Research map / Detection and segmentation / CNN features
ACAT: Adversarial Counterfactual Attention for Classification and Detection in Medical Imaging
Alessandro Fontanella, Antreas Antoniou, Wenwen Li, Joanna Wardlaw, Grant Mair, Emanuele Trucco, Amos Storkey
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add saliency-attention modules (ACAT) | baseline CNN · IST-3 | accuracy: 71.39% -> 72.55% | Improved |
| add saliency-attention modules (ACAT) | ResNet-50 baseline · MosMed | accuracy: 67.71% -> 70.84% | Improved |
| use image-feature attention (SpAtt) instead of saliency attention | baseline CNN · IST-3 | accuracy: 71.39% -> 70.96% | No clear effect |
| replace CNN with vision transformer | baseline CNN · IST-3 | accuracy: 71.39% -> 57.87% | Hurt |
| use gradient saliency maps instead of adversarial counterfactual maps (ablation) | ACAT · IST-3 | accuracy: 72.55% -> 72.16% | Hurt |
| use Grad-CAM saliency maps instead of ours (ablation) | ACAT · IST-3 | accuracy: 72.55% -> 69.42% | Hurt |
| also remove late attention layer (ablation) | ACAT · MosMed | accuracy: 69.79% -> 68.75% | Hurt |
| also remove middle attention layer (ablation) | ACAT · MosMed | accuracy: 68.75% -> 68.23% | Hurt |
| add random dropout (p=0.2) on image features instead of saliency attention (ablation) | baseline · IST-3 | accuracy: 71.39% -> 68.71% | Hurt |
| add random dropout (p=0.2) on image features instead of saliency attention (ablation) | baseline · MosMed | accuracy: 67.71% -> 53.13% | Hurt |
| progressive latent-space counterfactual saliency instead of latent shift | saliency generation · IST-3 lesion location | hit rate S: 58.39% -> 65.05% | Improved |
| use ACAT as classifier to generate counterfactual saliency maps | saliency generation · IST-3 lesion location | hit rate S: 65.05% -> 68.55% | Improved |
Closest papers by meaning
- Toward Robust Diagnosis: A Contour Attention Preserving Adversarial Defense for COVID-19 Detection (AAAI 2023 · similarity 0.85)
- NaFV-Net: An Adversarial Four-view Network for Mammogram Classification (AAAI 2025 · similarity 0.83)
- EMCAD: Efficient Multi-scale Convolutional Attention Decoding for Medical Image Segmentation (CVPR 2024 · similarity 0.82)
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.82)
- Attacking CNNs in Histopathology with SNAP: Sporadic and Naturalistic Adversarial Patches (Student Abstract) (AAAI 2024 · similarity 0.82)
- A World Model of Radiologist Reading for Medical Image Representation Learning (NeurIPS 2026 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
