RPS 214 - Image quality considerations and challenges: radiography and mammography

Purpose:

Mammography is the most effective method available for the early detection of breast cancer but many women report experiencing some anxiety, pain, and discomfort when they undergo these procedures. In recent years, substantial research claims that humour and laughter possess unique characteristics for coping with pain and stress. The aim of this study was to determine the effect of humour on levels of anxiety during mammography.

A total sample of 120 outpatients from public and private facilities was divided into a control group (n=60) and an experimental group (n=60). Sociodemographic interview and state-trait anxiety inventory were applied to both after consent of participation in the study. The experimental group watched funny videos before mammography while the control group did not watch any videos.

Cronbach's alpha coefficient was excellent (0.909). No significant correlations between the control group and the experimental group (r=0.831, P=0.831) were observed. However, in the private facility, there was a slight decrease in anxiety levels in the experimental group (66% of the patients who watched the videos had a low state of anxiety compared to 60% in the control group), but the levels were the same in the public facility for both groups (90% of women reported a low to moderate state of anxiety).

Results from this study have provided important information regarding anxiety levels during mammography but can't confirm that humour techniques can decrease anxiety. Additional research on the effects of humour and other techniques (such as music) is recommended to explore new skills that can help to decrease patient’s anxiety, pain, and discomfort.

The sample size.

The ethics committee approved the study and written informed consent was obtained.

No funding was received for this work.

Purpose:

The success of a screening program is guaranteed, among other things, by the synergy between the radiologist and the mammographer. The mammographer is the main actor, as both the quality of the services and the adherence to the program is subordinate to its technical-professional and relational skills. Thus, a mammography screening program needs mammographers trained, responsible, and aware; all qualities that can be developed and consolidated solely through a theoretical-practical training program tailored to the specific needs of each service. With this background, a training project was launched at the ASL of Latina.

The quality assessment and training activities were entrusted to an external senior mammographer consultant with specific skills and qualifications in breast imaging. The choice of an external professional ensured the establishment of a fiduciary and peer relationship with the recipients of the project. The project included theoretical and practical meetings.

A PGMI system was used to evaluate the images.

The scores of 136 images randomly selected before the project started were as follows: P 9.5%, G 7.5%, M 50%, and I 33%.

23 mammograms contained at least one inadequate image (67.50% of the total) and 8 inadequate to recall (6% of the total). At one month from the end of both training and corrective steps, the images classified as M presented a strong percentage reduction (30% vs 50%) for the benefit of the images classified as P and G (now, respectively 51.25% and 30%), with a total reset of I images.

The excellent results of the project confirmed that the verification and monitoring of a mammographer’s performance is a powerful tool for improving the technical-diagnostic quality of mammography screening programs.

n/a

n/a

No funding was received for this work.

Purpose:

To evaluate published evidence and calculate the diagnostic test performance of supplementary imaging procedures for breast cancer screening in women with dense breasts.

Data sources for identifying journal publications included MEDLINE (ProQuest), CINAHL, Health and Medicine Databases (ProQuest), Web of Science, and the Cochrane library. Reference lists of the studies and relevant systematic reviews that were retrieved were reviewed to identify relevant articles that were not identified in the literature searches. Two reviewers independently reviewed the titles, abstracts, and full-text articles to determine if studies met the inclusion and exclusion criteria. All studies were required to report diagnostic test performance characteristics using a supplementary screening with any imaging modality after a negative digital screening mammogram in asymptomatic women with dense breasts (BI-RADS C/D) as the sole risk factor. Discordance throughout the process was resolved by a third reviewer. Study quality was assessed using QUADAS-2 tool.

28 journal publications satisfied the inclusion criteria. 16 of these reported on hand-held ultrasound, 2 reported on automated whole breast ultrasound, 5 reported on MRI, 2 reported on digital breast-tomosynthesis, and 2 reported on molecular breast imaging. Only 1 study reported on contrast-enhanced digital mammography. Overall, the evidence indicates that supplementary screening among women with dense breasts and recent negative mammography can consistently identify additional cancers, most of which are invasive, but also leads to unnecessary recalls and biopsies.

Supplementary imaging procedures consistently detected additional breast cancers not identified by digital mammography in asymptomatic women with dense breasts. The inclusion of supplementary imaging into routine screening of dense breasts is recommended but will need to consider the availability and local yield.

n/a

n/a

No funding was received for this work.

Purpose:

In Japan, an objective structured clinical examination (OSCE) is an emerging trend. One of the most difficult positioning techniques to be acquired is breast positioning in mammography. In this study, we evaluated the effectiveness of the eye-tracking system as an assessment tool.

Two skilled mammographers (A and B) and two third-year university students (C and D) were recruited for a preliminary study. Three areas of interest (AOIs) were selected to detect visual attention: a) the juxtathoracic region (pectoral muscle), b) entire glandular tissue, and c) abdominal tissue (infra-mammary fold). Saccade-fixation eye movements were detected using gaze plots and heat maps. Dwell time was also assessed within selected AOIs.

Heat maps during the moving of the compression device into the position illustrated the differences of visual attention among participants. Expert A was gazing at the pectoral muscle and glandular tissue but had no gaze at the infra-mammary fold. Expert B was found to have strong visual attention on the pectoral muscle. Student C did not show any gaze within the selected AOIs. Student D was gazing at the pectoral muscle and infra-mammary fold. In addition, gaze at the display of the compression force was observed among students. Dwell time within AOIs also showed the differences in visual attention among participants.

An eye-tracking system can be used as an assessment tool for mammography positioning. Using both a heat map and dwell time within AOIs should be applied for a more accurate assessment.

We need to further examine the association between the obtained clinical images with visual attention to investigate the effectiveness of the eye-tracking system.

Informed consent was obtained from all participants.

No funding was received for this work.

Purpose:

To identify the most relevant image quality (IQ) criteria that should be used to assess breast implants mammograms (BIM) according to radiographers’ and radiologists' working in breast cancer screening programs perspectives.

A two-round Delphi method using a questionnaire was applied, asking the experts to rank each image criteria encountered in the literature according to 4 possible answers: 1=need to have/2=nice to have/3=not relevant/4=does not know. Criteria for craniocaudal, mediolateral-oblique, and lateral, performed with and without Ecklund manoeuvre, were included. This process was repeated modifying criteria in the second round by removing the less relevant criteria highlighted during the first round and including participants’ suggestions.

Visualisation of anatomical details (extension of the pectoral muscle, nipple in profile) was considered the most relevant criteria by radiographers during the first round, while general criteria were prioritised by radiologists in both rounds, namely the sharpness of glandular tissue, absence of artefacts, and the spread of breast tissue. The analysis of implant flow, the BI anterior edge, and its maximum retropulsion when the Eklund manoeuvre was performed were the specific criteria for BIM considered as the most relevant to be assessed. The visualisation of retroglandular adipose tissue and axillary tail were considered as “nice to have”.

The importance of each criterion used to assess BIM was not the same for radiographers and radiologists, which suggests the two groups of experts are looking for different aspects of the image. Further education and training are necessary to align the strategy to assess IQ of BIM and reduce subjectivity.

This study did not include all BCS centres available in the country, not being possible to generalise these results.

n/a

Funding provided by HES-SO.

Purpose:

Previous research indicates that radiologists and radiographers agree on the visibility of anatomical structures in radiographs but not on the clinical usability. The impact of the differences in the acceptance of a radiograph on the radiologist’s confidence to diagnose is unknown.

24 radiographers and 6 radiologists scored 40 PA chest radiographs (24 without and 16 with pathology) for clinical usability (RadLex) in the detection of lung lesions. Additionally, radiologists classified the radiographs as normal or abnormal and scored their confidence on a 5-point scale. AUC_ROCwere calculated for each RadLex category to evaluate the alignment between the radiographer’s assessment and the radiologist’s performance.

Based on the AUC_ROC, the classification of the radiographs by radiographers in terms of clinical usability was no predictor for the clinical performance of the radiologist. A significant different (p<0.01) classification was noticed between radiographers and radiologists. 25.3% of the cases, rejected by the radiographers (rejected 56%), sufficiently supported the detection of pathology for the radiologist (rejected 30.7%).

Almost half of the radiographs rejected by radiographers were potentially diagnostic, with useful doses likely being discarded. The discrepancy between professions in the evaluation of clinical usability alludes to a different approach in image evaluation. Future research is needed to align both professions, with agreement and collaboration being untapped niches for potential dose optimisation strategies.

The number of cases was relatively low to facilitate observers with limited availability and so the prevalence did not reflect the clinical practice.

n/a

No funding was received for this work.

Purpose:

Radiographers differ from radiologists in their assessment of image quality, which may have connotations for patient diagnosis and radiation dose. Understanding the perceptual basis for these judgements is warranted.

Expert radiologists are thought to be guided in pathology detection by an initial global impression, detecting abnormalities at a rate higher than chance in under 250 ms. This study aims to determine whether radiographers employ a similar process, detecting a global signal for “inadequate” image quality.

27 qualified radiographers attending the EFRS Research Hub 2019 viewed 24 chest radiographs of varying quality under two conditions and rated the image quality of each on a scale of 1 (very inadequate) to 5 (excellent); ratings of 1 or 2 indicated a repeat image was necessary.

In the first condition ("flash"), a sequence of fixation cross (750 ms), radiograph (200 ms), mask (50 ms), and a blank screen with rating scale (unlimited) were presented for each radiograph. In the second condition ("free"), images were presented in a different order with unlimited time, representing standard viewing conditions.

35.69% of the images had identical ratings in each condition. 69.61% were given the same decision to accept/reject in both conditions. Where the decision changed, 74.6% were initially accepted. Flash and free ratings were correlated (r=0.507, p<0.01); mean flash ratings were higher than free (p<0.01).

Radiographers may use a global impression in determining image quality, though further research is required.

Selection bias is possible as all participants were ECR attendees. The limited number and range of images limits the findings.

The work was submitted to the home institution human research ethics committee and declared exempt from a full review.

The study was conducted at the EFRS Research Hub and used monitors donated by Barco.

Purpose:

To determine the impact of reduced visual acuity on object counting and image evaluation of anatomical details in appendicular radiographs.

Phantom images were produced for the purpose of counting objects and image appraisal of anatomical structures was performed on anonymised radiographies. A 4-point Likert-scale was applied to characterise the level of anatomical definition (1-not defined; 4-clearly defined). The tasks were performed under 3 conditions: wearing 2, 1, and 0 pairs of simulation glasses, which provided different magnitudes of reduced visual acuity. Inferential statistics were carried out using SPSS and areas under the visual grading characteristic curve (AUC_VGC) were used as figure-of-merit. The tasks were completed by radiography students participating in a research summer school.

Observers were not able to count low-contrast objects with 2 pairs of glasses (13). The evaluation of anatomical details was performed by 7 radiography students wearing 0 and 2 pairs of glasses, but total interpretation time was longer when visual acuity was reduced (15.4 versus 8.9min). VGC analysis showed that observers can lose the ability to detect anatomical and contrast differences when they have a simulated visual acuity reduction, with low-contrast detection presenting more of a challenge (soft tissues). No simulation glasses compared with 1 pair gives binomial AUC_VGC of 0.302 (0.280, 0.333), that decreases to 0.197 (0.175, 0.223) when using 2 pairs.

Reduced visual acuity has a negative impact on the evaluation of test-objects and clinical images. Further work is required to test the reduced visual acuity impact on the technical evaluation of different imaging, pathologies, and levels of experience.

There was no optometrist available to measure the differences on observers’ visual acuity.

HSR1819-115.

No funding was received for this work.

Purpose:

To characterise differences in the evaluation of the quality of paediatric chest radiographs between radiographers and radiologists.

24 paediatric chest radiographs, performed before and after the implementation of a dose optimisation program, were analysed by 5 observers: 2 radiologists and 3 radiographers. This evaluation was performed blindly, independently, and without a time limit. The evaluation included an analysis of anatomical criteria (1 to 5-point Likert scale) and visual perception of physical image characteristics (1 to 3-point Likert scale) such as noise, resolution, and contrast (11 criteria). Visual grading analysis and visual grading characteristics analysis were performed.

The analysis of the scores was performed by criterion and by phase in the different weight ranges, and 1,320 scores were obtained. No statistically significant differences were found for scores between both phases (p=0.131). Regarding the differences between professional classes, radiographers gave lower scores than radiologists in almost all criteria and all weight levels, regardless of the phase. The average scores assigned by radiographers ranged from 1.3-4.9, with the most frequent average score being 4.7 (8.0%). The average scores given by the radiologists ranged from 2.0-5.0, with the most frequent mean score being 5.0 (43.2%). In the two analysed phases (before and after optimisation), in all weight ranges and criteria, statistically significant differences (p<0.05) were found between the evaluations performed by radiographers and radiologists.

The evaluation of the quality of chest radiographs was influenced by the evaluator's academic background, which may have implications for adherence to x-ray exposure optimisation programs.

The number of images and observers were limited.

Approved by the ethical board of the hosting hospital.

No funding was received for this work.

Click here to add a description...

Purpose:

Reject analyses (RA) in digital radiography report low reject rates for inadequate exposure because digital systems can produce diagnostic images over a broad range of exposure levels. A review of current literature indicates a lack of investigation into whether reduced rates of exposure error are actual or whether they are masked by post-processing techniques that optimally render images for display such that less-than-optimal exposure goes unnoticed. The current study aimed to investigate how exposure faults can be accounted for in the RA of digital radiographs.

A system to store digital reject images was set up. Rejects and their associated anonymised data were collected over an audit period of one month. To facilitate the analysis of exposure through the calculation of the deviation index (DI), examination specific target exposure values were derived in compliance with IEC advice. An objective quality analysis of the images was conducted at a RIS-PACS workstation at the researchers’ university.

A comparison of the frequency of exposure error reported in the RA (5.7%) and DI analysis (20%) revealed that RA of itself is not effective in identifying exposure error in digital radiographs. Image quality evaluation using Spearman's non-parametric test of correlation confirmed that no significant correlation existed between SNR or CNR and DI values.

RA without further consideration of exposure is not effective in identifying exposure error in digital radiographs. The addition of exposure analysis using DIs provides information that can be used to inform optimisation strategies for digital radiographs. Thresholds for SNR and CNR in terms of diagnostic efficacy should be investigated to determine whether target exposures could be reduced without an image detriment.

The study is specific to a site using Siemens technology. The method, however, is robust/reproducible and could be adapted/replicated in other sites.

Ethical approval from UCD SM UREC was granted (November 2018).

No funding was received for this work.