Currently, it is generally known that lens-free holographic microscopy, which has no imaging lens, can realize a large field-of-view imaging with a low-cost setup. However, in order to obtain colorful images, traditio...Currently, it is generally known that lens-free holographic microscopy, which has no imaging lens, can realize a large field-of-view imaging with a low-cost setup. However, in order to obtain colorful images, traditional lensfree holographic microscopy should utilize at least three quasi-chromatic light sources of discrete wavelengths,such as red LED, green LED, and blue LED. Here, we present a virtual colorization by deep learning methods to transfer a gray lens-free microscopy image into a colorful image. Through pairs of images, i.e., grayscale lens-free microscopy images under green LED at 550 nm illumination and colorful bright-field microscopy images, a generative adversarial network(GAN) is trained, and its effectiveness of virtual colorization is proved by applying it to hematoxylin and eosin stained pathological tissue samples imaging. Our computational virtual colorization method might strengthen the monochromatic illumination lens-free microscopy in medical pathology applications and label staining biomedical research.展开更多
In this paper we introduce an image-based virtual exhibition system especially for clothing product. It can provide a powerful material substitution function, which is very useful for customization clothing-built. A n...In this paper we introduce an image-based virtual exhibition system especially for clothing product. It can provide a powerful material substitution function, which is very useful for customization clothing-built. A novel color substitution algorithm and two texture morphing methods are designed to ensure realistic substitution result. To extend it to 3D, we need to do the model reconstruction based on photos. Thus we present an improved method for modeling human body. It deforms a generic model with shape details extracted from pictures to generate a new model. Our method begins with model image generation followed by silhouette extraction and segmentation. Then it builds a mapping between pixels inside every pair of silhouette segments in the model image and in the picture. Our mapping algorithm is based on a slice space representation that conforms to the natural features of human body.展开更多
AIM: To evaluate whether virtual chromoendoscopy can improve the delineation of small bowel lesions previously detected by conventional white light small bowel capsule endoscopy(SBCE). METHODS: Retrospective single ce...AIM: To evaluate whether virtual chromoendoscopy can improve the delineation of small bowel lesions previously detected by conventional white light small bowel capsule endoscopy(SBCE). METHODS: Retrospective single center study. One hundred lesions selected from forty-nine consecutive conventional white light SBCE(SBCE-WL) examinations were included. Lesions were reviewed at three Flexible Spectral Imaging Color Enhancement(FICE) settings and Blue Filter(BF) by two gastroenterologists with ex-perience in SBCE, blinded to each other's findings, whoranked the quality of delineation as better, equivalent or worse than conventional SBCE-WL. Inter-observer percentage of agreement was determined and analyzed with Fleiss Kappa(k) coefficient. Lesions selected for the study included angioectasias(n = 39), ulcers/ero-sions(n = 49) and villous edema/atrophy(n = 12). RESULTS: Overall, the delineation of lesions was im-proved in 77% of cases with FICE 1, 74% with FICE 2, 41% with FICE 3 and 39% with the BF, with a percent-age of agreement between investigators of 89%(k = 0.833), 85%(k = 0.764), 66%(k = 0.486) and 79%(k = 0.593), respectively. FICE 1 improved the delineation of 97.4% of angioectasias, 63.3% of ulcers/erosions and 66.7% of villous edema/atrophy with a percentage of agreement of 97.4%(k = 0.910), 81.6%(k = 0.714) and 91.7%(k = 0.815), respectively. FICE 2 improved the delineation of 97.4% of angioectasias, 57.1% of ulcers/erosions and 66.7% of villous edema/atrophy, with a percentage of agreement of 89.7%(k = 0.802), 79,6%(k = 0.703) and 91.7%(k = 0.815), respectively. FICE 3 improved the delineation of 46.2% of angioecta-sias, 24.5% of ulcers/erosions and none of the cases of villous edema/atrophy, with a percentage of agreement of 53.8% [k = not available(NA)], 75.5%(k = NA) and 66.7%(k = 0.304), respectively. The BF improved the delineation of 15.4% of angioectasias, 61.2% of ulcers/erosions and 25% of villous edema/atrophy, with a per-centage of agreement of 76.9%(k = 0.558), 81.6%(k = 0.570) and 25.0%(k = NA), respectively.CONCLUSION: Virtual chromoendoscopy can improve the delineation of angioectasias, ulcers/erosions and villous edema/atrophy detected by SBCE, with almost perfect interobserver agreement for FICE 1.展开更多
基金the National Natural Science Foundation of China(No.61775096)Fundamental Research Funds for the Central Uni vers让ies(No.30919011261)+1 种基金National Key Research and Development Program(No.2019YFB2005500)Key Laboratory of Optical System Advanced Manufacturing Technology(Chinese Academy of Sciences)(No.KLOMT190101).
文摘Currently, it is generally known that lens-free holographic microscopy, which has no imaging lens, can realize a large field-of-view imaging with a low-cost setup. However, in order to obtain colorful images, traditional lensfree holographic microscopy should utilize at least three quasi-chromatic light sources of discrete wavelengths,such as red LED, green LED, and blue LED. Here, we present a virtual colorization by deep learning methods to transfer a gray lens-free microscopy image into a colorful image. Through pairs of images, i.e., grayscale lens-free microscopy images under green LED at 550 nm illumination and colorful bright-field microscopy images, a generative adversarial network(GAN) is trained, and its effectiveness of virtual colorization is proved by applying it to hematoxylin and eosin stained pathological tissue samples imaging. Our computational virtual colorization method might strengthen the monochromatic illumination lens-free microscopy in medical pathology applications and label staining biomedical research.
基金This work was supported by 973 Project(No.2002CB312100)Key National Natural Science Foundation of China Project on Digital Olympic Museum(No.60533080),National 863 High-tech Project (No.2006AA01Z303).
文摘In this paper we introduce an image-based virtual exhibition system especially for clothing product. It can provide a powerful material substitution function, which is very useful for customization clothing-built. A novel color substitution algorithm and two texture morphing methods are designed to ensure realistic substitution result. To extend it to 3D, we need to do the model reconstruction based on photos. Thus we present an improved method for modeling human body. It deforms a generic model with shape details extracted from pictures to generate a new model. Our method begins with model image generation followed by silhouette extraction and segmentation. Then it builds a mapping between pixels inside every pair of silhouette segments in the model image and in the picture. Our mapping algorithm is based on a slice space representation that conforms to the natural features of human body.
文摘AIM: To evaluate whether virtual chromoendoscopy can improve the delineation of small bowel lesions previously detected by conventional white light small bowel capsule endoscopy(SBCE). METHODS: Retrospective single center study. One hundred lesions selected from forty-nine consecutive conventional white light SBCE(SBCE-WL) examinations were included. Lesions were reviewed at three Flexible Spectral Imaging Color Enhancement(FICE) settings and Blue Filter(BF) by two gastroenterologists with ex-perience in SBCE, blinded to each other's findings, whoranked the quality of delineation as better, equivalent or worse than conventional SBCE-WL. Inter-observer percentage of agreement was determined and analyzed with Fleiss Kappa(k) coefficient. Lesions selected for the study included angioectasias(n = 39), ulcers/ero-sions(n = 49) and villous edema/atrophy(n = 12). RESULTS: Overall, the delineation of lesions was im-proved in 77% of cases with FICE 1, 74% with FICE 2, 41% with FICE 3 and 39% with the BF, with a percent-age of agreement between investigators of 89%(k = 0.833), 85%(k = 0.764), 66%(k = 0.486) and 79%(k = 0.593), respectively. FICE 1 improved the delineation of 97.4% of angioectasias, 63.3% of ulcers/erosions and 66.7% of villous edema/atrophy with a percentage of agreement of 97.4%(k = 0.910), 81.6%(k = 0.714) and 91.7%(k = 0.815), respectively. FICE 2 improved the delineation of 97.4% of angioectasias, 57.1% of ulcers/erosions and 66.7% of villous edema/atrophy, with a percentage of agreement of 89.7%(k = 0.802), 79,6%(k = 0.703) and 91.7%(k = 0.815), respectively. FICE 3 improved the delineation of 46.2% of angioecta-sias, 24.5% of ulcers/erosions and none of the cases of villous edema/atrophy, with a percentage of agreement of 53.8% [k = not available(NA)], 75.5%(k = NA) and 66.7%(k = 0.304), respectively. The BF improved the delineation of 15.4% of angioectasias, 61.2% of ulcers/erosions and 25% of villous edema/atrophy, with a per-centage of agreement of 76.9%(k = 0.558), 81.6%(k = 0.570) and 25.0%(k = NA), respectively.CONCLUSION: Virtual chromoendoscopy can improve the delineation of angioectasias, ulcers/erosions and villous edema/atrophy detected by SBCE, with almost perfect interobserver agreement for FICE 1.
