Automatic graph scoring for neuropsychological assessments

Patent metadata

Field	Value
Publication number	US-2019304090-A1
Application number	US-201916373079-A
Country	US
Kind code	A1
Filing date	Apr 2, 2019
Priority date	Apr 2, 2018
Publication date	Oct 3, 2019
Grant date	—

Abstract

Systems and methods of the present invention provide for: receiving a digital image data; modifying the digital image data to reduce a width of a feature within the digital image data; executing a dimension reduction process on the feature; storing a feature vector comprising: at least one feature for each of the received digital image data, and a correct or incorrect label associated with each feature vector; selecting the feature vector from a data store; training a classification software engine to classify each feature vector according to the label; classifying the image data as correct or incorrect according to a classification software engine; and generating an output labeling a second digital image data as correct or incorrect.

First claim

The invention claimed is:

1 . A system comprising a server, the server comprising a hardware computing device coupled to a network and including at least one processor executing within a memory instructions comprising a specific device logic which, when executed, cause the system to:
receive a plurality of image inputs, each comprising a digital image data; modify the digital image data to reduce a width of at least one feature within the digital image data; execute a dimension reduction process on the at least one feature within the digital image data; store, within a data store:
a feature vector for each digital image data in the plurality of image input, the feature vector comprising the at least one feature of the digital image data; and
a correct or incorrect label associated with each feature vector;
select, from the data store, the feature vector for each digital image data; and train a classification software engine to classify each feature vector according to the label associated with the feature vector.

2 . The system of claim 1 , wherein the instructions further cause the system to:
receive a second image input comprising a second digital image data; classify the second digital image data as correct or incorrect according to the classification software engine; and generate an output labeling the second digital image data as correct or incorrect.

3 . The system of claim 1 , wherein the instructions further cause the system to, for each of a first plurality of pixels within the digital image data:
identify a pixel within the first plurality of pixels; identify a total number of black pixels within a second plurality of pixels immediately adjacent to the pixel and a third plurality of pixels immediately adjacent to the second plurality of pixels determine whether the total number of black pixels is greater than a threshold number; responsive to a determination that the total number of black pixels is greater than the threshold number, characterize the pixel as a black pixel; responsive to a determination that the total number of black pixels is not greater than the threshold number, characterize the pixel as a white or an empty pixel.

4 . The system of claim 1 , wherein the dimension reduction process includes execution of an orthogonal transformation to reduce a number of the at least one feature within the feature vector data for each digital image data.

5 . The system of claim 1 , wherein:
each at least one feature is a pixel in the digital image data; and the dimension reduction algorithm reduces the number of pixels and features in the feature vector.

6 . The system of claim 1 , wherein the classification software engine uses the feature vector to train the model with a non-linear Radial Basis Function (RBF) kernel method.

7 . The system of claim 1 , wherein the classification software engine is trained by:
partitioning the plurality of feature vectors into a plurality of folds; designating a fold within the plurality of folds as a validation dataset; designating each remaining fold in the plurality of folds as a training dataset; running a performance analysis for each of a plurality of subsets in the training dataset, the performance analysis comprising:
performing an analysis on each of the plurality of training subsets in the training dataset; and
validating the analysis against the validation dataset;
designating each remaining fold in the plurality of folds as the validation dataset, and the validation dataset as being within the training dataset; and repeating the performance analysis for each remaining fold in the plurality of folds.

8 . A system comprising a server, the server comprising a hardware computing device coupled to a network and including at least one processor executing within a memory instructions comprising a specific device logic which, when executed, cause the system to:
receive an image input comprising a digital image data; modify the digital image data to reduce a width of at least one feature within the digital image data; execute a dimension reduction process on the at least one feature within the digital image data; classify the digital image data as correct or incorrect according to a classification software engine; and generate an output labeling the digital image data as correct or incorrect.

9 . The system of claim 8 , wherein the instructions further cause the system to:
receive a plurality of image inputs, each comprising the digital image data; store, within a data store:
a feature vector for each digital image data in the plurality of image inputs, the feature vector comprising the at least one feature of the digital image data; and
a correct or incorrect label associated with each feature vector;
select, from the data store, the feature vector for each digital image data; and train the classification software engine to classify each feature vector according to the label associated with the feature vector.

10 . The system of claim 8 , wherein the computer executable instructions further cause the system to, for each of a first plurality of pixels within the digital image data:
identify a pixel within the first plurality of pixels; identify a total number of black pixels within a second plurality of pixels immediately adjacent to the pixel and a third plurality of pixels immediately adjacent to the second plurality of pixels determine whether the total number of black pixels is greater than a threshold number; responsive to a determination that the total number of black pixels is greater than the threshold number, characterize the pixel as a black pixel; responsive to a determination that the total number of black pixels is not greater than the threshold number, characterize the pixel as a white or an empty pixel.

11 . The system of claim 8 , wherein the dimension reduction process includes execution of an orthogonal transformation to reduce a number of the at least one feature within the feature vector data for each digital image data.

12 . The system of claim 9 , wherein:
each at least one feature is a pixel in the digital image data; and the dimension reduction algorithm reduces the number of pixels and features in the feature vector.

13 . The system of claim 8 , wherein the classification software engine uses the feature vector to train the model with a non-linear Radial Basis Function (RBF) kernel method.

14 . The system of claim 9 , wherein the classification software engine is trained by:
partitioning the plurality of feature vectors into a plurality of folds; designating a fold within the plurality of folds as a validation dataset; designating each remaining fold in the plurality of folds as a training dataset; running a performance analysis for each of a plurality of subsets in the training dataset, the performance analysis comprising:
performing an analysis on each of the plurality of training subsets in the training dataset; and
validating the analysis against the validation dataset;
designating each remaining fold in the plurality of folds as the validation dataset, and the validation dataset as being within the training dataset; and repeat the performance analysis for each remaining fold in the plurality of folds.

15 . A method comprising the steps of:
receiving, by a server comprising a hardware computing device coupled to a network and comprising at least one processor executing computer-executable instructions within a memory, a plurality of image inputs, each including a digital image data; modifying, by the server hardware computing

Assignees

• Pearson Education Inc

Inventors

• Wang Zhaodong

Classifications

• G16H50/70

  for mining of medical data, e.g. analysing previous cases of other patients · CPC title

• A61B5/7267

  involving training the classification device · CPC title

• A61B5/4076

  Diagnosing or monitoring particular conditions of the nervous system · CPC title

• A61B5/1128

  using image analysis (A61B5/1127 takes precedence) · CPC title

• A61B5/7253

  characterised by using transforms · CPC title