Accurate colorization for holographic imaging of pathology slides using absorbance spectrum estimation of histochemical stains (Conference Presentation)
Optics and Biophotonics in Low-Resource Settings V(2019)
摘要
Digital holographic microscopy (DHM) has various unique advantages, such as large depth-of-field, simplicity of the optical setup, and the capability to reconstruct both the amplitude and phase images of samples. However, due to the use of narrow-band illumination sources, holographic imaging of pathology samples is limited by imperfect color representation, which might negatively impact diagnostic decisions. Here, an accurate-color holographic microscopy framework is presented using absorbance spectrum estimation of histochemical stains with a minimum mean square error (MMSE) criterion. Using this method, a pathology slide is imaged using a holographic microscope at a small number of wavelengths (e.g., three to six). These multispectral images are then used to estimate the absorbance spectrum of the sample at each pixel location, and to calculate a color-corrected tristimulus image. Based on this approach, we further optimize the selection of wavelengths by minimizing the color error of the reconstructed image compared to the ground truth, color-accurate image that is obtained using 31 illumination wavelengths (acquired sequentially). Based on this absorbance spectrum estimation method and the selection of optimal illumination wavelengths, we significantly improved the average color error of holographic images of 25 samples with different tissue-stain combinations, including breast, kidney, esophagus, lung, liver, artery, and Pap smear samples combined with H&E, PAS, MT, EVG, Congo Red, GMS, Alcian Blue, Jones, Gram, and Pap stains. The presented method provides a practical guide for accurate-color holographic imaging of stained tissue samples and cells for digital pathology and telemedicine applications.
更多查看译文
关键词
holographic imaging,histochemical stains,accurate colorization,absorbance spectrum estimation
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要