Search

EP-3698276-B1 - IMAGE ANALYSIS USING DEVIATION FROM NORMAL DATA

EP3698276B1EP 3698276 B1EP3698276 B1EP 3698276B1EP-3698276-B1

Inventors

  • ZHANG, MIN
  • AVINASH, GOPAL BILIGERI

Dates

Publication Date
20260506
Application Date
20180208

Claims (9)

  1. A machine learning system (100), comprising: a memory (112) that stores computer executable components; a processor (110) configured to execute the computer executable components stored in the memory (112), wherein the computer executable components comprise: an atlas map component (104) configured to generate atlas map data (404) that includes a first portion of patient image data (402) from a plurality of reference pediatric patients and a second portion of the patient image data (402) from a plurality of target pediatric patients, wherein the first portion of patient image data (402) is normal patient image data, wherein the second portion of the patient image data (402) is abnormal patient image data and is associated with a plurality of clinical conditions and wherein the patient image data (402) are a set of pediatric brain scan images, wherein the atlas map component is configured to match the first portion of the patient image data to a corresponding age group for a set of patient identities associated with the first portion of the patient image data, wherein the atlas map component (104) is configured to register patient image data of a given age group from the first portion of the patient image data (402) to an atlas map to form a standardized age matched normal patient grouping for the age group, and wherein the atlas map component (104) is configured to register patient image data from the second portion of the patient image data (402) to the atlas map to form a standardized abnormal patient grouping; wherein the atlas map component (104) is configured to generate an image intensity value for each data value of the atlas map data, and wherein the atlas map component (104) is configured to generate a statistical representation of the first portion of the patient image data (402) comprising point-by-point intensity value mean and intensity value standard deviation; a deviation map component (106) configured to generate deviation map data (406) that represents an amount of deviation for the second portion of the patient image data (402) compared to the first portion of the patient image data (402), wherein the deviation map component (106) is configured to generate deviation map data for each abnormal patient image in the second portion of the patient image data (402) by subtracting, from an image intensity value, an intensity value mean associated with a corresponding location within the atlas map to generate a difference value, and dividing the difference value by an intensity standard deviation associated with a corresponding location within the atlas map; and a neural network component (108) configured to train a neural network (502) based on the deviation map data (406) to determine one or more clinical conditions included in image data.
  2. The machine learning system (100) of claim 1, wherein the atlas map component (104) normalizes the first portion of the patient image data and the second portion of the patient image data to generate the atlas map data (404).
  3. The machine learning system (100) of claim 1 or claim 2, wherein the atlas map component (104) formats the atlas map data (404) as a matrix of numerical data values that represent the first portion of the patient image data and the second portion of the patient image data.
  4. A method (800), comprising using a processor operatively coupled to memory to execute computer executable components to perform the following acts: generating (802) atlas map data that includes a first portion of patient image data associated with a plurality of reference pediatric patients and a second portion of the patient image data associated with a plurality of pediatric patients, wherein the first portion of patient image data (402) is normal patient image data, wherein the second portion of the patient image data (402) is abnormal patient image data and is associated with a plurality of clinical conditions and wherein the patient image data (402) are a set of pediatric brain scan images, wherein generating the atlas map data comprises matching the first portion of the patient image data to a corresponding age group for a set of patient identities associated with the first portion of the patient image data, wherein generating atlas map data comprises registering patient image data of a given age group from the first portion of the patient image data (402) to an atlas map to form a standardized age matched normal patient grouping for the age group, and registering patient image data from the second portion of the patient image data (402) to the atlas map to form a standardized abnormal patient grouping; and wherein generating the atlas map data comprises generating an image intensity value for each data value of the atlas map data, and generating a statistical representation of the first portion of the patient image data (402) comprising point-by-point intensity value mean and intensity value standard deviation; generating (804) deviation map data that represents an amount of deviation between the second portion of the patient image data and the first portion of the patient image data, wherein deviation map data for each abnormal patient image in the second portion of the patient image data (402) is generated by subtracting, from an image intensity value, an intensity value mean associated with a corresponding location within the atlas map to generate a difference value, and dividing the difference value by an intensity standard deviation associated with a corresponding location within the atlas map; and training (806) a neural network based on the deviation map data to determine one or more clinical conditions included in image data.
  5. The method (800) of claim 4, wherein the generating (802) the atlas map data comprises grouping the first portion of the patient image data based on a set of age groups for the set of patient identities.
  6. A computer readable storage device (924) comprising instructions that, in response to execution, cause a system comprising a processor to perform operations, comprising: generating (802) atlas map data that includes a first portion of patient image data (402) associated with a plurality of reference pediatric patients and a second portion of the patient image data (402) associated with a plurality of target pediatric patients, wherein the first portion of patient image data (402) is normal patient image data, wherein the second portion of the patient image data (402) is abnormal patient image data and is associated with a plurality of clinical conditions and wherein the patient image data (402) are a set of pediatric brain scan images, wherein generating the atlas map data comprises matching the first portion of the patient image data to a corresponding age group for a set of patient identities associated with the first portion of the patient image data; wherein generating atlas map data comprises registering patient image data of a given age group from the first portion of the patient image data (402) to an atlas map to form a standardized age matched normal patient grouping for the age group, and registering patient image data from the second portion of the patient image data (402) to the atlas map to form a standardized abnormal patient grouping; and wherein generating the atlas map data comprises generating an image intensity value for each data value of the atlas map data, and generating a statistical representation of the first portion of the patient image data (402) comprising point-by-point intensity value mean and intensity value standard deviation; modifying the atlas map data to generate (804) deviation map data that represents an amount of deviation between the second portion of the patient image data and the first portion of the patient image data, wherein deviation map data for each abnormal patient image in the second portion of the patient image data (402) is generated by subtracting, from an image intensity value, an intensity value mean associated with a corresponding location within the atlas map to generate a difference value, and dividing the difference value by an intensity standard deviation associated with a corresponding location within the atlas map; and training (806) a neural network based on the deviation map data to determine one or more clinical conditions included in image data.
  7. The computer readable storage device (924) of claim 6, wherein the generating (802) the atlas map data comprises normalizing the first portion of the patient image data and the second portion of the patient image data.
  8. The computer readable storage device (924) of claim 6 or claim 7, wherein the generating (802) the atlas map data comprises formatting the atlas map data as a matrix of numerical data values that represent the first portion of the patient image data and the second portion of the patient image data.
  9. The computer readable storage device of claim 8, wherein the modifying the atlas map data comprises converting the matrix of numerical data values into a matrix of colorized data values formatted based on the amount of deviation for the second portion of the patient image data compared to the first portion of the patient image data.

Description

RELATED APPLICATION This application claims priority to U.S. Provisional Application No. 62/574,333, filed October 19, 2017, and entitled "DEEP LEARNING ARCHITECTURE FOR AUTOMATED IMAGE FEATURE EXTRACTION". TECHNICAL FIELD This disclosure relates generally to artificial intelligence. BACKGROUND Artificial Intelligence (AI) can be employed for classification and/or analysis of digital images. For instance, AI can be employed for image recognition. In certain technical applications, AI can be employed to enhance imaging analysis. In an example, region-of-interest based deep neural networks can be employed to localize a feature in a digital image. However, accuracy and/or efficiency of a classification and/or an analysis of digital images using conventional artificial techniques is generally difficult to achieve. Furthermore, conventional artificial techniques for classification and/or analysis of digital images generally requires labor-intensive processes such as, for example, pixel annotations, voxel level annotations, etc. As such, conventional artificial techniques for classification and/or analysis of digital images can be improved. US 2012/0070044 describes a system and method for analyzing and visualizing a local feature of interest which includes access of a clinical image dataset comprising clinical image data acquired from a patient, identification of a region of interest (ROI) from the clinical image dataset, and extraction of at least one local feature corresponding to the ROI. The system and method also include definition of a local feature dataset comprising data representing at least one local feature, access of a pre-computed reference dataset comprising image data representing an expected value of the at least one identified derived characteristic of interest, and comparison of the characteristic dataset to the pre-computed reference dataset. Further, the system and method include calculation of at least one deviation metric from the comparison and output of a visualization of the at least one deviation metric. US 2017/0213339 describes a method and system for segmenting medical images. US 2009/0082637 describes disease or biomedical condition assessments or classifications computed with scores from multiple different imaging modalities. Aljabar, P et al: "Multi-atlas based segmentation of brain images: Atlas selection and its effect on accuracy", Neuroimage, vol. 46, no. 3, 1 July 2009 describes a framework to address the problems of scale in multi-atlas segmentation. US 2012/051608 describes a system and method for analyzing and visualizing local clinical features which includes identification of a first region of interest (ROI) from a medical image dataset acquired from a patient and extraction of a feature dataset representing a feature of interest specific to the ROI. The system also includes identification of a second ROI from the medical image dataset, extraction of a reference dataset comprising reference data representing an expected behavior of the feature of interest, comparison of the feature dataset to the reference dataset, generation of a deviation metric representing a deviation of the feature of interest based on the comparison, and creation of a visual representation of the deviation metric. SUMMARY Aspects of the presently claimed invention are set out in the independent claims. Particular embodiments of these aspects are set out in the dependent claims. Any subject matter contained herein that does not fall within the scope of the appended claims is considered as being useful for understanding the invention. The following presents a simplified summary of the specification in order to provide a basic understanding of some aspects of the specification. This summary is not an extensive overview of the specification. It is intended to neither identify key or critical elements of the specification, nor delineate any scope of the particular implementations of the specification or any scope of the claims. Its sole purpose is to present some concepts of the specification in a simplified form as a prelude to the more detailed description that is presented later. According to an embodiment, a system includes an atlas map component, a deviation map component, and a neural network component. The atlas map component generates atlas map data indicative of an atlas map that includes a first portion of patient image data from a plurality of reference patients that satisfies a first defined criterion and a second portion of the patient image data from a plurality of target patients that satisfies a second defined criterion. The first portion of the patient image data is matched to a corresponding age group for a set of patient identities associated with the first portion of the patient image data. The second portion of the patient data is associated with a plurality of clinical conditions. The deviation map component generates deviation map data that represents an amount of deviation for th