TY - GEN
T1 - Reducing Annotation Need in Self-explanatory Models for Lung Nodule Diagnosis
AU - Lu, Jiahao
AU - Yin, Chong
AU - Krause, Oswin
AU - Erleben, Kenny
AU - Nielsen, Michael Bachmann
AU - Darkner, Sune
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022/10/1
Y1 - 2022/10/1
N2 - Feature-based self-explanatory methods explain their classification in terms of human-understandable features. In the medical imaging community, this semantic matching of clinical knowledge adds significantly to the trustworthiness of the AI. However, the cost of additional annotation of features remains a pressing issue. We address this problem by proposing cRedAnno, a data-/annotation-efficient self-explanatory approach for lung nodule diagnosis. cRedAnno considerably reduces the annotation need by introducing self-supervised contrastive learning to alleviate the burden of learning most parameters from annotation, replacing end-to-end training with two-stage training. When training with hundreds of nodule samples and only 1 % of their annotations, cRedAnno achieves competitive accuracy in predicting malignancy, meanwhile significantly surpassing most previous works in predicting nodule attributes. Visualisation of the learned space further indicates that the correlation between the clustering of malignancy and nodule attributes coincides with clinical knowledge. Our complete code is open-source available: https://github.com/diku-dk/credanno.
AB - Feature-based self-explanatory methods explain their classification in terms of human-understandable features. In the medical imaging community, this semantic matching of clinical knowledge adds significantly to the trustworthiness of the AI. However, the cost of additional annotation of features remains a pressing issue. We address this problem by proposing cRedAnno, a data-/annotation-efficient self-explanatory approach for lung nodule diagnosis. cRedAnno considerably reduces the annotation need by introducing self-supervised contrastive learning to alleviate the burden of learning most parameters from annotation, replacing end-to-end training with two-stage training. When training with hundreds of nodule samples and only 1 % of their annotations, cRedAnno achieves competitive accuracy in predicting malignancy, meanwhile significantly surpassing most previous works in predicting nodule attributes. Visualisation of the learned space further indicates that the correlation between the clustering of malignancy and nodule attributes coincides with clinical knowledge. Our complete code is open-source available: https://github.com/diku-dk/credanno.
KW - Explainable AI
KW - Intrinsic explanation
KW - Lung nodule diagnosis
KW - Self-explanatory model
KW - Self-supervised learning
UR - http://www.scopus.com/inward/record.url?scp=85141792446&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-17976-1_4
DO - 10.1007/978-3-031-17976-1_4
M3 - Conference proceeding
AN - SCOPUS:85141792446
SN - 9783031179754
SN - 9783031179761
T3 - Lecture Notes in Computer Science
SP - 33
EP - 43
BT - Interpretability of Machine Intelligence in Medical Image Computing
A2 - Reyes, Mauricio
A2 - Henriques Abreu, Pedro
A2 - Cardoso, Jaime
PB - Springer Cham
T2 - 5th International Workshop on Interpretability of Machine Intelligence in Medical Image Computing, iMIMIC 2022, held in conjunction with the 25th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2022
Y2 - 22 September 2022 through 22 September 2022
ER -
