TY - GEN
T1 - Novel Structural-Scale Uncertainty Measures and Error Retention Curves
T2 - 20th IEEE International Symposium on Biomedical Imaging, ISBI 2023
AU - Molchanova, Nataliia
AU - Raina, Vatsal
AU - Malinin, Andrey
AU - La Rosa, Francesco
AU - Muller, Henning
AU - Gales, Mark
AU - Granziera, Cristina
AU - Graziani, Mara
AU - Cuadra, Meritxell Bach
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - This paper focuses on the uncertainty estimation for white matter lesions (WML) segmentation in magnetic resonance imaging (MRI). On one side, voxel-scale segmentation errors cause the erroneous delineation of the lesions; on the other side, lesion-scale detection errors lead to wrong lesion counts. Both of these factors are clinically relevant for the assessment of multiple sclerosis patients. This work aims to compare the ability of different voxel- and lesion-scale uncertainty measures to capture errors related to segmentation and lesion detection, respectively. Our main contributions are (i) proposing new measures of lesion-scale uncertainty that do not utilise voxel-scale uncertainties; (ii) extending an error retention curves analysis framework for the evaluation of lesion-scale uncertainty measures. Our results obtained on the multi-center testing set of 58 patients demonstrate that the proposed lesion-scale measure achieves the best performance among the analysed measures. All code implementations are provided at https://github.com/Medical-Image-Analysis-Laboratory/MS_WML_uncs.
AB - This paper focuses on the uncertainty estimation for white matter lesions (WML) segmentation in magnetic resonance imaging (MRI). On one side, voxel-scale segmentation errors cause the erroneous delineation of the lesions; on the other side, lesion-scale detection errors lead to wrong lesion counts. Both of these factors are clinically relevant for the assessment of multiple sclerosis patients. This work aims to compare the ability of different voxel- and lesion-scale uncertainty measures to capture errors related to segmentation and lesion detection, respectively. Our main contributions are (i) proposing new measures of lesion-scale uncertainty that do not utilise voxel-scale uncertainties; (ii) extending an error retention curves analysis framework for the evaluation of lesion-scale uncertainty measures. Our results obtained on the multi-center testing set of 58 patients demonstrate that the proposed lesion-scale measure achieves the best performance among the analysed measures. All code implementations are provided at https://github.com/Medical-Image-Analysis-Laboratory/MS_WML_uncs.
KW - Magnetic resonance imaging
KW - Multiple sclerosis
KW - Reliable AI
KW - Structural-scale uncertainty
KW - White matter multiple sclerosis lesions
UR - http://www.scopus.com/inward/record.url?scp=85172098674&partnerID=8YFLogxK
U2 - 10.1109/ISBI53787.2023.10230563
DO - 10.1109/ISBI53787.2023.10230563
M3 - Conference contribution
AN - SCOPUS:85172098674
T3 - Proceedings - International Symposium on Biomedical Imaging
BT - 2023 IEEE International Symposium on Biomedical Imaging, ISBI 2023
PB - IEEE Computer Society
Y2 - 18 April 2023 through 21 April 2023
ER -