A5015803667- AI in cancer detection
- Fetal and Pediatric Neurological Disorders
- Medical Image Segmentation Techniques
- Topic Modeling
- Multimodal Machine Learning Applications
- Radiomics and Machine Learning in Medical Imaging
- Artificial Intelligence in Healthcare and Education
- Neonatal and fetal brain pathology
- Retinal Imaging and Analysis
- Retinal Diseases and Treatments
- 3D Surveying and Cultural Heritage
- Machine Learning in Healthcare
- Machine Fault Diagnosis Techniques
- 3D Shape Modeling and Analysis
- Glaucoma and retinal disorders
- Domain Adaptation and Few-Shot Learning
- Explainable Artificial Intelligence (XAI)
- Generative Adversarial Networks and Image Synthesis
- Gear and Bearing Dynamics Analysis
- Engineering Diagnostics and Reliability
- Human Pose and Action Recognition
- Advanced Text Analysis Techniques
- COVID-19 diagnosis using AI
- Underwater Vehicles and Communication Systems
- Advanced Numerical Analysis Techniques
Technical University of Denmark
2021-2025
Huazhong University of Science and Technology
2022
Rolling bearings based rotating machinery are widely used in various industrial applications. The failure of rolling bearings, as one the most critical components, would lead to disastrous consequences machinery. Therefore, it's paramount deliver an effective intelligent fault diagnosis method for ensure machinery's stability and reliability. With this aim, article proposes a novel approach that features extracted via improved complete ensemble empirical mode decomposition with adaptive...
Obstetric ultrasound image quality is crucial for accurate diagnosis and monitoring of fetal health. However, producing high-quality standard planes difficult, influenced by the sonographer's expertise factors like maternal BMI or fetus dynamics. In this work, we propose using diffusion-based counterfactual explainable AI to generate realistic from low-quality non-standard ones. Through quantitative qualitative evaluation, demonstrate effectiveness our method in plausible counterfactuals...
We introduce the notion of semantic image quality for applications where relies on requirements. Working in fetal ultrasound, ranking is challenging and annotations are noisy, we design a robust coarse-to-fine model that ranks images based their endow our predicted rankings with an uncertainty estimate. To annotate training data, efficient annotation scheme merge sort algorithm. Finally, compare algorithm to number state-of-the-art algorithms ultrasound assessment task, showing superior...
Scene graph generation (SGG) of surgical procedures is crucial in enhancing holistically cognitive intelligence the operating room (OR). However, previous works have primarily relied on multi-stage learning that generates semantic scene graphs dependent intermediate processes with pose estimation and object detection, which may compromise model efficiency efficacy, also impose extra annotation burden. In this study, we introduce a novel single-stage bimodal transformer framework for SGG OR,...
Shortcut learning is a phenomenon where machine models prioritize simple, potentially misleading cues from data that do not generalize well beyond the training set. While existing research primarily investigates this in realm of image classification, study extends exploration shortcut into medical segmentation. We demonstrate clinical annotations such as calipers, and combination zero-padded convolutions center-cropped sets dataset can inadvertently serve shortcuts, impacting segmentation...
Despite the rapid development of AI models in medical image analysis, their validation real-world clinical settings remains limited. To address this, we introduce a generic framework designed for deploying image-based such settings. Using this framework, deployed trained model fetal ultrasound standard plane detection, and evaluated it real-time sessions with both novice expert users. Feedback from these revealed that while offers potential benefits to practitioners, need navigational...
A comprehensive understanding of surgical scenes allows for monitoring the process, reducing occurrence accidents and enhancing efficiency medical professionals. Semantic modeling within operating rooms, as a scene graph generation (SGG) task, is challenging since it involves consecutive recognition subtle actions over prolonged periods. To address this challenge, we propose Tri-modal (i.e., images, point clouds, language) confluence with Temporal dynamics framework, termed TriTemp-OR....
The Prostate Imaging Reporting and Data System (PI-RADS) is pivotal in the diagnosis of clinically significant prostate cancer through MRI imaging. Current deep learning-based PI-RADS scoring methods often lack incorporation essential clinical guidelines~(PICG) utilized by radiologists, potentially compromising accuracy. This paper introduces a novel approach that adapts multi-modal large language model (MLLM) to incorporate PICG into without additional annotations network parameters. We...
Scene graph generation (SGG) of surgical procedures is crucial in enhancing holistically cognitive intelligence the operating room (OR). However, previous works have primarily relied on multi-stage learning, where generated semantic scene graphs depend intermediate processes with pose estimation and object detection. This pipeline may potentially compromise flexibility learning multimodal representations, consequently constraining overall effectiveness. In this study, we introduce a novel...
Congenital malformations of the brain are among most common fetal abnormalities that impact development. Previous anomaly detection methods on ultrasound images based supervised learning, rely manual annotations, and risk missing underrepresented categories. In this work, we frame as an unsupervised task using diffusion models. To end, employ inpainting-based Noise Agnostic Anomaly Detection approach identifies abnormality diffusion-reconstructed from multiple noise levels. Our only requires...
Concept bottleneck models (CBMs) include a of human-interpretable concepts providing explainability and intervention during inference by correcting the predicted, intermediate concepts. This makes CBMs attractive for high-stakes decision-making. In this paper, we take quality assessment fetal ultrasound scans as real-life use case CBM decision support in healthcare. For case, simple binary are not sufficiently reliable, they mapped directly from images highly variable quality, which model...
Confounding information in the form of text or markings embedded medical images can severely affect training diagnostic deep learning algorithms. However, data collected for clinical purposes often have such them. In dermatology, known examples include drawings rulers that are overrepresented malignant lesions. this paper, we encounter and calipers placed on found national databases containing fetal screening ultrasound scans, which correlate with standard planes to be predicted. order...
Examination of the umbilical artery with Doppler ultrasonography is performed to investigate blood supply fetus through cord, which vital for monitoring fetal health. Such examination involves several steps that must be correctly: identifying suitable sites on measurement, acquiring flow curve in form a spectrum, and ensuring compliance set quality standards. These rely heavily operator's skill, shortage experienced sonographers has thus created demand machine assistance. In this work, we...
Curvilinear structure segmentation is important in medical imaging, quantifying structures such as vessels, airways, neurons, or organ boundaries 2D slices. Segmentation via pixel-wise classification often fails to capture the small and low-contrast curvilinear structures. Prior topological information typically used address this problem, at an expensive computational cost, sometimes requiring prior knowledge of expected topology. We present DTU-Net, a data-driven approach...