Mammography classification with multi-view deep learning techniques: Investigating graph and transformer-based architectures.

Autor: Manigrasso F; Politecnico di Torino, Dipartimento di Automatica e Informatica, Corso Duca degli Abruzzi 24, 10129, Turin, Italy., Milazzo R; Politecnico di Torino, Dipartimento di Automatica e Informatica, Corso Duca degli Abruzzi 24, 10129, Turin, Italy., Russo AS; Politecnico di Torino, Dipartimento di Automatica e Informatica, Corso Duca degli Abruzzi 24, 10129, Turin, Italy., Lamberti F; Politecnico di Torino, Dipartimento di Automatica e Informatica, Corso Duca degli Abruzzi 24, 10129, Turin, Italy., Strand F; Department of Oncology-Pathology, Karolinska Institute, Stockholm, Sweden; Department of Breast Radiology, Karolinska University Hospital, Stockholm, Sweden., Pagnani A; Politecnico di Torino, Dipartimento di Scienza Applicata e Tecnologia, Corso Duca degli Abruzzi 24, 10129, Turin, Italy., Morra L; Politecnico di Torino, Dipartimento di Automatica e Informatica, Corso Duca degli Abruzzi 24, 10129, Turin, Italy. Electronic address: lia.morra@polito.it.
Jazyk: angličtina
Zdroj: Medical image analysis [Med Image Anal] 2024 Sep 02; Vol. 99, pp. 103320. Date of Electronic Publication: 2024 Sep 02.
DOI: 10.1016/j.media.2024.103320
Abstrakt: The potential and promise of deep learning systems to provide an independent assessment and relieve radiologists' burden in screening mammography have been recognized in several studies. However, the low cancer prevalence, the need to process high-resolution images, and the need to combine information from multiple views and scales still pose technical challenges. Multi-view architectures that combine information from the four mammographic views to produce an exam-level classification score are a promising approach to the automated processing of screening mammography. However, training such architectures from exam-level labels, without relying on pixel-level supervision, requires very large datasets and may result in suboptimal accuracy. Emerging architectures such as Visual Transformers (ViT) and graph-based architectures can potentially integrate ipsi-lateral and contra-lateral breast views better than traditional convolutional neural networks, thanks to their stronger ability of modeling long-range dependencies. In this paper, we extensively evaluate novel transformer-based and graph-based architectures against state-of-the-art multi-view convolutional neural networks, trained in a weakly-supervised setting on a middle-scale dataset, both in terms of performance and interpretability. Extensive experiments on the CSAW dataset suggest that, while transformer-based architecture outperform other architectures, different inductive biases lead to complementary strengths and weaknesses, as each architecture is sensitive to different signs and mammographic features. Hence, an ensemble of different architectures should be preferred over a winner-takes-all approach to achieve more accurate and robust results. Overall, the findings highlight the potential of a wide range of multi-view architectures for breast cancer classification, even in datasets of relatively modest size, although the detection of small lesions remains challenging without pixel-wise supervision or ad-hoc networks.
Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lia Morra, Fabrizio Lamberti, Andrea Pagnani report financial support was provided by HealthTriage srl. Lia Morra, Francesco Manigrasso, Alessandro Russo, Rosario Milazzo has patent pending to HealthTriage srl. Fredrik Strand reports speaker fees from Lunit and Pfizer. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
(Copyright © 2024 The Author(s). Published by Elsevier B.V. All rights reserved.)
Databáze: MEDLINE
Externí odkaz: