Method for characterizing images acquired through a video medical device

{
  "publication_number": "US10607346B2",
  "country": "US",
  "kind": "B2",
  "title": "Method for characterizing images acquired through a video medical device",
  "abstract": "According to a first aspect, the invention relates to a method to support clinical decision by characterizing images acquired in sequence through a video medical device. The method comprises defining at least one image quantitative criterion, storing sequential images in a buffer, for each image (10) in the buffer, automatically determining, using a first algorithm, at least one output based on said image quantitative criterion and attaching said output to a timeline (11).",
  "claims": [
    "1. A method to support clinical decision by characterizing images acquired in sequence through a video medical device, comprising: storing images in a buffer; for each image in the buffer, automatically determining, using a first algorithm, at least one output based on at least one image quantitative criterion, the first algorithm computing discrepancy measurement between local image descriptors, wherein the at least one output of the first algorithm comprises a cluster associated with each image; and extracting, from the buffer, the images according to said at least one output for each image.",
    "2. The method according to claim 1, wherein the at least one output of the first algorithm is a boolean, scalar or vector value.",
    "3. The method according to claim 1, wherein a timeline is formed of temporal regions corresponding to images having equal outputs.",
    "4. The method according to claim 3, wherein the timeline is formed of temporal regions corresponding to consecutive images with said equal outputs.",
    "5. The method according to claim 1, further comprising: selecting images with equal outputs of the first algorithm; and processing the selected images using a second algorithm, the second algorithm providing at least one second output.",
    "6. The method according to claim 5, wherein the second algorithm is a content-based image or video retrieval algorithm.",
    "7. The method according to claim 5, wherein the second algorithm is an image or video mosaicing algorithm.",
    "8. The method according to claim 5, wherein the second algorithm is based on image or video classification.",
    "9. The method according to claim 5, wherein the second algorithm uses an external database.",
    "10. The method according to claim 5, wherein the second algorithm is based on machine learning.",
    "11. The method according to claim 5, further comprising displaying said at least one second output.",
    "12. The method according to claim 1, wherein the first algorithm uses an external database.",
    "13. The method according to claim 1, wherein the first algorithm is based on machine learning.",
    "14. The method according to claim 1, wherein the quantitative criterion is one selected from the group consisting of kinematic stability, similarity between images, probability of belonging to a category, image or video typicity, image or video atipicity, image quality, and presence of artifacts.",
    "15. A system to support clinical decision by characterizing images acquired in sequence through a video medical device, the system comprising means for implementing the steps of a method according to claim 1."
  ],
  "description_excerpt": "The invention relates generally to image and video processing and in particular to a system and method to characterize the interpretability of images acquired in sequences and especially images acquired through a video medical device.\n\nVideo acquisition devices generate massive amounts of data. Efficient use of this data is of importance for video editing, video summarization, fast visualization and many other applications related to video management and analysis.\n\nAs illustrated in Koprinskaa et al., (“Temporal video segmentation: A survey.”, Signal Processing: Image Communication, 16 (5), 477-500 (2001)), temporal video segmentation is a key step in most existing video management tools. Many different types of algorithms have been developed to perform the temporal segmentation.\n\nEarly techniques focused on cut-boundary detection or image grouping using pixel differences, histogram comparisons, edge differences, motion analysis and the like, while more recent methods such as presented in U.S. Pat. No. 7,783,106B2 and U.S. Pat. No. 8,363,960B2 have also used image similarity metrics, classification and clustering to achieve the same goal.\n\nIn some applications as the ones in Sun, Z. et al. (“Removal of non-informative frames for wireless capsule endoscopy video segmentation”, Proc. ICAL pp. 294-299 (2012)) and Oh, J.-H. et al. (“Informative frame classification for endoscopy video”, Medical Image Analysis, 11 (2), 110-127 (2007)), the problem of temporal video segmentation may be reformulated as a classification problem that distinguishes between informative and noise images.",
  "cpc": [
    "G06T 7/0016",
    "G06T 2207/10068",
    "G06T 7/174"
  ],
  "ipc": [
    "G06T 7/00",
    "G06T 7/174",
    "A61B 5/00"
  ],
  "assignees": [
    "Mauna Kea Technologies SA"
  ],
  "inventors": [
    "Nicolas Linard",
    "Barbara Andre",
    "Julien Dauguet",
    "Tom Vercauteren"
  ],
  "filing_date": "2018-06-05",
  "publication_date": "2020-03-31",
  "grant_date": "2020-03-31",
  "priority_date": "2013-10-11",
  "application_number": "US-201815997915-A",
  "family_id": "51868938",
  "cited_by_count": 21,
  "citations": [
    "US5807256A",
    "US20120281923A1",
    "US10002427B2"
  ]
}