Rosalind Given-Wilson

Observer variability in cancer detection during routine repeat (incident) mammographic screening in a study of two versus one view mammography.

Objective To examine the reasons for observer variability of cancer detection using one and two view mammography at incident (subsequent) screening and determine whether false negative results (non-recall of a cancer) are due to failure to detect the associated features(s) of the cancer on the mammogram, or misinterpretation of the observed feature(s) as not indicative of malignancy. Setting A random selection of cancers (invasive and in situ) seen as incident cases during the second screening round (January 1994–January 1997) in the South West London Breast Screening Service were used. This service uses two view mammography and double reading with arbitration by a third or further readers for all screens. Methods Mammograms of cases were mixed with those of controls in a 1:2 ratio in two test sets. Eleven experienced film readers, each reading both test sets, took part in the study. Initially the oblique view only was read, then, additionally, the craniocaudal view. Previous films were available to the readers. Data on abnormalities noted on the films and probability of recall were recorded and analysed. Results 387 valid readings of 36 cancers (30 invasive and six ductal carcinoma in situ) were made by 11 readers. The overall sensitivity increased from 79% with one view to 85% with two views. For invasive cancers <10 mm the sensitivity was 71% with one view, but increased to 85% with two views. Recall of individual cancers by the readers varied substantially. With one view 15 (50%) of the 30 invasive cancers were recalled by all 11 readers, increasing to 18 (60%) with two views. Of the invasive cancers not recalled by all 11 readers, there was considerable disagreement, particularly for the smaller cancers.With one view 69% of invasive cancers <10 mm were correctly marked on the proforma compared with 87% with two views. Invasive cancers >10 mm were almost all marked on the proforma with one or two views. For invasive cancers, the misinterpreted feature that did not lead to recall was most commonly an asymmetry (42%), whereas for in situ cancers it was calcifications (67%). The finding of an irregular mass was the least misinterpreted feature. Conclusion The study showed that of those invasive cancers detected at routine repeat screening by a programme using two view mammography and double reading with arbitration, at least 50% could be described as “difficult” (for example, “minimal” signs) to recall using the single reading of one view, even under “favourable” study conditions with two normal subjects per case.The finding that at least 87% of invasive cancers <10 mm are detected (marked on the proforma) with two views, but only 69% with the one view, suggests that for single reading of mammograms with one view the detection of small invasive cancers is a major problem. This problem is helped by the second view. For invasive cancers ≥10 mm, interpretation (benign or malignant) rather than detection (under these study conditions) was the major cause of recall failure. The most common signs to be misinterpreted were calcifications and asymmetries; once visualised an irregular mass was least likely to be misinterpreted.This study provides evidence that detection and interpretation of most invasive cancers is improved by increasing the number of views, and by increasing the number of readers.

Effect of image quality on calcification detection in digital mammography.

PURPOSE This study aims to investigate if microcalcification detection varies significantly when mammographic images are acquired using different image qualities, including: different detectors, dose levels, and different image processing algorithms. An additional aim was to determine how the standard European method of measuring image quality using threshold gold thickness measured with a CDMAM phantom and the associated limits in current EU guidelines relate to calcification detection. METHODS One hundred and sixty two normal breast images were acquired on an amorphous selenium direct digital (DR) system. Microcalcification clusters extracted from magnified images of slices of mastectomies were electronically inserted into half of the images. The calcification clusters had a subtle appearance. All images were adjusted using a validated mathematical method to simulate the appearance of images from a computed radiography (CR) imaging system at the same dose, from both systems at half this dose, and from the DR system at quarter this dose. The original 162 images were processed with both Hologic and Agfa (Musica-2) image processing. All other image qualities were processed with Agfa (Musica-2) image processing only. Seven experienced observers marked and rated any identified suspicious regions. Free response operating characteristic (FROC) and ROC analyses were performed on the data. The lesion sensitivity at a nonlesion localization fraction (NLF) of 0.1 was also calculated. Images of the CDMAM mammographic test phantom were acquired using the automatic setting on the DR system. These images were modified to the additional image qualities used in the observer study. The images were analyzed using automated software. In order to assess the relationship between threshold gold thickness and calcification detection a power law was fitted to the data. RESULTS There was a significant reduction in calcification detection using CR compared with DR: the alternative FROC (AFROC) area decreased from 0.84 to 0.63 and the ROC area decreased from 0.91 to 0.79 (p < 0.0001). This corresponded to a 30% drop in lesion sensitivity at a NLF equal to 0.1. Detection was also sensitive to the dose used. There was no significant difference in detection between the two image processing algorithms used (p > 0.05). It was additionally found that lower threshold gold thickness from CDMAM analysis implied better cluster detection. The measured threshold gold thickness passed the acceptable limit set in the EU standards for all image qualities except half dose CR. However, calcification detection varied significantly between image qualities. This suggests that the current EU guidelines may need revising. CONCLUSIONS Microcalcification detection was found to be sensitive to detector and dose used. Standard measurements of image quality were a good predictor of microcalcification cluster detection.

Do General Practitioners Influence the Uptake of Breast Cancer Screening

Objectives — To investigate the relative importance of patient and general practice characteristics in explaining variations between practices in the uptake of breast cancer screening. Design — Ecological study examining variations in breast cancer screening rates among 131 general practices using routine data. Setting — Merton, Sutton, and Wandsworth Family Health Services Authority, which covers parts of inner and outer London. Main outcome measure — Percentage of eligible women aged 50–64 who attended for mammography during the first round of screening for breast cancer (1991–1994). Results — Of the 43 063 women eligible for breast cancer screening, 25 826 (60%) attended for a mammogram. Breast cancer screening rates in individual practices varied from 12·5% to 84·5%. The estimated percentage list inflation for the practices was the variable most highly correlated with screening rates (r= −0·69). There were also strong negative correlations between screening rates and variables associated with social deprivation, such as the estimated percentage of the practice population living in households without a car (r= −0·61), and with variables that measured the ethnic make-up of practice populations, such as the estimated percentage of people in non-white ethnic groups (r= −0·60). Screening rates were significantly higher in practices with a computer than in those without (59·5% v 53·9%, difference 5·6%, 95% confidence interval 1·1 to 10·2%). There was no significant difference in screening rates between practices with and without a female partner; with and without a practice nurse; and with and without a practice manager. In a forward stepwise multiple regression model that explained 58% of the variation in breast cancer screening rates, four factors were significant independent predictors (at P = 0·05) of screening rates: list inflation and people living in households without a car were both negative predictors of screening rates, and chronic illness and the number of partners in a practice were both positive predictors of screening rates. The practice with the highest screening rate (84·5%) contacted all women invited for screening to encourage them to attend for their mammogram and achieved a rate 38% higher than predicted from the regression model. Breast cancer screening rates were on average lower than cervical cancer screening rates (mean difference 14·5%, standard deviation 12·0%) and were less strongly associated with practice characteristics. Conclusions — The strong negative correlation between breast cancer screening rates and list inflation shows the importance of accurate age-sex registers in achieving high breast cancer screening rates. Breast cancer screening units, family health services authorities, and general practitioners need to collaborate to improve the accuracy of the age-sex registers used to generate invitations for breast cancer screening. The success of the practice with the highest screening rate suggests that practices can influence the uptake of breast cancer screening among their patients. Giving general practitioners a greater role in breast cancer screening, either by offering them financial incentives or by giving them clerical support to check prior notification lists and contact non-attenders, may also help to increase breast cancer screening rates.

The number of women who would need to be screened regularly by mammography to prevent one death from breast cancer

The number of women who would need to be screened regularly by mammography to prevent one death from breast cancer depends strongly on several factors, including the age at which regular screening starts, the period over which it continues, and the duration of follow-up after screening. Furthermore, more women would need to be INVITED for screening than would need to be SCREENED to prevent one death, since not all women invited attend for screening or are screened regularly. Failure to consider these important factors accounts for many of the major discrepancies between different published estimates. The randomised evidence indicates that, in high income countries, around one breast cancer death would be prevented in the long term for every 400 women aged 50–70 years regularly screened over a ten-year period.

Reproducibility of Prompts in Computer-aided Detection (CAD) of Breast Cancer

AIM We evaluated the reproducibility of prompts using the R2 ImageChecker M2000 computer-aided detection (CAD) system. MATERIALS AND METHODS Forty selected two-view mammograms of women with breast cancer were digitized and analysed using the ImageChecker on 10 separate occasions. The mammograms were chosen to provide both straightforward and subtle signs of malignancy. Data analysed included mammographic abnormality, pathology, and whether the cancer was prompted or given an emphasized prompt. RESULTS Correct prompts were generated in 86 out of 100 occasions for screen-detected cancers. Reproducibility was less in the other categories of more subtle cancers: 21% for cancers previously missed by CAD, a group that contained more grade 1 and small (<10 mm) tumours. Prompts for calcifications were more reproducible than those for masses (76% versus 53%) and these cancers were more likely to have an emphasized prompt. CONCLUSIONS Probably the most important cause of variability of prompts is shifts in film position between sequential digitizations. Consequently subtle lesions that are only just above the threshold for display may not be prompted on repeat scanning. However, users of CAD should be aware that even emphasized prompts are not consistently reproducible.

The Use of Ultrasound for Monitoring Breast Tumour Response to Pro-adjuvant Therapy

In a prospective study, use of serial ultrasound (US) for monitoring tumour response to pro-adjuvant chemotherapy was assessed in 16 patients. Comparison was made with mammographic and pathological tumour size measurements. Clinical and radiological response to treatment was assessed using UICC (International Union Against Cancer) criteria. Comparison of clinical and US response to treatment showed some agreement in 60% and disagreement in 40%. This was comparable with clinical versus mammographic responses (55% and 45%). Correlation between calliper and pathological measurement was similar to that between US and pathological measurement (r = 0.51, P = 0.05; r = 0.50, P < 0.05). Mammography showed poorer correlation (NS). For assessment of final tumour size, US clinical measurements were comparable and better than mammography. US may be a useful tool in monitoring the response of breast tumours to pro-adjuvant therapy.

Efficiency of cancer detection during routine repeat (incident) mammographic screening: two versus one view mammography

Objective To examine the influence of one view versus two view mammography on cancer detection and recall for further investigation of women attending incident (subsequent) screening. Setting All cancers (invasive and in situ) detected as incident cases during the second screening round (January 1994 to January 1997) at the South West London Breast Screening Service were used. This service uses two view mammography and double reading, with arbitration by a third or further readers for all screens. Methods Mammograms of cases were mixed with those of controls in a 1:2 ratio in nine test sets; each set was read independently by three film readers. Fourteen readers, each reading from one to four test sets, took part in the study. Initially, the oblique view only was read, then the craniocaudal view was read in addition. Previous films were available to the readers. Data on abnormalities noted on the films and probability of recall were recorded and analysed. Results 10 of the 14 readers obtained increased sensitivity using two views (p=0.04), for two readers there was no difference, and for two readers sensitivity decreased. The mean sensitivity increase was 6.1% (p=0.01). The overall increase in sensitivity from all readings of invasive cancers was 8.9%, with no increase seen for in situ cancers. 11 of the 14 readers obtained an increase in specificity (p=0.006), two readers showed no increase, and the specificity for one reader was decreased. The mean increase in specificity using two views was 5.7% (p=0.006). Conclusion This study showed an increase of 8.9% in sensitivity for the detection of invasive cancers when two views are used at incident screening, with a ratio of two control mammograms for every case. This is equivalent to a sample from population screening with a cancer detection rate of 333 per 1000. Such a study is considered to be likely to underestimate the benefit of two views in screening under non-test conditions where the cancer detection rate is of the order of five per 1000. The use of two view mammography for the detection of in situ cancers showed no increased benefit. A randomised controlled trial is needed to obtain a reliable estimate of the increase in cancer detection rate for incident screening in normal populations.

Ductal carcinoma in situ presenting as a stellate lesion on mammography

Ductal carcinoma in situ (DCIS) usually manifests as microcalcification on mammography, but several other unusual forms of presentation on mammography are also described. One such atypical appearance is the stellate mass without calcification. This may occur with DCIS alone or when DCIS is associated with a complex sclerosing lesion (CSL). We retrospectively analysed the histopathological specimens and mammograms of women who were referred for biopsy from two large breast screening programmes, and were found to have DCIS as the dominant histological lesion. Of 677 women referred for surgical biopsy, 86 (13%) showed histological evidence of DCIS as the predominant lesion, and of these, seven (8%) showed a stellate appearance on mammography without associated calcification. In three cases the mammographic appearance was due to DCIS alone, and four were due to a CSL with associated DCIS. Only one case showed microinvasion (< 1 mm), and this was not large enough to account for the stellate lesion. We advocate biopsy of all radial lesions which are not surgical scars, as malignancy associated with benign lesions such as CSLs could be missed by cytological sampling errors.

CAD in mammography: lesion-level versus case-level analysis of the effects of prompts on human decisions

ObjectTo understand decision processes in CAD-supported breast screening by analysing how prompts affect readers’ judgements of individual mammographic features (lesions). To this end we analysed hitherto unexamined details of reports completed by mammogram readers in an earlier evaluation of a CAD tool.Material and methodsAssessments of lesions were extracted from 5,839 reports for 59 cancer cases. Statistical analyses of these data focused on what features readers considered when recalling a cancer case and how readers reacted to CAD prompts.ResultsAbout 13.5% of recall decisions were found to be caused by responses to features other than those indicating actual cancer. Effects of CAD: lesions were more likely to be examined if prompted; the presence of a prompt on a cancer increased the probability of both detection and recall especially for less accurate readers in subtler cases; lack of prompts made cancer features less likely to be detected; false prompts made non-cancer features more likely to be classified as cancer.ConclusionThe apparent lack of impact reported for CAD in some studies is plausibly due to CAD systematically affecting readers’ identification of individual features, in a beneficial way for certain combinations of readers and features and a damaging way for others. Mammogram readers do not ignore prompts. Methodologically, assessing CAD by numbers of recalled cancer cases may be misleading.

The Effect of Image Processing on the Detection of Cancers in Digital Mammography

OBJECTIVE. The objective of our study was to investigate the effect of image processing on the detection of cancers in digital mammography images. MATERIALS AND METHODS. Two hundred seventy pairs of breast images (both breasts, one view) were collected from eight systems using Hologic amorphous selenium detectors: 80 image pairs showed breasts containing subtle malignant masses; 30 image pairs, biopsy-proven benign lesions; 80 image pairs, simulated calcification clusters; and 80 image pairs, no cancer (normal). The 270 image pairs were processed with three types of image processing: standard (full enhancement), low contrast (intermediate enhancement), and pseudo-film-screen (no enhancement). Seven experienced observers inspected the images, locating and rating regions they suspected to be cancer for likelihood of malignancy. The results were analyzed using a jackknife-alternative free-response receiver operating characteristic (JAFROC) analysis. RESULTS. The detection of calcification clusters was significantly affected by the type of image processing: The JAFROC figure of merit (FOM) decreased from 0.65 with standard image processing to 0.63 with low-contrast image processing (p = 0.04) and from 0.65 with standard image processing to 0.61 with film-screen image processing (p = 0.0005). The detection of noncalcification cancers was not significantly different among the image-processing types investigated (p > 0.40). CONCLUSION. These results suggest that image processing has a significant impact on the detection of calcification clusters in digital mammography. For the three image-processing versions and the system investigated, standard image processing was optimal for the detection of calcification clusters. The effect on cancer detection should be considered when selecting the type of image processing in the future.