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

Mammography interpretation is challenging with high error rates. This study aims to reduce the errors in mammography reading by mapping diagnostic errors against global mammographic characteristics using a radiomics-based machine learning approach. A total of 36 radiologists from cohort A (n = 20) and cohort B (n = 16) read 60 high-density mammographic cases. Radiomic features were extracted from three regions of interest (ROIs), and random forest models were trained to predict diagnostic errors for each cohort. Performance was evaluated using sensitivity, specificity, accuracy, and AUC. The impact of ROI placement and normalization on prediction was investigated. Our approach successfully predicted both the false positive and false negative errors of both cohorts but did not consistently predict location errors. The errors produced by radiologists from cohort B were less predictable compared to those in cohort A. The performance of the models did not show significant improvement after feature normalization, despite the mammograms being produced by different vendors. Our novel radiomics-based machine learning pipeline focusing on global radiomic features could predict false positive and false negative errors. The proposed method can be used to develop group-tailored mammographic educational strategies to help improve future mammography reader performance.

[1]  P. Brennan,et al.  Test-set training improves the detection rates of invasive cancer in screening mammography. , 2022, Clinical radiology.

[2]  C. McCollough,et al.  Individualized and generalized models for predicting observer performance on liver metastasis detection using CT , 2022, Journal of medical imaging.

[3]  P. Brennan,et al.  Varying performance levels for diagnosing mammographic images depending on reader nationality have AI and educational implications , 2022, Medical Imaging.

[4]  P. Brennan,et al.  A machine learning model based on readers’ characteristics to predict their performances in reading screening mammograms , 2022, Breast Cancer.

[5]  P. Brennan,et al.  Differences in lesion interpretation between radiologists in two countries: Lessons from a digital breast tomosynthesis training test set , 2021, Asia-Pacific journal of clinical oncology.

[6]  P. Brennan,et al.  Mammography-based Radiomics in Breast Cancer: A Scoping Review of Current Knowledge and Future Needs. , 2021, Academic radiology.

[7]  H. Frazer,et al.  Clinical performance progress of BREAST participants: the impact of test-set participation. , 2021, Clinical radiology.

[8]  J. Wolfe,et al.  Global processing provides malignancy evidence complementary to the information captured by humans or machines following detailed mammogram inspection , 2021, Scientific Reports.

[9]  P. Brennan,et al.  Improving radiologist's ability in identifying particular abnormal lesions on mammograms through training test set with immediate feedback , 2021, Scientific Reports.

[10]  H. Alkadhi,et al.  Radiomics in medical imaging—“how-to” guide and critical reflection , 2020, Insights into Imaging.

[11]  Lonie R. Salkowski,et al.  Artificial Intelligence and Machine Learning in Radiology Education Is Ready for Prime Time. , 2020, Journal of the American College of Radiology : JACR.

[12]  K. Regmi,et al.  A systematic review of the factors – enablers and barriers – affecting e-learning in health sciences education , 2020, BMC medical education.

[13]  P. Brennan,et al.  The roles of clinical audit and test sets in promoting the quality of breast screening: a scoping review. , 2020, Clinical radiology.

[14]  W. Cai,et al.  Predicting the response to neoadjuvant chemotherapy for breast cancer: wavelet transforming radiomics in MRI , 2020, BMC Cancer.

[15]  Till Bärnighausen,et al.  Evaluation of e-learning for medical education in low- and middle-income countries: A systematic review , 2020, Comput. Educ..

[16]  Ziba Gandomkar,et al.  Radiologists can detect the ‘gist’ of breast cancer before any overt signs of cancer appear , 2018, Scientific Reports.

[17]  P. Brennan,et al.  Errors in Mammography Cannot be Solved Through Technology Alone , 2018, Asian Pacific journal of cancer prevention : APJCP.

[18]  S. Lewis,et al.  An investigation into the mammographic appearances of missed breast cancers when recall rates are reduced. , 2017, The British journal of radiology.

[19]  Warwick B. Lee,et al.  Impact of Breast Reader Assessment Strategy on mammographic radiologists' test reading performance , 2016, Journal of medical imaging and radiation oncology.

[20]  Paul Kinahan,et al.  Radiomics: Images Are More than Pictures, They Are Data , 2015, Radiology.

[21]  Yuanjie Zheng,et al.  Parenchymal texture analysis in digital mammography: A fully automated pipeline for breast cancer risk assessment. , 2015, Medical physics.

[22]  Maciej A Mazurowski,et al.  Using computer-extracted image features for modeling of error-making patterns in detection of mammographic masses among radiology residents. , 2014, Medical physics.

[23]  B. Keller,et al.  Estimation of breast percent density in raw and processed full field digital mammography images via adaptive fuzzy c-means clustering and support vector machine segmentation. , 2012, Medical physics.

[24]  Karla K. Evans,et al.  The gist of the abnormal: Above-chance medical decision making in the blink of an eye , 2013, Psychonomic Bulletin & Review.

[25]  Berkman Sahiner,et al.  Association of computerized mammographic parenchymal pattern measure with breast cancer risk: a pilot case-control study. , 2011, Radiology.

[26]  Huiman X Barnhart,et al.  Individualized computer-aided education in mammography based on user modeling: concept and preliminary experiments. , 2010, Medical physics.

[27]  Michael J. Carston,et al.  Texture Features from Mammographic Images and Risk of Breast Cancer , 2009, Cancer Epidemiology Biomarkers & Prevention.

[28]  Mongi A. Abidi,et al.  Evaluation of sharpness measures and search algorithms for the auto focusing of high-magnification images , 2006, SPIE Defense + Commercial Sensing.

[29]  Claudia Mello-Thoms,et al.  The perception of breast cancer: what differentiates missed from reported cancers in mammography? , 2002, Academic radiology.

[30]  Yung-Chang Chen,et al.  Statistical feature matrix for texture analysis , 1992, CVGIP Graph. Model. Image Process..

[31]  D. Sagi,et al.  Gabor filters as texture discriminator , 1989, Biological Cybernetics.

[32]  Azriel Rosenfeld,et al.  A Comparative Study of Texture Measures for Terrain Classification , 1975, IEEE Transactions on Systems, Man, and Cybernetics.

[33]  Mary M. Galloway,et al.  Texture analysis using gray level run lengths , 1974 .

[34]  Robert M. Haralick,et al.  Textural Features for Image Classification , 1973, IEEE Trans. Syst. Man Cybern..

[35]  H. Yamashita,et al.  Standardization of imaging features for radiomics analysis. , 2019, The journal of medical investigation : JMI.

[36]  Anna Bilyatdinova,et al.  Artificial Intelligence trends in education: a narrative overview , 2018 .

[37]  Yung-Chang Chen,et al.  Texture features for classification of ultrasonic liver images , 1992, IEEE Trans. Medical Imaging.

[38]  Robert King,et al.  Textural features corresponding to textural properties , 1989, IEEE Trans. Syst. Man Cybern..