Method and apparatus for measuring endolymphatic hydrops ratio of inner ear organ using artificial neural network
Provided are a method and an apparatus for measuring an endolymphatic hydrops ratio of inner ear organs using an artificial neural network. The method of measuring an endolymphatic hydrops ratio includes obtaining a plurality of frame images obtained by capturing inner ear organs, obtaining a plurality of pieces of mask data corresponding to each of the plurality of frame images by inputting the plurality of frame images into a neural network, clustering the plurality of pieces of mask data according to the inner ear organs and obtaining representative images according to the inner ear organs according to certain conditions, and overlapping a target image synthesized by using the plurality of frame images and the representative images according to the inner ear organs so as to measure an endolymphatic hydrops ratio.
1. A method of measuring an endolymphatic hydrops ratio, the method comprising:
obtaining a plurality of frame images obtained by capturing inner ear organs;
obtaining a plurality of pieces of mask data corresponding to each of the plurality of frame images by inputting the plurality of frame images into a neural network;
clustering the plurality of pieces of mask data according to the inner ear organs and obtaining representative images according to the inner ear organs according to certain conditions; and
overlapping a target image synthesized by using the plurality of frame images and the representative images according to the inner ear organs
so as to measure
an endolymphatic hydrops ratio.
2. The method of claim 1 , wherein
the obtaining of the representative images according to the inner ear organs further comprises generating three-dimensional (3D) segmentation corresponding to each of the inner ear organs by using the plurality of pieces of mask data clustered according to the inner ear organs, and
the measuring of the endolymphatic hydrops ratio further comprises overlapping the target image and the 3D segmentation so as to measure an endolymphatic hydrops ratio.
3. The method of claim 1 , wherein the obtaining of the representative images according to the inner ear organs comprises obtaining mask data having a largest inner ear organ region included in each of the mask data among the plurality of pieces of mask data clustered according to the inner ear organs as representative images of the inner ear organs.
4. The method of claim 1 , wherein the measuring of the endolymphatic hydrops ratio comprises
overlapping the target image and the representative image according to the inner ear organs so as to obtain an endolymphatic hydrops detection image; and
measuring an endolymphatic hydrops ratio by using pixel data having negative values among inner ear organ mask regions included in the endolymphatic hydrops detection image.
5. The method of claim 1 , wherein
the plurality of frame images comprise frame images included in a MR cisternography (MRC) image obtained by capturing the inner ear organs, and
the target image comprises a HYDROPS image Multiplied with heavily T2-weighted MR cisternography (Hydrops-Mi2) image generated by synthesizing the MRC image and the Hydrops image.
6. An apparatus for measuring an endolymphatic hydrops ratio, the apparatus comprising a processor, wherein
the processor obtains a plurality of frame images obtained by capturing inner ear organs, obtains a plurality of pieces of mask data corresponding to each of the plurality of frame images by inputting the plurality of frame images into a neural network, clusters the plurality of pieces of mask data according to the inner ear organs, obtains representative images according to the inner ear organs according to certain conditions, and overlaps a target image synthesized by using the plurality of frame images and the representative images according to the inner ear organs so as to measure an endolymphatic hydrops ratio.
7. The apparatus of claim 6 , wherein
the processor further generates three-dimensional (3D) segmentation corresponding to each of the inner ear organs by using the plurality of pieces of mask data clustered according to the inner ear organs and overlaps the target image and the 3D segmentation so as to measure an endolymphatic hydrops ratio.
8. The apparatus of claim 6 , wherein
the processor obtains mask data having a largest inner ear organ region included in each of the mask data among the plurality of pieces of mask data clustered according to the inner ear organs as representative images of the inner ear organs.
9. The apparatus of claim 6 , wherein the processor overlaps the target image and the representative image according to the inner ear organs so as to obtain an endolymphatic hydrops detection image and measure an endolymphatic hydrops ration by using pixel data having negative values included in the endolymphatic hydrops detection image.
10. The apparatus of claim 6 , wherein
the plurality of frame images comprise frame images included in a MR cisternography (MRC) image obtained by capturing the inner ear organs, and
the target image comprises a HYDROPS image Multiplied with heavily T2-weighted MR cisternography (Hydrops-Mi2) image generated by synthesizing the MRC image and the Hydrops image.