A5037433952- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Domain Adaptation and Few-Shot Learning
- Medical Imaging Techniques and Applications
- Explainable Artificial Intelligence (XAI)
- Advanced MRI Techniques and Applications
- Medical Imaging and Analysis
- Lung Cancer Diagnosis and Treatment
- Molecular Biology Techniques and Applications
- Topic Modeling
- Robotics and Automated Systems
- Prostate Cancer Treatment and Research
- Neural Networks and Applications
- Generative Adversarial Networks and Image Synthesis
- Bayesian Modeling and Causal Inference
- Energy Efficient Wireless Sensor Networks
- Cardiac Imaging and Diagnostics
- Artificial Intelligence in Healthcare and Education
- Adversarial Robustness in Machine Learning
- Time Series Analysis and Forecasting
- Advanced X-ray and CT Imaging
- Prostate Cancer Diagnosis and Treatment
- COVID-19 diagnosis using AI
- Image Processing and 3D Reconstruction
- Software System Performance and Reliability
University of Pisa
2022-2024
Istituto di Scienza e Tecnologie dell'Informazione "Alessandro Faedo"
2022-2024
National Research Council
2022-2024
European Institute of Oncology
2022
We present a novel technique to discover and exploit weak causal signals directly from images via neural networks for classification purposes. This way, we model how the presence of feature in one part image affects appearance another different image. Our method consists convolutional network backbone causality-factors extractor module, which computes weights enhance each map according its influence scene. develop architecture variants empirically evaluate all models on two public datasets...
Causality and eXplainable Artificial Intelligence (XAI) have developed as separate fields in computer science, even though the underlying concepts of causation explanation share common ancient roots. This is further enforced by lack review works jointly covering these two fields. In this paper, we investigate literature to try understand how what extent causality XAI are intertwined. More precisely, seek uncover kinds relationships exist between one can benefit from them, for instance,...
In this paper, we present a novel method for the automatic classification of medical images that learns and leverages weak causal signals in image. Our framework consists convolutional neural network backbone causality-extractor module which extracts cause-effect relationships between feature maps can inform model on appearance one place image, given presence another within some other To evaluate effectiveness our approach low-data scenarios, train causality-driven architecture One-shot...
Abstract Objectives Radiomics-based analyses encompass multiple steps, leading to ambiguity regarding the optimal approaches for enhancing model performance. This study compares effect of several feature selection methods, machine learning (ML) classifiers, and sources radiomic features, on models’ performance diagnosis clinically significant prostate cancer (csPCa) from bi-parametric MRI. Methods Two multi-centric datasets, with 465 204 patients each, were used extract 1246 features per...
Due to domain shift, deep learning image classifiers perform poorly when applied a different from the training one. For instance, classifier trained on chest X-ray (CXR) images one hospital may not generalize another due variations in scanner settings or patient characteristics. In this paper, we introduce our CROCODILE framework, showing how tools causality can foster model's robustness shift via feature disentanglement, contrastive losses, and injection of prior knowledge. This way, model...
The aim of this paper is threefold. We inform the AI practitioner about human visual system with an extensive literature review; we propose a novel biologically motivated neural network for image classification; and, finally, present new plug-and-play module to model context awareness. focus on effect incorporating circuit motifs found in biological brains address recognition. Our convolutional architecture inspired by connectivity cortical and subcortical streams, implement bottom-up...
Researchers nowadays may take advantage of broad collections medical data to develop personalized medicine solutions. Imaging bio-banks play a fundamental role, in this regard, by serving as organized repositories images associated with imaging biomarkers. In context, the NAVIGATOR Project aims advance colorectal, prostate, and gastric oncology translational research leveraging quantitative multi-omics analyses. As Project's core, an bio-bank is being designed implemented web-accessible...
In this paper, we present a novel method to automatically classify medical images that learns and leverages weak causal signals in the image. Our framework consists of convolutional neural network backbone causality-extractor module extracts cause-effect relationships between feature maps can inform model on appearance one place image, given presence another within some other To evaluate effectiveness our approach low-data scenarios, train causality-driven architecture One-shot learning...
We present a novel technique to discover and exploit weak causal signals directly from images via neural networks for classification purposes. This way, we model how the presence of feature in one part image affects appearance another different image. Our method consists convolutional network backbone causality-factors extractor module, which computes weights enhance each map according its influence scene. develop architecture variants empirically evaluate all models on two public datasets...
Cine Magnetic Resonance Imaging (MRI) allows for understanding of the heart's function and condition in a non-invasive manner. Undersampling $k$-space is employed to reduce scan duration, thus increasing patient comfort reducing risk motion artefacts, at cost reduced image quality. In this challenge paper, we investigate use convolutional recurrent neural network (CRNN) architecture exploit temporal correlations supervised cine cardiac MRI reconstruction. This combined with single-image...