Application of the Kaiser score to increase diagnostic accuracy in equivocal lesions on diagnostic mammograms referred for MR mammography
03 medical and health sciences
0302 clinical medicine
ROC Curve
Area Under Curve
Humans
Breast Neoplasms
Magnetic Resonance Imaging
Sensitivity and Specificity
Mammography
Retrospective Studies
DOI:
10.1016/j.ejrad.2020.109413
Publication Date:
2020-11-30T07:06:10Z
AUTHORS (10)
ABSTRACT
We aimed to interpret MR mammography (MRM) using the Kaiser scores for equivocal or inconclusive lesions on mammography (MG).Retrospective IRB-approved evaluation of 3623 MG for which MRM was deployed as a problem-solving tool, after inclusion-exclusion criteria were met. Three readers with different levels of experience assigned a final score from 1 to 11 based on the previously established tree classification system. Area under the curve (AUC) derived from receiver operating characteristic (ROC) analysis was used to determine the overall diagnostic performance for all lesions and separately for mass and non-mass enhancement. Sensitivity, specificity, and likelihood ratio values were obtained at different cut-off values of >4, > 5, and > 8 to rule in and rule out malignancy.Histopathology of 183 mass and 133 non-mass enhancement (NME) lesions show benign etiology in 95 and malignant in 221. The AUC was 0.796 [0.851 for mass and 0.715 for NME]. Applying the Kaiser score upgraded 202 lesions with correct prediction in 77 %, and downgraded 28 lesions with correct prediction in 60.8 %. Using a score <5 instead of <4 to rule out malignancy improved our diagnostic ability to correctly identify 100 % benign lesions. Applying Kaiser score correctly downgraded 60.8 % (17/28) lesions; thus avoiding biopsies in these. Using a high cut-off value>8 to rule-in malignancy, we correctly identified 59.7 % of lesions with 80 % specificity and positive likelihood ratio of 3.The Kaiser score has clinical translation benefits when used as a problem-solving tool for inconclusive MG findings.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (36)
CITATIONS (32)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....