Patient motion analysis for behavior identification based on video frames with user selecting the head and torso from a frame

A method comprising:

• obtaining video information of patient motion captured over a period of time, wherein the video information comprises a plurality of frames;
• receiving, with respect to one frame of the plurality of frames, a user selection of a sample area within the one frame of the plurality of frames, the sample area being representative of an anatomical region, wherein the user selection is defined by a user;
• analyzing, by one or more processors, each of the other plurality of frames for respective areas corresponding to the sample area;
• calculating, by the one or more processors, one or more movement parameters of the anatomical region during the period of time from at least one difference between the sample area and one or more respective areas of at least a subset of the plurality of frames, wherein calculating the one or more movement parameters comprises:
• generating raw motion pixels for each of the other plurality of frames, wherein the raw motion pixels are a difference between pixels of each of the other plurality of frames and its respective prior frame;
• for each raw motion pixel of each of the other plurality of frames, determining a gray intensity value;
• for each raw motion pixel of each of the other plurality of frames, comparing the respective gray intensity value to a gray intensity threshold;
• identifying raw motion pixels greater than the gray intensity threshold as intensity motion pixels;
• and
• generating a motion track map comprising the intensity motion pixels of each of the other plurality of frames;
• for each of the intensity motion pixels of each of the other plurality of frames, determining the intensity motion pixels occurring in a continuous or an adjacent area of a subsequent frame;
• identifying the determined intensity motion pixels as real motion pixels representative of movement of the anatomical region;
• updating the motion track map to include only the identified real motion pixels;
• for each of the other plurality of frames, determining a center of the corresponding areas representing the anatomical region;
• for each of the other plurality of frames, calculating, based on movement of the center of the corresponding area from a prior frame of the plurality of frames, a velocity of the anatomical region as a value of a first movement parameter of the one or more movement parameters;
• and
• for each of the other plurality of frames, calculating a ratio of a distance the center of the corresponding area moved from the prior frame to a distance a center of the anatomical region moved during the period of time, wherein the ratio is a value of a second movement parameter of the one or more movement parameters;
• comparing, by the one or more processors, the one or more movement parameters of the period of time to respective criteria for each of a plurality of predetermined patient behaviors;
• identifying, based on the comparison and by the one or more processors, each one of the predetermined patient behaviors that occurred during the period of time;
• and
• outputting an indication of the patient behaviors that occurred during the period of time.

A method comprising:

• obtaining video information of patient motion captured over a period of time, wherein the video information comprises a plurality of frames;
• receiving, with respect to one frame of the plurality of frames, a user selection of a sample area within the one frame of the plurality of frames, the sample area being representative of an anatomical region, wherein the user selection is defined by a user;
• analyzing, by one or more processors, each of the other plurality of frames for respective areas corresponding to the sample area;
• calculating, by the one or more processors, one or more movement parameters of the anatomical region during the period of time from at least one difference between the sample area and one or more respective areas of at least a subset of the plurality of frames, wherein calculating the one or more movement parameters of the anatomical region comprises, for each of at least a subset of the plurality of frames:
• applying a fast Fourier transform to the real motion pixels of the frame to generate amplitude spectrums in a frequency domain;
• selecting three frequencies having the highest amplitude spectrums in the frequency domain;
• selecting, based on the comparison, one of the three frequencies that falls within a predetermined range of frequencies and has the highest amplitude spectrum of the three frequencies;
• and
• storing the selected one of the three frequencies as a detected frequency for the respective frame, wherein the detected frequency is a value of one of the one or more movement parameters;
• comparing, by the one or more processors, the one or more movement parameters of the period of time to respective criteria for each of a plurality of predetermined patient behaviors;
• identifying, based on the comparison and by the one or more processors, each one of the predetermined patient behaviors that occurred during the period of time, wherein identifying each one of the predetermined patient behaviors comprises:
• determining an occurrence with which the subset of the plurality of frames comprises the detected frequency;
• determining that the occurrence is greater than an occurrence threshold;
• determining that an uncertainty of the respective areas corresponding to the anatomical region occurred at a frequency less than an uncertainty occurrence threshold;
• and
• in response to determining that the occurrence is greater than the occurrence threshold and determining that the uncertainty occurred at the frequency less than the uncertainty occurrence threshold, identifying tremor for the anatomical region during the period of time;
• and
• outputting an indication of the patient behaviors that occurred during the period of time.