Publications by authors named "Haoxuan Che"
Rethinking Self-Training for Semi-Supervised Landmark Detection: A Selection-Free Approach.
IEEE Trans Image Process
September 2024
Self-training is a simple yet effective method for semi-supervised learning, during which pseudo-label selection plays an important role for handling confirmation bias. Despite its popularity, applying self-training to landmark detection faces three problems: 1) The selected confident pseudo-labels often contain data bias, which may hurt model performance; 2) It is not easy to decide a proper threshold for sample selection as the localization task can be sensitive to noisy pseudo-labels; 3) coordinate regression does not output confidence, making selection-based self-training infeasible. To address the above issues, we propose Self-Training for Landmark Detection (STLD), a method that does not require explicit pseudo-label selection.
View Article and Find Full Text PDFArticle Synopsis
- The study develops and tests deep learning models to assess the quality of 3D macular scans from two optical coherence tomography devices, Cirrus and Spectralis.
- Researchers collected and analyzed 3D scans from over 3,800 patients, and utilized a specialized deep learning network to classify scans as gradable or ungradable.
- The models demonstrated high accuracy in internal validation and external testing, indicating they could effectively filter out low-quality scans and be integrated with disease detection systems for automated eye disease diagnosis.
Article Synopsis
- The "DRAC - Diabetic Retinopathy Analysis Challenge" was held at the MICCAI 2022 conference, introducing the DRAC ultra-wide optical coherence tomography angiography dataset containing 1,103 images to tackle diabetic retinopathy (DR) analysis tasks.
- The challenge focused on three main clinical tasks: segmenting DR lesions, assessing image quality, and grading diabetic retinopathy, attracting participation from multiple teams with 11, 12, and 13 solutions submitted for each task.
- The paper summarizes the best-performing solutions, which can aid in developing better classification and segmentation models for DR diagnosis, and the dataset is now available to enhance computer-aided diagnostic systems in the healthcare field.
Diabetic retinopathy (DR) is the leading cause of preventable blindness worldwide. The risk of DR progression is highly variable among different individuals, making it difficult to predict risk and personalize screening intervals. We developed and validated a deep learning system (DeepDR Plus) to predict time to DR progression within 5 years solely from fundus images.
View Article and Find Full Text PDF