR),
the control arm for the study,
were designed with manufacturer advice to effect a radiation exposure 80-90% that of a standard departmental CT abdomen and pelvis.
The low dose CT studies imaged an identical area to the conventional dose studies but with an effective dose approximating 1mSv,
or 10-20% that of the standard departmental CT (ethical approval for acquisition of two CT data sets in each patient was justified because the cumulative radiation exposure was less than or equivalent to that of our standard departmental CT abdomen and pelvis).
The low dose studies were used for study purposes only and did not guide clinical management.
Reconstructions were retrospectively performed on raw data,
using inbuilt hybrid iterative reconstruction software on the CT system.
Fig. 2: Conventional (6.5 mSv) and low-dose (1.6mSv) coronal CT images reconstructed with ASIR40 demonstrating an enterocolic fistula (arrow) in a 34-year-old male patient.
This study was able to demonstrate that diagnostic sub-millisievert CT abdomen and pelvis studies are feasible in patients with Crohn’s disease at an average effective dose of 1.3 mSv (and allow a 74% radiation dose reduction),
which is equivalent to two abdominal radiographs and is capable of accurately detecting acute complications of Crohn’s disease.
Contrast resolution was lower on the low dose studies,
which did limit solid organ assessment but no significant pathology was missed.
The role of pure iterative reconstruction in conventional dose CT enterography(15)
In 2015,
our group published a paper which compared standard protocol CT abdomen and pelvis studies with a pure iterative reconstruction of the same study.
Clinical interpretation was then performed by two blinded radiologists to assess the benefit of this iterative reconstruction algorithm.
We were able to show that pure iterative reconstruction images significantly outperformed the original study carried out using a standard reconstruction algorithm (p < 0.001).This study demonstrated that conventional studies reconstructed with pure iterative reconstruction demonstrated excellent noise reduction,
spatial resolution,
contrast resolution and diagnostic acceptability The most impressive subjective performance of pure IR was in terms of streak artefact reduction,
which in most cases,
was completely eliminated.
This study demonstrates that pure iterative reconstruction considerably improves image quality of conventional CT enterography studies and that even outside the low dose sphere,
that this technology can confer significant value to the diagnostic process.
Fig. 3: Coronal reformatted images in a 35-year-old female patient (BMI 23.4) showing hybrid IR and pure IR images at the same anatomical points. A comb sign is seen in the right iliac fossa along with adjacent prominent lymph nodes (arrow).
Pure Iterative Reconstruction Improves Image Quality in Computed Tomography of the Abdomen and Pelvis Acquired at Substantially Reduced Radiation Doses in Patients With Active Crohn Disease (14)
Here we assessed the diagnostic accuracy and image quality of modified protocol computed tomography of the abdomen and pelvis reconstructed using pure iterative reconstruction in patients with Crohn’s disease presenting with suspected extramural complications. Following ethical approval,
two contemporaneous CT datasets were acquired in all patients – a standard protocol (80-90% of the dose of routine abdominopelvic CT) and a modified protocol (using 10-20% of the routine dose).
The datasets acquired using the modified low dose protocol were reconstructed using ASIR and MBIR.
While,
as expected,
median diagnostic acceptability,
spatial resolution and contrast resolution were significantly lower and subjective noise scores were higher when utilising the low dose modified protocol compared with the standard protocol,
we demonstrated perfect clinical agreement when assessing for extramural complications between the standard protocol compared with the low dose protocol reconstructed with MBIR and we concluded that this modified protocol CT using pure iterative reconstruction is feasible for assessment of active Crohn’s disease
Fig. 4: Routine dose (standard protocol) CT and low dose CT datasets reconstructed with ASIR and MBIR. While noise is higher in the ASIR and MBIR images, diagnostic acceptability is adequate in all three
Accurate tissue characterization in low-dose CT imaging with pure iterative reconstruction(20)
In a follow on to the previous study,
using sub-millisievert abdominopelvic CT for Crohn’s disease patients,
we used the same data-sets to assess the ability of the low-dose hybrid reconstruction and “pure” model-based IR (MBIR) images to maintain accurate Hounsfield unit (HU)-determined tissue characterisation when compared with standard CT abdomen and pelvis studies.
This study used the same standard-protocol and low-dose modified protocol data sets as described above (14).
This study showed the Hounsfield tissue volume measurement generated using the MBIR reconstruction algorithm closely correlated with the Hounsfield units calculated using a standard CT abdomen and pelvis reconstruction technique.
When considering the potential of introducing low dose MBIR protocols into clinical practice it is important to be confident that these protocols can produce consistent Hounsfield unit measurement as these values are of core importance when defining tissue density and in tissue and lesion characterisation during CT assessment.
Effects of oral contrast on dose in abdominopelvic computed tomography with pure iterative reconstruction (21)
Having demonstrated the clinical utility of pure iterative reconstruction within a Crohn’s patient cohort we have also tried to further investigate protocol modification to further decrease dose.
We randomised Crohn’s patients (n =79) undergoing clinically indicated CT studies using positive contrast (n = 35,
1L 2% gastrograffin) or negative contrast (n = 44,
1.5L polyethylene glycol). We showed that higher doses were seen in those patients who had taken negative oral contrast,
likely due to the osmotic effect of the agent resulting in larger intraluminal fluid volumes and resultant increased overall beam attenuation.
Low-dose abdominopelvic CT using pure iterative reconstruction in patients presenting to the emergency department with acute abdominal symptoms (22)
In an effort to assess the potential of this low dose pure iterative reconstruction technique in the patient presenting with acute abdominal pain [without a known diagnosis as in the earlier studies(12,14)] we prospectively recruited 57 patients presenting to the emergency department of Cork University Hospital.
Following written consent,
each patient was scanned using a split dose technique whereby 80-90% of the standard effective dose was used to generate a control study and this was immediately followed by a low dose CT using the remaining 10-20% of the standard CT dose.
Blinded reads by two gastrointestinal radiologists were performed in order to objectively and subjectively assess image quality from both datasets and the ability to identify pathology and reach a diagnosis was compared between both groups.
Fig. 5: Coronal image of the abdomen and pelvis reconstructed using MBIR at 10-20% of the dose of a standard CT abdomen and pelvis demonstrating acute appendicitis
Of the 57 patients scanned,
28 patients had an identifiable pathology.
The low dose studies were able to correctly identify this pathology in all but one of the abnormal studies.
The positive predictive value and negative predictive value of the low dose compared with the conventional dose studies were 96.4% and 96.6% respectively.
While these low dose studies had increased subjective image noise and poorer contrast and diagnostic resolution,
the low dose pure iterative reconstruction provided diagnostic imaging in patients presenting with acute abdominal pain while imparting a significantly lower patient dose.
Fig. 6: Axial images of the abdomen and pelvis demonstrating routine dose (standard protocol) CT (left) and low dose pure iterative reconstruction CT (right)
Low-Dose Carotid Computed Tomography Angiography Using Pure Iterative Reconstruction (15)
While much of our clinical work on the potential utility of iterative reconstruction has concentrated on abdominal imaging,
we have also assessed whether a low dose carotid computed tomography (CTA) angiography study performed with pure iterative reconstruction is comparable to a conventional dose CTA protocol.
In this study we recruited 20 patients,
and radiation dose was divided into a low-dose acquisition reconstructed with pure IR and a conventional dose acquisition reconstructed with 40% hybrid.
As expected in these patients the subjective image noise was higher in the low dose studies and contrast and spatial resolution were lower but there was excellent agreement for stenosis grading accuracy between low and conventional dose studies.
This study demonstrated that a low dose protocol is able to effect a dose reduction of 49.6% while maintaining adequate image quality and diagnostic accuracy.
Minimizing dose in cystic fibrosis patients
Our unit has a firm research interest in the area of cystic fibrosis and the minimisation of radiation exposure in this cohort.
Patients with cystic fibrosis can expect exposure to a significant radiation dose during their lifetime,
over and above that of the general population (10,11,23).
Our group is currently working to develop and assess modified low-dose volumetric chest CT protocols using pure iterative reconstruction techniques which will allow imaging of the thorax at radiation doses similar to that of a standard chest radiograph.
Initial experience with this ultra-low dose technique has demonstrated promising results and images acquired to date have been deemed to be of diagnostic quality.
We hope that this CT protocol can be utilised to allow follow-up imaging of cystic fibrosis to occur via CT as opposed to standard chest radiograph,
with the benefit of allowing increased bronchial and pulmonary parenchymal visualisation.
Fig. 7: Modified low-dose volumetric CT thorax using pure iterative reconstruction techniques acquired at radiation dose equivalent to a PA erect and lateral chest radiograph in a 32 year old male with FEV1 of 47%