CNN-based Whole Breast Segmentation in Longitudinal High-risk MRI Study

Ambroladze, Ani; Hahn, Horst K.; Amer, Heba; Ingrisch, Michael; Gerken, Annika; Wenzel, Markus; Püsken, Michael; Mittermeier, Andreas; Engel, Christoph; Schmutzler, Rita; Fallenberg, Eva M.

doi:10.1007/978-3-658-41657-7_35

Cited by 1 publication

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The masks are computed by a deep learningbased breast segmentation model that uses a U-Net architecture and was developed in previous work. 10 We switched from the sheetness-based segmentation to a deep learning-based one since it failed to produce a useful breast mask in some cases and had long and unpredictable running times in other cases. To crop the images, a bounding box around the breast mask is computed, and a margin of 21 voxels in a plane and 7 voxels in a slice direction are added to prevent cutting out a too-tight image region.…”

Section: Breast Segmentation and Croppingmentioning

confidence: 99%

Deformable current-prior registration of DCE breast MR images on multi-site data

Geißler,

Ambroladze,

Papenberg

et al. 2024

Medical Imaging 2024: Image Processing

View full text Add to dashboard Cite

Recent studies indicate that malignant breast lesions can be predicted from structural changes in prior exams of preventive breast MRI examinations. Due to non-rigid deformation between studies, spatial correspondences between structures in two consecutive studies are lost. Thus, deformable image registration can contribute to predicting individual cancer risks. This study evaluates a registration approach based on a novel breast mask segmentation and non-linear image registration based on data from 5 different sites. The landmark error (mean ± standard deviation [1 st quartile, 3 rd quartile]), annotated by three radiologists, is 2.9 ± 2.8 [1.3, 3.2] mm when leaving out two outlier cases from the evaluation for which the registration failed completely. We assess the inter-observer variabilities of keypoint errors and find an error of 3.6 ± 4.7 [1.6, 4.0] mm, 4.4 ± 4.9 [1.8, 4.8] mm, and 3.8 ± 4.0 [1.7, 4.1] mm when comparing each radiologist to the mean keypoints of the other two radiologists. Our study shows that the current state of the art in registration is well suited to recover spatial correspondences of structures in cancerous and non-cancerous cases, despite the high level of difficulty of this task.

show abstract

Section: Breast Segmentation and Croppingmentioning

confidence: 99%