Fourier Ptychographic Microscopy 영상에서의 딥러닝 기반 디지털 염색 방법 연구
Deep Learning Based Digital Staining Method in Fourier Ptychographic Microscopy Image
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 연구에서 세포를 분별하기 위해 H&E 염색이 필요하다. 직접 염색하는 많은 비용과 시간이 필요하다. H&E 염색되지 않은 세포의 Phase image에서 H&E 염색이 된 세포의 Amplitude image로 변환하는 것이 목적이다. FPM으로 촬영한 Image data를 가지고 Matlab을 이용해 매개변수를 변경해 Phase image와 Amplitude image를 만들었다. 정규화를 통해 육안으로 식별이 가능한 이미지를 얻었다. GAN 알고리즘을 이용해 Phase image를 기반으로 Real Amplitude image와 비슷한 Fake Amplitude image를 만들고 Fake Amplitude image를 가지고 MASK R-CNN을 이용하여 세포를 분별하여 객체화를 통해 구분했다. 연구 결과 D loss의 max는 3.3e-1, min은 6.8e-2, G loss max는 6.9e-2, min은 2.9e-2, A loss는 max 5.8e-1, min는 1.2e-1, mask rcnn max는 1.9e0, min은 3.2e-1이다.
Abstract
In this study, H&E staining is required to distinguish cells, but there are many disadvantages, so the purpose of this study is to convert from H&E staining cell phase image to H&E staining cell amplitude image. To this end, phase image and amplitude image were created by changing parameters using Matlab with image data captured by AAA stained cells with FPM, and images that can be visually identified were obtained through normalization. Using the GAN algorithm, a fake amplitude image similar to a conventional amplitude image was created based on a phase image, and cells were distinguished and objectified using MASK R-CNN.
참고문헌 (27)
- H Hwang, "Future risk factors of artificial intelligence technology and the direction of our response", Collection of materials for academic conferences of the Korean Local Government Society, pp.49~81 (2021)
- 주강진, "4차 산업혁명과 인공지능 : 현황, 사례, 규제에 대한 개괄적 고찰", 한국경영과학회지, 42(4), pp.1~14 (2017), 한국경영과학회[DOI]
- Mayerich. Author, "Bhargava, Rohit Stain-less staining for computed histopathology", Technology, 3(1), pp.27~31 (2015)
- JT Gwak, "Digital Pathology Technology Status and Development Direction", Electronic Engineering Society, 44(2), pp.36~41 (2017)
- JW Oh, "Improvement activities of the examination process to improve the quality of general biopsy (H&E Stay)", Tissue Cell Examination Society abstract, 2010(2), pp.2~3 (2010)
- Faliu Yi, "Automatic extraction of cell nuclei from H&E-stained histopathological images", Journal of Medical Imaging, 4(2), pp.1~12 (2017)
- de Haan, "Deep learning-based transformation of H&E stained tissues into special stains", Nature Communications, 12(1), pp.1~13 (2021)
- Avwioro, G., "Histochemical Uses of Haematoxylin - A Review", JPCS, 1, pp.24~34 (2011)
- 오영호, "임상병리사 인력의 수급전망과 정책방향", 대한임상검사과학회지, 50(4), pp.511~524 (2018), 대한임상검사과학회[DOI]
- JT Kwak, "Digital Pathology Technology Status and Development Direction", Electronic Engineering Society, 44(2), pp.36~41 (2017)
- Marshall PN, "The oxidation products of haematoxylin and their role in biological staining", The Histochemical Journal, 4, pp.493~503 (1972)
- Clark George, "Comparison of various oxidants for alum hematoxylin", Stain Technology, 49(4), pp.225~227 (1974)
- Kulkarni, "Deep Learning Based on Standard H&E Images of Primary Melanoma Tumors Identifies Patients at Risk for Visceral Recurrence and Death", Clinical cancer research :an official journal of the American Association for Cancer Research (2020)
- SY Song, "Introduction to Pathological Artificial Intelligence", Hanyang Medical Reviews, pp.77~85 (2017)
- Kyoohyun Kim, "The Relative Densities of Cytoplasm and Nuclear Compartments Are Robust against Strong Perturbation", Biophysical Journal, 119(10), pp.1946~1957 (2020)
- IJ. Goodfellow, "Generative Adversarial Networks", NIPS (2014)
- K. He, "Mask R-CNN", Proceedings of the IEEE International Conference on Computer Vision(ICCV), pp.2961~2969 (2017)
- J. Hosang, "What makes for effective detection proposals?", PAMI (2015)
- JRR Uijlings, "Selective search for object recognition", IJCV (2013)
- Alex Krizhevsky, "ImageNet classification with deep convolutional neural networks", NIPS (2012)
- Y. LeCun, "Backpropagation appFaked to handwritten zip code recognition", Neural computation (1989)
- J. Huang, "Speed", accuracy trade-offs for modern convolutional object detectors, 2017, ~
- R. Girshick, "Fast R-CNN", ICCV (2015)
- 황석민, "다중 트레이닝 기법을 이용한 MASK R-CNN의 초음파 DDH 각도 측정 진단 시스템 연구", 융합신호처리학회 논문지, 21(4), pp.183~194 (2020), 한국융합신호처리학회[DOI]
- C Zheng, "Robust Fourier ptychographic microscopy via a physics-based defocusing strategy for calibrating angle-varied LED illumination", Biomedical Optics Express, 13(3), pp.1581~1594 (2022)
- Y Rivenson, "PhaseStain: the digital staining of label-free quantitative phase microscopy images using deep learning", Light: Science &Applications, 8(23), pp.2~11 (2019)
- F. Zernike, "Phase contrast, a new method for the microscopic observation of transparent objects", Physica, 9(7), pp.686~698 (1942)