First published online July 9, 2007
British Journal of Radiology (2007) 80, 563-568
© 2007 British Institute of Radiology
doi: 10.1259/bjr/24633984
A retrospective analysis of digital stereotaxis in breast screening
K Faulkner, FIPEM, Flnst P, FSRP
S McCormack, BSc
and
K Bennison, BSc
Quality Assurance Reference Centre, Unit 9 Kingfisher Way, Wallsend, Tyne & Wear NE28 9ND, UK
Correspondence: Miss Susan McCormack, Information Audit Officer, Quality Assurance Reference Centre, Unit 9 Kingfisher Way, Silverlink Business Park, Wallsend, Newcastle-upon-Tyne, Tyne & Wear NE28 9ND, UK. E-mail: susan.mccormack{at}nhs.net
 |
Abstract
|
|---|
Digital imaging systems have been introduced into the assessment process for breast screening and should help to make the procedure quicker and reduce the trauma. However, there is concern that digital stereotaxis could have a negative impact on the accuracy of tissue sampling procedures. The objective of this study was to deduce the impact of digital stereotaxis on the assessment process in breast screening. A retrospective study was undertaken in 28 centres that performed assessments in the UK breast screening programme. Quality assurance (QA) data from 7750 women assessed using conventional stereotaxis (from 25 units) were compared with those from 4743 women assessed using digital stereotaxis (from 14 units). All the data relate to women having biopsies. The data used show the pre-operative result recorded at the core biopsy compared with the overall outcome after surgery. From this, QA parameters are calculated which are used to deduce the performance of screening programmes. Complete sensitivity, specificity (full), the positive predictive value (PPV) for B3 biopsies and the suspicious rate had statistically significant increases (p<0.05, p<0.01, p<0.05 and p<0.05, respectively) and the specificity (biopsy), false-negative rate and inadequate rate from cancers had statistically significant decreases (p<0.01, p<0.0005 and p<0.0005, respectively). One of the other QA measures improved to meet the target standard set, i.e. PPV (B5). It is concluded that digital stereotaxis has contributed to significant improvements in the assessment process, although the improvements shown in this study cannot be wholly attributed to the introduction of digital stereotaxis.
|
Introduction
|
|---|
In breast screening, if an abnormality is found on a woman's mammograms, she is invited for further examinations at an assessment clinic. For some women, a stereo pair of mammograms will be acquired, as part of a tissue sampling procedure. These images are taken from different angles to determine where to place a needle or biopsy device within the breast. Small digital field imaging has been introduced to replace film/screens for stereoscopic imaging. One advantage of digital imaging is that the images can be linked to a computer to determine accurately where the sampling device should be placed. Digital stereotaxis should make the procedure quicker and thereby reduce the trauma to women.
Most lesions referred for assessment are benign. These lesions may be evaluated using cytology alone, histology alone or a combination of both. Whitlock et al [1] compared digital imaging with film screens for stereotactic core biopsy of mammographic microcalcifications. They noted that digital stereotaxis increased the radiographic calcification retrieval rate, the absolute and complete sensitivity for the detection of ductal carcinoma in situ (DCIS) and DCIS with/without an invasive component. Becker et al [2] also noted that lesions with fewer microcalcifications could be sampled with digital stereotaxis, which achieves a greater biopsy success rate.
However, digital stereotactic attachments have poorer spatial resolution than the film/screen systems they replace. This reduction in resolution could affect the visibility of microcalcifications and other lesions, as well as the accuracy of tissue sampling procedures. The objective of this study was to assess the impact of digital stereotaxis on assessment in breast screening retrospectively.
|
Methods and materials
|
|---|
In order to compare the use of film/screen and digital stereotaxis, 32 assessment units in England were surveyed. Each unit was asked to provide data for the period 1 April 1996 to 31 March 2003. Units were asked to provide data on how many wide bore needle core biopsies were performed using stereotaxis within the UK National Health Service Breast Screening Programme (NHSBSP). None of the data related to fine needle aspiration cytology (FNAC), but some women could also have had this procedure. Only core biopsy data were obtained. Data from a centre using a prone biopsy table and centres using vacuum-assisted biopsy have been excluded. In addition, the summary results of these biopsy tests were requested. The date when digital stereotaxis was introduced was also requested from each unit. Replies were received from 28 of the 32 assessment centres surveyed. Data were excluded from the analysis if the date of introduction of digital stereotaxis was uncertain or if the data were incomplete. A total of 7750 procedures using conventional core biopsy methods (from 25 units) and 4743 procedures using digital stereotactic methods (from 14 units) were analysed using 12 quality assurance (QA) parameters defined by the NHSBSP to deduce the performance of screening programmes [3]. As a consequence, 14 units only supplied data on conventional stereotaxis and 3 units only on digital stereotaxis.
These parameters were absolute sensitivity, complete sensitivity, specificity (biopsy), specificity (full), positive predictive values (PPV(B5), PPV(B4) and PPV(B3)), false-positive rate, false-negative rate, inadequate rate, inadequate rate from cancers and suspicious rate. These were calculated using the data provided by units according to the NHSBSP guidelines [3].
Each core biopsy is classified using five categories:
- B5 Malignant
- B4 Suspicious
- B3 Uncertain malignant potential
- B2 Benign
- B1 Unsatisfactory/normal tissue only
The definitions for the quality assurance parameters for core biopsy are:
- Absolute sensitivity—The number of carcinomas diagnosed as such (B5) expressed as a percentage of the total number of carcinomas sampled.
- Complete sensitivity—The number of carcinomas that were not definitely negative or inadequate on the core expressed as a percentage of the total number of carcinomas.
- Specificity (biopsy)—The number of benign lesions (B2 results only) expressed as a percentage of the total number of cases classed as benign following any further histology.
- Specificity (full)—The number of correctly identified benign lesions (B2 results minus the number of false-negative results) expressed as a percentage of the total number of benign lesions aspirated.
- Positive predictive value of a B5 diagnosis—The number of correctly identified cancers (number of B5 results minus the number of false-positive results) expressed as a percentage of the total number of positive results (B5).
- Positive predictive value of a B4 diagnosis—The number of cancers identified as suspicious (number of B4 results minus the number of false suspicious results) expressed as a percentage of the total number of suspicious results (B4).
- Positive predictive value of a B3 diagnosis—The number of cancers identified as atypical (B3 results minus the number of benign atypical results) expressed as a percentage of the total atypical results (B3).
- False-positive rate—The number of false-positive results (cases with a B5 result that turn out to be benign after surgery) expressed as a percentage of the total number of carcinomas sampled.
- False-negative rate—The number of false-negative results (cases that turn out to be carcinomas despite a negative core result) expressed as a percentage of the total number of carcinomas sampled.
As the number of cases for each centre varied greatly, the data were analysed using a weighted t-test. This was because the standard t-test would not give appropriate weight to all the centres. After calculating the weighted estimates of the means and common variance for each parameter, the test statistic could be calculated using the usual t-test formula [4]. The level of statistical significance is then evaluated using standard statistical tables [5].
|
Results
|
|---|
Table 1
compares the weighted means for conventional and digital stereotaxis using the 12 QA parameters used by the NHSBSP with the statistical test results. Absolute sensitivity, complete sensitivity, specificity (full), PPV (B5), false-negative rate, false-positive rate and suspicious rate have minimum standards and targets [3]. The data for complete sensitivity, specificity (full), false-negative rate and suspicious rate before and after the introduction of digital stereotaxis are shown in Figures 1–4, respectively.
View larger version (22K):
[in this window]
[in a new window]
|
Figure 1. A comparison of complete sensitivity before and after the introduction of digital stereotaxis using box and whisker plots with current NHSBSP QA standards[4].
|
|
View larger version (23K):
[in this window]
[in a new window]
|
Figure 2. A comparison of specificity(full) before and after the introduction of digital stereotaxis using box and whisker plots with current NHSBSP QA standards [4].
|
|
View larger version (18K):
[in this window]
[in a new window]
|
Figure 3. A comparison of false negative rate before and after the introduction of digital stereotaxis using box and whisker plots with current NHSBSP QA standards[4].
|
|
View larger version (25K):
[in this window]
[in a new window]
|
Figure 4. A comparison of suspicious rate before and after the introduction of digital stereotaxis using box and whisker plots with current NHSBSP QA standards[4].
|
|
|
Discussion
|
|---|
The introduction of digital stereotaxis has increased absolute sensitivity by 3.1% and complete sensitivity by 7.3%, with the increase in complete sensitivity being statistically significant at the 95% level. Before the introduction of digital stereotaxis, these parameters were comparable to the current minimum NHSBSP standard [3]. Using digital stereotaxis, absolute sensitivity is just below the target standard, and complete sensitivity exceeds the target standard.
Specificity (biopsy) decreased from 34.3% to 20.1% with the introduction of digital stereotaxis, whereas full specificity increased by 1.4% from 69.9% to 71.3%, which is above the current minimum standard [3]. These were both statistically significant at the 99% level. Positive predictive values (PPV) for B5 and B3 biopsy results both increased, by 0.8% from 98.8% to 99.6% and by 0.8% from 31.7% to 32.5%, respectively, with the increase in the PPV for B3 biopsies being a statistically significant difference at the 95% level. However, PPV for B4 biopsies decreased by just under 10% from 84.3% to 74.5%.
Both the false-negative rate and false-positive rate decreased with the introduction of digital stereotaxis. They decreased from 5.6% to 1.9% and from 1.0% to 0.3%, respectively, with the decrease in the false-negative rate being statistically significant at the 99.95% level. While the false-negative rate met the target standard using the conventional stereotaxis methods, the introduction of the digital methods has further reduced the rate, representing an improved technique.
The inadequate rates both decreased following the introduction of digital stereotaxis, falling from 17.2% to 15.7% (inadequate rate) and from 6.5% to 2.9% (inadequate rate from cancers), with the latter being statistically significant at the 99.95% level. The suspicious rate increased by just under a quarter from 8.8% to 10.8%. The suspicious rate does not meet the current minimum standard. This increase was statistically significant at the 95% level. The QA parameters are interrelated, which means that improvement in one parameter will affect the others. The increase in the suspicious rate may be, in part, a consequence of the decrease in the inadequate rates.
There are a number of reasons why an improvement in the QA parameters is to be expected in the time period covered by this audit. The introduction of digital stereotaxis is just one. Given that data relating to vacuum-assisted biopsy and the use of a prone biopsy table have been excluded, then the improvement could also be due to greater experience, the use of larger biopsy needles or a reluctance to refer women for surgical biopsy so more core biopsies are repeated. While only cases that had stereotactic core biopsy were looked at, it was not recorded whether the patients also had a fine needle aspiration, the number of samples taken for each patient, the type and bore of the needle used, the amount of experience the operator had, the patient's position (i.e. prone) or if the protocol was to repeat core biopsies rather than go to surgical biopsy.
In an earlier study, Whitlock et al [1] looked at the improvement in the accuracy of core biopsy for microcalcification following the introduction of digital imaging (replacing film/screens) on an upright stereotactic unit. They studied a series of 144 women: 104 using film/screens and 40 using digital imaging. They found a statistically significant improvement in the retrieval of calcifications from 55% to 85%. Statistically significant increases in the absolute and complete sensitivity for pure DCIS were observed. Irrespective of whether there was an invasive component to DCIS, digital imaging significantly improved both the absolute and the complete sensitivity of core biopsy.
While our study has some limitations due to its retrospective nature, it is larger than that of Whitlock et al [1] in terms of both the number of units surveyed and the number of patients included. This study also confirms the findings of both Whitlock et al [1] and Becker et al [2] that digital imaging improves the core biopsy process.
The radiation dose from digital stereotaxis procedures has been assessed by Faulkner and Bennison [6]. The estimated mean glandular dose was 4.5 mGy, which is comparable to the dose received by the woman at initial screening stage [7].
|
Summary and conclusions
|
|---|
Some of the QA parameters used to monitor the performance of the assessment process in the NHSBSP exhibited statistically significant improvement with the introduction of digital stereotaxis (i.e. complete sensitivity, specificity (biopsy), specificity (full), PPV (B3), false-negative rate and inadequate rate from cancers). There was also improvement in the absolute sensitivity, PPV for B5 biopsies and the false-positive rate, although this was not significant. The improvement in PPV for B5 biopsies now meets the target standard set, and the false-positive rate now meets the minimum standard. Only the suspicious rate exhibited a statistically significant decline in performance, falling just below the minimum standard.
It would appear that the introduction of digital stereotaxis has meant that it is easier to meet the current minimum and target standards for the assessment process in the NHSBSP [3]. Owing to the nature of this study and the fact that the data were collected retrospectively, there are a number of potential biases that have not been removed. It can be expected that many of these biases would also have contributed to the improvement in the NHSBSP standards, but we conclude that digital stereotaxis has contributed to significant improvements in the assessment process.
Digital stereotaxis in breast screening
|
Acknowledgments
|
|---|
This research was partially supported by the European Commission's Radiation Protection Research Programme DIMOND III and SENTINEL (FP6 012909). We would like to thank all the assessment units who took part in the survey.
Received for publication November 15, 2005.
Revision received October 10, 2006.
Accepted for publication October 24, 2006.
|
References
|
|---|
- Whitlock JP, Evans AJ, Burrell HC, Pinder SE, Ellis IO, Blamey RW, et al. Digital imaging improves upright stereotactic core biopsy of mammographic microcalcifications. Clin Radiol 2000;55:374–7.[CrossRef][Medline]
- Becker L, Taves D, McCurdy L, Muscedere G, Karlik S, Ward S. Stereotactic core biopsy of breast microcalcifications: comparison of film versus digital mammography, both using an add-on unit. Am J Roentgenol 2001;177:1451–7.[Abstract/Free Full Text]
- NHSBSP guidelines for non-operative diagnostic procedures and reporting in breast cancer screening. NHSBSP Publication No. 50, 2001:46–51
- Bland JM, Kerry SM. Statistics notes: weighted comparison of means. Br Med J 1998;316:129[Free Full Text]
- Neave HR. Statistics tables. London: Routledge, 1978:41
- Faulkner K, Bennison K. An assessment of digital stereotaxis in the National Health Service Breast Screening Programme. Rad Prot Dosim 2005;117:327–9.
- Young KC, Burch A, Oduko JM. Radiation doses received in the UK Breast Screening Programme in 2001 and 2002. Br J Radiol 2005;78:207–18.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
BJR review of the year -- 2007
Br. J. Radiol.,
April 1, 2008;
81(964):
265 - 269.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
