An investigation into the mammographic appearances of missed breast cancers when recall rates are reduced

Abstract

Cancers with a stellate mass were more easily detected and were more likely to continue to be recalled, even at lower recall rates. Cancers with non-specific density and mixed features were most likely to be missed at reduced recall rates. Advances in knowledge: Internationally, recall rates vary within screening mammography programs considerably, with a range between 1% and 15%, and very little is known about the type of breast cancer appearances found when radiologists interpret screening mammograms at these various recall rates. Therefore, understanding the lesion types and the mammographic appearances of breast cancers that are affected by readers' recall decisions should be investigated.

Characterization of invasive breast cancer lesions in breast x-ray imaging: a reference dataset for virtual imaging trials

“…Regarding the BI-RADS characterization of the lesions, we found that poorly visible lesions show a larger variation in mass margins compared to well visible mass lesions, which were mostly irregular shaped and spiculated. These findings correlate with the study of Mohd Norsuddin et al (4) which indicated that cancers with non-specific density and mixed features were most likely missed at reduced recall rates. Nevertheless, detailed BI-RADS imaging descriptors of the lesions are lacking.…”

Using Radiomics-Based Machine Learning to Create Targeted Test Sets to Improve Specific Mammography Reader Cohort Performance: A Feasibility Study

“…This finding was consistent with previous research showing variability in the type of breast lesions most likely to be missed among radiologists across countries [ 39 ]. For example, one study found that spiculated stellate lesions were more likely to be neglected by Vietnamese radiologists (31); however, non-specific density was difficult to detect by Australian radiologists [ 40 ]. The reasons for these variations are multifaceted and may be due to different focuses in mammographic training in different countries and/or radiologists’ various experience with certain types of mammographic densities or features [ 15 ].…”

CNN-based transfer learning with 10-fold cross-validation: a novel approach for customized education of mammography training

“…False negative errors can lead to missed diagnoses, while false positive errors result in unnecessary costs and potential patient morbidity. The interpretation of screening mammograms is susceptible to high error rates, and certain errors are region-specific due to readers' varying experience to particular types of disease appearances presented on images and different mammography training programs received [3][4][5]. For instance, Tao et al conducted a study that revealed a significant variation in diagnostic agreement between Chinese and Australian radiologists when interpreting the same mammography cases.…”

“…In contrast, there was greater agreement observed within the groups of radiologists from within China [3]. Similarly, Mohd Norsuddin et al demonstrated that Australian radiologists faced challenges in detecting breast lesions with non-specific density, while Trieu et al found Vietnamese radiologists tended to miss spiculated stellate masses more easily [4,5]. These findings underscore the importance of identifying common errors within specific radiology communities and tailoring mammography training to address their improvement needs.…”