Publications by authors named "Siyu Steven Lin"
Dome-APIC illumination design for high space-bandwidth product analytic imaging.
Biomed Opt Express
April 2025
Breaking the tradeoff between resolution and field-of-view, while obtaining distortion-free images, can be achieved through computational imaging techniques. A recent approach, Angular Ptychographic Imaging with Close-form method (APIC), has showcased its capability to analytically recover both intricate aberrations and high space-bandwidth product complex optical fields with NA-matching and darkfield illuminations. However, its flat illumination setup limits its ability to efficiently reconstruct a large field-of-view simultaneously with high resolution, owing to the curvature in the wavefront from NA-matching illuminations and the finite beam angle of the Lambertian LED light source.
View Article and Find Full Text PDFArticle Synopsis
- Whole slide imaging enhances pathological analysis by providing detailed images of biopsy or surgical samples, improving clinical diagnosis and patient treatment planning.
- The newly developed WSI-APIC system utilizes angular ptychographic imaging for high-quality, tens-of-gigapixels images while eliminating the need for z-scanning and reducing background scanning.
- This system achieves an optical resolution of 772 nm and seamlessly stitches together phase images, allowing for efficient and aberration-free visualization on standard microscopic slides.
Article Synopsis
- Recent advances in deep neural networks (DNNs) have shown they can sometimes outperform expert pathologists in analyzing pathology images by learning intricate details from large datasets.
- The research highlighted issues with DNNs not being able to generalize well across different batches of histological slides, even when stain normalization techniques were applied.
- The study emphasizes the urgent need for improved methods in preparing and standardizing microscopy images so that DNNs can effectively train and predict outcomes consistently across various datasets.