TY  - JOUR
AU  - Wald, Tassilo
AU  - Hamm, Benjamin
AU  - Holzschuh, Julius
AU  - El Shafie, Rami
AU  - Kudak, Andreas
AU  - Kovacs, Balint
AU  - Pflüger, Irada
AU  - von Nettelbladt, Bastian
AU  - Ulrich, Constantin
AU  - Baumgartner, Michael
AU  - Vollmuth, Philipp
AU  - Debus, Jürgen
AU  - Maier-Hein, Klaus
AU  - Welzel, Thomas
TI  - Enhancing deep learning methods for brain metastasis detection through cross-technique annotations on SPACE MRI.
JO  - European radiology experimental
VL  - 9
IS  - 1
SN  - 2509-9280
CY  - [Cham]
PB  - Springer International Publishing
M1  - DKFZ-2025-00303
SP  - 15
PY  - 2025
N1  - #EA:E230#
AB  - Gadolinium-enhanced 'sampling perfection with application-optimized contrasts using different flip angle evolution' (SPACE) sequence allows better visualization of brain metastases (BMs) compared to 'magnetization-prepared rapid acquisition gradient echo' (MPRAGE). We hypothesize that this better conspicuity leads to high-quality annotation (HAQ), enhancing deep learning (DL) algorithm detection of BMs on MPRAGE images.Retrospective contrast-enhanced (gadobutrol 0.1 mmol/kg) SPACE and MPRAGE data of 157 patients with BM were used, either annotated on MPRAGE resulting in normal annotation quality (NAQ) or on coregistered SPACE resulting in HAQ. Multiple DL methods were developed with NAQ or HAQ using either SPACE or MRPAGE images and evaluated on their detection performance using positive predictive value (PPV), sensitivity, and F1 score and on their delineation performance using volumetric Dice similarity coefficient, PPV, and sensitivity on one internal and four additional test datasets (660 patients).The SPACE-HAQ model reached 0.978 PPV, 0.882 sensitivity, and 0.916 F1-score. The MPRAGE-HAQ reached 0.867, 0.839, and 0.840, the MPRAGE NAQ 0.964, 0.667, and 0.798, respectively (p ≥ 0.157). Relative to MPRAGE-NAQ, the MPRAGE-HAQ F1-score detection increased on all additional test datasets by 2.5-9.6 points (p < 0.016) and sensitivity improved on three datasets by 4.6-8.5 points (p < 0.001). Moreover, volumetric instance sensitivity improved by 3.6-7.6 points (p < 0.001).HAQ improves DL methods without specialized imaging during application time. HAQ alone achieves about 40
KW  - Humans
KW  - Brain Neoplasms: diagnostic imaging
KW  - Brain Neoplasms: secondary
KW  - Deep Learning
KW  - Magnetic Resonance Imaging: methods
KW  - Retrospective Studies
KW  - Male
KW  - Female
KW  - Middle Aged
KW  - Contrast Media
KW  - Aged
KW  - Organometallic Compounds
KW  - Adult
KW  - Brain neoplasms (Other)
KW  - Deep learning (Other)
KW  - Image interpretation (computer-assisted) (Other)
KW  - Image processing (computer-assisted) (Other)
KW  - Magnetic resonance imaging (Other)
KW  - Contrast Media (NLM Chemicals)
KW  - gadobutrol (NLM Chemicals)
KW  - Organometallic Compounds (NLM Chemicals)
LB  - PUB:(DE-HGF)16
C6  - pmid:39913077
DO  - DOI:10.1186/s41747-025-00554-5
UR  - https://inrepo02.dkfz.de/record/298593
ER  -