3 METHOD AND APPARATUS FOR RAPID DIAGNOSIS OF HEMATOLOGIC MALIGNANCY USING 3D QUANTITATIVE PHASE IMAGING AND DEEP LEARNING

摘要

A method and apparatus for rapid non-labeled blood cancer diagnosis using three-dimensional quantitative phase image measurement and deep learning are presented. An apparatus for diagnosing unlabeled blood cancer using three-dimensional quantitative phase image measurement and deep learning according to an embodiment includes a blood smear (including puncture and biopsy section) obtained by smearing a patient's blood (including bone marrow and other body fluids) on a slide glass. ) by taking a three-dimensional refractive index image in the form of a three-dimensional refractive index cell image measuring unit to generate a three-dimensional refractive index slide image of the blood smear; an ROI detection unit that samples a portion of a suspicious cell from the blood smear through the three-dimensional refractive index slide image, and determines cells determined to be abnormal cells as ROI (Region of Interest) patches; The quality of the sample is evaluated using the cancer cell subclassification discrimination model built on the basis of a deep learning algorithm, and the subclassification of cancer cells corresponding to each ROI patch is distinguished, and the subclassification of the ROI patches is performed. It may include a diagnostic unit that collects the results to generate a blood cancer diagnosis and classification result.