Assessment of Enhancement Kinetics Improves the Specificity of Abbreviated Breast MRI: Performance in an Enriched Cohort.
Haejung KimEun Young KoKa Eun KimMyoung Kyoung KimJi Soo ChoiEun Sook KoBoo-Kyung HanPublished in: Diagnostics (Basel, Switzerland) (2022)
Objective: To investigate the added value of kinetic information for breast lesion evaluation on abbreviated breast MRI (AB-MRI). Methods: This retrospective study analyzed 207 breast lesions with Breast Imaging Reporting and Data System categories 3, 4, or 5 on AB-MRI in 198 consecutive patients who had breast MRI for screening after breast cancer surgery between January 2017 and December 2019. All lesions were pathologically confirmed or stable on follow-up images for 2 years or more. Kinetic information of the lesions regarding the degree and rate of enhancement on the first post-contrast-enhanced image and the enhancement curve type from two post-contrast-enhanced images were analyzed on a commercially available computer-assisted diagnosis system. The diagnostic performances of AB-MRI with morphological analysis alone and with the addition of kinetic information were compared using the McNemar test. Results: Of 207 lesions, 59 (28.5%) were malignant and 148 (71.5%) were benign. The addition of an enhancement degree of ≥90% to the morphological analysis significantly increased the specificity of AB-MRI (29.7% vs. 52.7%, p < 0.001) without significantly reducing the sensitivity (94.9% vs. 89.8%, p = 0.083) compared to morphological analysis alone. Unnecessary biopsy could have been avoided in 34 benign lesions, although three malignant lesions could have been missed. For detecting invasive cancer, adding an enhancement degree ≥107% to the morphological analysis significantly increased the specificity (26.5% vs. 57.6%, p < 0.001) without significantly decreasing the sensitivity (94.6% vs. 86.5%, p = 0.083). Conclusion: Adding the degree of enhancement on the first post-contrast-enhanced image to the morphological analysis resulted in higher AB-MRI specificity without compromising its sensitivity.
Keyphrases
- young adults
- contrast enhanced
- magnetic resonance imaging
- diffusion weighted
- diffusion weighted imaging
- magnetic resonance
- computed tomography
- deep learning
- minimally invasive
- emergency department
- squamous cell carcinoma
- coronary artery disease
- machine learning
- social media
- convolutional neural network
- coronary artery bypass
- percutaneous coronary intervention
- ultrasound guided
- papillary thyroid