Invented by Joseph Ralph Ferrantelli, Douglas BOBERG, JR., Postureco Inc, Posture Co Inc
The Postureco Inc, Posture Co Inc invention works as followsA method of using machine learning to assist in anatomical predictions. The method comprises identifying parameters with a processor in a plurality training images in order to generate a dataset of parameters, the dataset having data linking parameters to respective images. Training at least one machine-learning algorithm using the parameters from the dataset and validating it, identifying digitized points in a plurality anatomical landmarks on an image of a human displayed on a touch screen digital display by using the validated machine-learning algorithm and a scaling factor for the displayed picture, and making a circumferential anatomical prediction based on
Background for Method for measuring anatomical measurements and postural dimensions using a digital image and machine learning
Commonly-assigned U.S. Pat. Nos. Nos.
For example, U.S. Pat. No. No. 8,721,567 describes methods and systems to calculate a patient’s postural displacement in an image displayed by a display device using the pixel-to-distance ratio. U.S. Pat. No. No. 9,801,550 describes methods and systems that allow for the acquisition of multiple two-dimensional images (2D) of a human being and the prediction of anatomical circumferential measurements of the individual based on linear anatomical dimensions determined and a known relationship between morphology and anatomy. U.S. Pat. No. No.
Despite the improvements in postural analyses and anatomical prediction, there is still a need to develop systems and practice methods that are more accurate and efficient. According to conventional practices, human interaction with digital touchscreen displays, cameras, and image analysis may result in mistakes and oversights. The conventional systems and techniques do not currently use any learning procedures to make them’smarter.’ Recent inventions have used machine learning techniques to develop systems and practice methods for predicting anatomical structures that addresses these and other shortcomings.
Machine learning algorithms enable computers to perform a wide range of tasks, including solving problems, answering questions, and performing other tasks, based on not only pre-programmed instruction, but also inferences derived from training data. Training data can be used as a basis for ‘training? The machine learning algorithms can be ‘trained’ by creating representations or generalizations which are then applied to future data. Machine learning “learns” the weights and parameters for different representations and generalizations. Machine learning is used to “learn” the parameters and weights of different representations.
The inventors have looked to using machine learning algorithms and classifier software implemented on a specialized computer to enhance, among other things, (1) digitizing points on a plurality of anatomical landmarks on the displayed images, (2) determining linear anatomical dimensions of at least a portion of a body of a person in the displayed images using the digitized points and a scale factor for the displayed images, and (3) making an anatomical circumferential measurement prediction of the person based on the determined linear anatomical dimensions and a known morphological relationship. These algorithms and classifiers utilize machine learning techniques to develop a model for distinguishing and measuring various body dimensions. The training data consisting of image data may be annotated by a domain expert (such as a physiologist) and fed into the classifier, and the classifier analyzes the training data to identify patterns in the data that indicate when a given sample corresponds to a known dimensions stored in a database. After the classifier has been trained, a set of similarly annotated validation data is typically used to test the accuracy of the classifier. This type of machine learning is known as ?supervised learning,? since the training and validation data is annotated by a human ?supervisor? or ?teacher.? Unsupervised learning is also contemplated.
The use of machine-learning algorithms allows for a faster, more accurate, and more precise identification than was previously possible. The ability to recognize patterns in data sets that cannot be processed by the human brain or eye is a key factor.
In a first embodiment is provided a machine learning algorithm for computer-assisted prediction of anatomical features. The method includes identifying parameters with a processor in a plurality training images in order to generate a dataset of parameters, the dataset containing data linking parameters to respective images, training and validating at least one machine-learning algorithm using the parameters from the dataset, identifying digitized points in an image of an individual displayed on a touch screen digital display by determining the linear anatomical dimension of at least a part of the body of the person and a scaling factor for the image displayed, and making an anatomical circumferential predictions based on the digitized landmarks on the a a a a a a a a a a a a a a a based on anatomical landmarks based on a a a a a a based on the validated a a a based on the a a a
The system is a machine learning algorithm for computer-assisted prediction. The system includes a memory that stores at least one machine-learning algorithm and datasets. A processor is programmed to (i) identify parameters from a plurality training images and generate a dataset with data linking parameters to respective images. (ii), train the machine-learning algorithm using the parameters of the dataset and validate it. (iii), identify digitized landmarks on an image of a human displayed on a touch screen digital display by using the validated machine-learning algorithm and a scaling factor for the image.
The program causes the processor to execute identifying with a processor parameters in a plurality of training images, generating a training dataset that has data linking the parameters to respective training images, and then training at least one machine learning algorithm based on the parameters in said dataset. Validating this trained machine learning algorithm. The program instructs the processor to perform identifying parameters in a plurality training images in order to generate a dataset of parameters, which contains data linking parameters to training images. Training at least one machine-learning algorithm using the parameters from the dataset and validating it. Identifying digitized points in a plurality anatomical landmarks on an image of a human displayed on a touch screen digital display by using the validated machine-learning algorithm and a scaling factor for the displayed picture. Making an anatomical circumferential predictions
BRIEF DESCRIPTION DES DRAWINGS
FIG. The screen of the device shown in FIG. “Figure 1 shows a step in the postural screening technique where a reference is superimposed on the image to provide vertical, horizontal, and center reference points and where a corresponding line of reference is anchored onto the displayed image of the patient.
FIG. The screen of the FIG. The two reference lines of FIG. The two reference lines in FIG.
FIG. The screen of the device shown in FIG. The two reference lines of FIG. The vertical plane has been aligned by tilting the device toward the patient at the top to level the image capture device.
FIG. The screen of the device shown in FIG. The method of postural screening is shown in Figure 1. Two horizontal lines at the top and bottom of the screen are displayed and the image has then been centered and scaled with the camera by panning.
FIG. The screen of the FIG. The image of the patient is similar to that in FIG. “5 but in the direction the frontal plane the patient.
FIG. “FIG. The image acquired in FIG. Optionally, the image acquired in FIG. 5 is displayed behind an overlay grid of horizontal and vertical lines that allows a qualitative assessment of postural displacement to be observed.
FIG. “FIG. The image acquired in FIG. Optionally, the grid overlay is shown behind which you can observe a qualitative assessment of your postural movement.
FIG. “FIG. 1.
FIG. “FIG. “Figures 1-10 can be used to measure the dimensions of the body, as described in other embodiments.
Click here to view the patent on Google Patents.