Method and apparatus for rapid diagnosis of hematologic malignancy using 3D quantitative phase imaging and deep learning

1. A non-label diagnosis apparatus for a hematologic malignancy using three-dimensional (3-D) quantitative phase imaging and deep learning, the apparatus comprising:

a 3-D refractive index cell imaging unit configured to generate a 3-D refractive index slide image of a blood smear specimen by capturing a 3-D refractive index image of the blood smear specimen in which blood of a patient has been smeared on a slide glass or in which the 3-D refractive index slide image has a single layer or a plurality of layers on an in vitro slide glass;

a region-of-interest (ROI) detection unit configured to sample a suspected cell segment in the blood smear specimen based on the 3-D refractive index slide image and to determine, as ROI patches, cells determined as abnormal cells; and

a diagnosis unit configured to determine a sub-classification of a cancer cell corresponding to each of the ROI patches using a cancer cell sub-classification determination model constructed based on a deep learning algorithm and to generate hematologic malignancy diagnosis results by gathering sub-classification results of the ROI patches;

wherein the 3-D refractive index cell imaging unit comprises a 3-D patch image capturing unit configured to simultaneously capture 3-D refractive index images of the blood smear specimen in which the blood has been smeared on the slide glass.

2. The apparatus of claim 1 , further comprising:

a light source configured to have light incident on the blood smear specimen;

an interferometer configured to obtain multiple 2-D holograms by measuring transmitted light diffracted from the blood smear specimen; and

a measurement unit configured to measure a 3-D refractive index distribution of the blood smear specimens using the multiple 2-D holograms,

wherein the 3-D refractive index distribution of the blood smear specimens is measured using the multiple 2-D holograms, measured by the interferometer, by rotating an angle of the light incident on the blood smear specimens.

3. The apparatus of claim 1 , further comprising:

a light source configured to have light incident on the blood smear specimen;

an interferometer configured to obtain multiple 2-D holograms by measuring transmitted light diffracted from the blood smear specimen; and

a measurement unit configured to measure a 3-D refractive index distribution of the blood smear specimens using the multiple 2-D holograms,

wherein the 3-D refractive index distribution of the blood smear specimens is measured using the multiple 2-D holograms, measured by the interferometer, by directly rotating the blood smear specimens.

4. The apparatus of claim 1 , wherein the 3-D refractive index cell imaging unit further comprises:

a patch image connector configured to generate the 3-D refractive index slide image by connecting the 3-D refractive index images captured.

5. The apparatus of claim 1 , wherein the ROI detection unit comprises:

a cell region extraction unit configured to extract the suspected cell segment from the 3-D refractive index slide image which is a 3-D refractive index image of all of the blood smear specimens;

a cell patch sampling unit configured to generate a patch by sampling the extracted suspected cell segment; and

an ROI determination unit configured to determine, as the ROI patches, the cells determined as the abnormal cells.

6. The apparatus of claim 1 , wherein the diagnosis unit generates the hematologic malignancy diagnosis results comprising a number of cancer cells for each sub-classification by gathering sub-classification results of the ROI patches.

7. The apparatus of claim 1 , wherein the cancer cell sub-classification determination model determines the sub-classification of the cancer cell corresponding to each ROI patch by merging results of the deep learning algorithm to which the measured 3-D refractive index image has been input based on a probability.

8. A non-label diagnosis method for a hematologic malignancy using 3-D quantitative phase imaging and deep learning, the method comprising:

generating a 3-D refractive index slide image of a blood smear specimen by capturing a 3-D refractive index image of the blood smear specimen in which blood of a patient has been smeared on a slide glass or in which the 3-D refractive index slide image has a single layer or a plurality of layers on an in vitro slide glass;

sampling a suspected cell segment in the blood smear specimen based on the 3-D refractive index slide image and determining, as ROI patches, cells determined as abnormal cells; and

determining a sub-classification of a cancer cell corresponding to each of the ROI patches using a cancer cell sub-classification determination model constructed based on a deep learning algorithm and generating hematologic malignancy diagnosis results by gathering sub-classification results of the ROI patches;

wherein generating the 3-D refractive index slide image of the blood smear specimen comprises simultaneously capturing 3-D refractive index images of the blood smear specimen in which the blood has been smeared on the slide glass.

9. The method of claim 8 , wherein generating the 3-D refractive index slide image of the blood smear specimen further comprises:

generating the 3-D refractive index slide image by connecting the 3-D refractive index images captured.

10. The method of claim 8 , wherein:

generating the hematologic malignancy diagnosis results by gathering the sub-classification results of the ROI patches comprises generating the hematologic malignancy diagnosis results comprising a number of cancer cells for each sub-classification by gathering sub-classification results of the ROI patches, and

the cancer cell sub-classification determination model determines the sub-classification of the cancer cell corresponding to each ROI patch by merging results of the deep learning algorithm to which the measured 3-D refractive index image has been input based on a probability.