Title | ||
---|---|---|
SSMD: Semi-Supervised medical image detection with adaptive consistency and heterogeneous perturbation |
Abstract | ||
---|---|---|
•We propose a semi-supervised medical image detector with a novel adaptive consistency cost function which takes into account the confidence of proposals at each spatial position.•We develop novel heterogeneous perturbation strategies that consist of two novel components: a noisy residual block for feature space, and an instance-level adversarial perturbation strategy for image space. The proposed heterogeneous perturbation strategies can improve the detection accuracy by enhancing the robustness of image features.•We experimentally verify the effectiveness of the proposed modules. The experiments suggest that the proposed detector outperforms the existing state-of-the-arts considerably. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1016/j.media.2021.102117 | Medical Image Analysis |
Keywords | DocType | Volume |
Lesion detection,Nuclei detection,Semi-Supervised learning | Journal | 72 |
ISSN | Citations | PageRank |
1361-8415 | 0 | 0.34 |
References | Authors | |
0 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Hong-Yu Zhou | 1 | 9 | 2.19 |
Chengdi Wang | 2 | 0 | 1.35 |
Hao-Feng Li | 3 | 8 | 3.15 |
Gang Wang | 4 | 1 | 0.81 |
Shu Zhang | 5 | 28 | 4.46 |
Weimin Li | 6 | 9 | 2.19 |
Yizhou Yu | 7 | 2907 | 181.26 |