- Email: [email protected]
Comparison of ultrasound with mammography in the symptomatic patient
366
CLINICAL RADIOLOGY
services in the metropolitan area around Perth where most women live are similar to those in other cities but special provision has to be made for w o m e n in country areas. The results from the first three screening units are presented and the very different areas covered are illustrated. The first unit is at a fixed site in suburban Perth, the second is a bus which covers the southwest of the state (13 500 women in 26 819 sq km) and the third is a van moved by a prime mover which covers the whole of the north west of the state (10 000 women in 1 687 517 sq kin). The service in the north west is unique, covering such an area with a significant number of the women being aboriginal.
CAN W E P R E D I C T W H I C H W O M E N W I L L FIND M A M M O G R A P H Y PAINFUL? E. J. W Y L I E and R. M. A D A M S O N
Department of Diagnostic Radiology, Royal Perth Hospital, Perth, Western Australia Anticipation of a painful examination may discourage women from presenting for screening m a m m o g r a p h y . To determine if we could predict which women might find m a m m o g r a p h y painful, a questionnaire was completed by 315 consecutive women presenting to a general hospital radiology department for two-view film screen m a m m o g r a p h y . The questions included: menstrual history, breast pain symptoms, if they anticipated that m a m m o g r a p h y would be unpleasant, and if this had delayed their attending the procedure. After the examination women were asked to grade the comfort of the procedure. The pressure applied to each breast, the a m o u n t of glandular tissue and the breast size from the m a m m o g r a m s was recorded. 81% of women found the examination painless or only slightly uncomfortable, 14.5% found the examination painful and 4.5% considered the procedure very painful. 15% of women who anticipated that m a m m o g r a p h y would be painful delayed attending for the examination. A painful examination was more likely if a w o m a n anticipated pain, suffered from frequent breast pain or had a previous history of breast surgery. As the expectation of a painful examination may not only prevent women from presenting for m a m m o g r a p h y but appears to be associated with a higher incidence of painful examinations, women need to be reassured that the majority of examinations cause minimal discomfort.
TO SWING OR NOT TO SWING E. M c C A N N , J. S U R R E Y and S. A. M E R C E R
X-Ray Department, Princess Grace Hospital, London M a n y different views can be taken to show different parts of the breast anatomy. In breast screening it is important to get m a x i m u m information using the m i n i m u m number of views. It is known that the whole of the medial part of the breast m a y not be demonstrated on the lateral oblique projection, and some authorities suggest rotating the craniocaudal to show as much as possible of the medial part of the breast. However, most cancers occur in the lateral part of the breast, so that to miss the axillary tail is not satisfactory. The areas of the breast lost on conventional radiography are demonstrated, and the percentage of tumours in the areas of the breast are quoted.
COMPARISON OF ULTRASOUND WITH MAMMOGRAPHY IN T H E S Y M P T O M A T I C P A T I E N T D. M A R T I N E Z , B. C L A R K S O N and A. C H A P M A N
Department t?fRadiology, St James's University Hospital, Leeds Over a 5 m o n t h period ultrasound was performed in addition to m a m m o g r a p h y by two trained radiographers on all symptomatic patients. Both examinations were reported by one of four radiologists and scored into five groups: normal, benign disease, probably benign, probably malignant and malignant disease. Patients presented with palpable lump, pain or discharge ( N = 549, 608 symptomatic breasts [S] and 465 asymptomatic breasts [AS]). Four groups emerged: normal m a m m o g r a m , normal ultrasound (288 S, 333 AS); abnormal m a m m o g r a m , abnormal ultrasound (188 S, 82 AS); normal m a m m o g r a m , abnormal ultrasound (120 S, 49 AS); abnormal m a m m o g r a m , normal ultrasound (12 S, 1 AS). The majority of the abnormal ultrasound, normal m a m m o g r a m group showed benign disease (cysts, duct ectasia) with the palpable lump accounted for in 60% of cases. Four of this group had a malignancy on ultrasound (confirmed histologically) not suspected on m a m m o g r a p h y .
Five patients in the abnormal m a m m o g r a m , abnormal ultrasound group had benign disease on m a m m o g r a m upgraded to probably malignant by ultrasound (confirmed histologically). In the abnormal m a m m o g r a m , normal ultrasound group 12 had benign disease (three with calcifications only) and one had malignant microcalcifications confirmed on histology. In conclusion ultrasound has provided useful additional information in a large number o f symptomatic breast patients. Of particular interest are the four patients with malignancy seen on ultrasound alone.
C O M P U T E R I N T E R P R E T A T I O N OF D I G I T A L MAMMOGRAMS D. H. DAVIES and D. R. D A N C E
Department of Physics, Institute of Cancer Research, The Royal Marsden Hospital, London Analysis of the large n u m b e r of images produced by the U K breast screening programme m a y be enhanced if an automatic computer system could be used to pre-read the m a m m o g r a m s . Two preliminary studies into the automatic detection of calcifications have been conducted. In the first study 75 randomly selected conventional screen film m a m m o g r a m s were digitized at 100 microns and 8 bits. These were divided into training and test sets of 25 and 50 fihns respectively. The results for the test set of 50 complete clinical m a m m o g r a m s show that the computer system achieves a 25/25 true positive film classification with false positive clusters detected in 4/50 films. There were no false negative film classifications. In the second study a small n u m b e r of m a m m o g r a m s of radiographically very dense breasts containing examples of small (<0.014 m m 2) subtle (optical density differences < 0.035) calcifications were selected. These films were digitized using a high resolution (43 micron), high quality (12 bit) laser scanner and were selected to challenge the algorithm developed in the initial study. This is the first time that the detection of small subtle calcifications has been able to be investigated. Improvements were made to the algorithm used in the first study, but our results show that the automatic detection rates for these small subtle calcifications have yet to reach an acceptable level.
SELF A S S E S S M E N T O F F I L M READING P E R F O R M A N C E 1N BREAST S C R E E N I N G A. G. G A L E , J. A. T O W L E , E. J. R O E B U C K * and A. R. M. W I L S O N *
Department of Academic Radiology and Breast Screening Training Centre and *University Hospital and City Hospital, Nottingham A national self-assessment programme of radiologists' film reading performance has been established by the NHSBSP and the Royal College of Radiologists. The purpose of this is to enable each radiologist involved in screening to learn how they interpret sets of known cases. Following piloting of the methodology a three region trial has been carried out. A set of exemplary cases encompassing a diverse range of screening appearances has been amassed from screening training centres and a gold standard of radiological interpretation obtained. Individually each screening radiologist in the three regions has reviewed these cases and measures of their film reading performance recorded. Performance indices monitored include: the number of cases recalled for assessment; sensitivity and specificity; identification of key radiological features; and R O C measures of detection and classification ability. Furthermore the data analyses allow identification of areas where an individual has particular difficulty which then permits the suggestion of appropriate revision strategies. It is proposed that this self-assessment programme will be extended nationally and will become an annual exercise so that every radiologist can follow the progression of their performance. This paper details the results of the trial phase which encompasses the Trent, East Anglian and Northern regions.
M A M M O G R A M S AT D I F F E R E N T C O M P R E S S I O N S F O R T H E D E T E C T I O N OF BREAST C A N C E R R. P. H I G H N A M , J. M. B R A D Y * and B. SHEPSTONE~
Computing Laboratory and *Department of Engineering, Oxford University, and ~fBreast Care Unit, Churchill Hospital, Oxford Mammographic signs are sometimes non-specific if not ambiguous. It is proposed that by taking two m a m m o g r a m s of the same breast at different compressions important new information becomes available.
CLINICAL RADIOLOGY
services in the metropolitan area around Perth where most women live are similar to those in other cities but special provision has to be made for w o m e n in country areas. The results from the first three screening units are presented and the very different areas covered are illustrated. The first unit is at a fixed site in suburban Perth, the second is a bus which covers the southwest of the state (13 500 women in 26 819 sq km) and the third is a van moved by a prime mover which covers the whole of the north west of the state (10 000 women in 1 687 517 sq kin). The service in the north west is unique, covering such an area with a significant number of the women being aboriginal.
CAN W E P R E D I C T W H I C H W O M E N W I L L FIND M A M M O G R A P H Y PAINFUL? E. J. W Y L I E and R. M. A D A M S O N
Department of Diagnostic Radiology, Royal Perth Hospital, Perth, Western Australia Anticipation of a painful examination may discourage women from presenting for screening m a m m o g r a p h y . To determine if we could predict which women might find m a m m o g r a p h y painful, a questionnaire was completed by 315 consecutive women presenting to a general hospital radiology department for two-view film screen m a m m o g r a p h y . The questions included: menstrual history, breast pain symptoms, if they anticipated that m a m m o g r a p h y would be unpleasant, and if this had delayed their attending the procedure. After the examination women were asked to grade the comfort of the procedure. The pressure applied to each breast, the a m o u n t of glandular tissue and the breast size from the m a m m o g r a m s was recorded. 81% of women found the examination painless or only slightly uncomfortable, 14.5% found the examination painful and 4.5% considered the procedure very painful. 15% of women who anticipated that m a m m o g r a p h y would be painful delayed attending for the examination. A painful examination was more likely if a w o m a n anticipated pain, suffered from frequent breast pain or had a previous history of breast surgery. As the expectation of a painful examination may not only prevent women from presenting for m a m m o g r a p h y but appears to be associated with a higher incidence of painful examinations, women need to be reassured that the majority of examinations cause minimal discomfort.
TO SWING OR NOT TO SWING E. M c C A N N , J. S U R R E Y and S. A. M E R C E R
X-Ray Department, Princess Grace Hospital, London M a n y different views can be taken to show different parts of the breast anatomy. In breast screening it is important to get m a x i m u m information using the m i n i m u m number of views. It is known that the whole of the medial part of the breast m a y not be demonstrated on the lateral oblique projection, and some authorities suggest rotating the craniocaudal to show as much as possible of the medial part of the breast. However, most cancers occur in the lateral part of the breast, so that to miss the axillary tail is not satisfactory. The areas of the breast lost on conventional radiography are demonstrated, and the percentage of tumours in the areas of the breast are quoted.
COMPARISON OF ULTRASOUND WITH MAMMOGRAPHY IN T H E S Y M P T O M A T I C P A T I E N T D. M A R T I N E Z , B. C L A R K S O N and A. C H A P M A N
Department t?fRadiology, St James's University Hospital, Leeds Over a 5 m o n t h period ultrasound was performed in addition to m a m m o g r a p h y by two trained radiographers on all symptomatic patients. Both examinations were reported by one of four radiologists and scored into five groups: normal, benign disease, probably benign, probably malignant and malignant disease. Patients presented with palpable lump, pain or discharge ( N = 549, 608 symptomatic breasts [S] and 465 asymptomatic breasts [AS]). Four groups emerged: normal m a m m o g r a m , normal ultrasound (288 S, 333 AS); abnormal m a m m o g r a m , abnormal ultrasound (188 S, 82 AS); normal m a m m o g r a m , abnormal ultrasound (120 S, 49 AS); abnormal m a m m o g r a m , normal ultrasound (12 S, 1 AS). The majority of the abnormal ultrasound, normal m a m m o g r a m group showed benign disease (cysts, duct ectasia) with the palpable lump accounted for in 60% of cases. Four of this group had a malignancy on ultrasound (confirmed histologically) not suspected on m a m m o g r a p h y .
Five patients in the abnormal m a m m o g r a m , abnormal ultrasound group had benign disease on m a m m o g r a m upgraded to probably malignant by ultrasound (confirmed histologically). In the abnormal m a m m o g r a m , normal ultrasound group 12 had benign disease (three with calcifications only) and one had malignant microcalcifications confirmed on histology. In conclusion ultrasound has provided useful additional information in a large number o f symptomatic breast patients. Of particular interest are the four patients with malignancy seen on ultrasound alone.
C O M P U T E R I N T E R P R E T A T I O N OF D I G I T A L MAMMOGRAMS D. H. DAVIES and D. R. D A N C E
Department of Physics, Institute of Cancer Research, The Royal Marsden Hospital, London Analysis of the large n u m b e r of images produced by the U K breast screening programme m a y be enhanced if an automatic computer system could be used to pre-read the m a m m o g r a m s . Two preliminary studies into the automatic detection of calcifications have been conducted. In the first study 75 randomly selected conventional screen film m a m m o g r a m s were digitized at 100 microns and 8 bits. These were divided into training and test sets of 25 and 50 fihns respectively. The results for the test set of 50 complete clinical m a m m o g r a m s show that the computer system achieves a 25/25 true positive film classification with false positive clusters detected in 4/50 films. There were no false negative film classifications. In the second study a small n u m b e r of m a m m o g r a m s of radiographically very dense breasts containing examples of small (<0.014 m m 2) subtle (optical density differences < 0.035) calcifications were selected. These films were digitized using a high resolution (43 micron), high quality (12 bit) laser scanner and were selected to challenge the algorithm developed in the initial study. This is the first time that the detection of small subtle calcifications has been able to be investigated. Improvements were made to the algorithm used in the first study, but our results show that the automatic detection rates for these small subtle calcifications have yet to reach an acceptable level.
SELF A S S E S S M E N T O F F I L M READING P E R F O R M A N C E 1N BREAST S C R E E N I N G A. G. G A L E , J. A. T O W L E , E. J. R O E B U C K * and A. R. M. W I L S O N *
Department of Academic Radiology and Breast Screening Training Centre and *University Hospital and City Hospital, Nottingham A national self-assessment programme of radiologists' film reading performance has been established by the NHSBSP and the Royal College of Radiologists. The purpose of this is to enable each radiologist involved in screening to learn how they interpret sets of known cases. Following piloting of the methodology a three region trial has been carried out. A set of exemplary cases encompassing a diverse range of screening appearances has been amassed from screening training centres and a gold standard of radiological interpretation obtained. Individually each screening radiologist in the three regions has reviewed these cases and measures of their film reading performance recorded. Performance indices monitored include: the number of cases recalled for assessment; sensitivity and specificity; identification of key radiological features; and R O C measures of detection and classification ability. Furthermore the data analyses allow identification of areas where an individual has particular difficulty which then permits the suggestion of appropriate revision strategies. It is proposed that this self-assessment programme will be extended nationally and will become an annual exercise so that every radiologist can follow the progression of their performance. This paper details the results of the trial phase which encompasses the Trent, East Anglian and Northern regions.
M A M M O G R A M S AT D I F F E R E N T C O M P R E S S I O N S F O R T H E D E T E C T I O N OF BREAST C A N C E R R. P. H I G H N A M , J. M. B R A D Y * and B. SHEPSTONE~
Computing Laboratory and *Department of Engineering, Oxford University, and ~fBreast Care Unit, Churchill Hospital, Oxford Mammographic signs are sometimes non-specific if not ambiguous. It is proposed that by taking two m a m m o g r a m s of the same breast at different compressions important new information becomes available.