Molecular block-matching method for gel image analysis

A method for gel image analysis implemented in a medical system with one or more processors, said method comprising:

• (a) receiving the approximate coordinates of a first protein spot on a first gel image;
• (b) identifying the perimeter of said first protein spot using a wave search based on the pixel-intensity of the area surrounding said approximate coordinates;
• (c) identifying the center coordinates of said first protein spot using a Levenberg-Marquardt fitting algorithm to model pixel intensity within said perimeter of said first protein spot, said Levenberg-Marquardt fitting algorithm based on a Gaussian function;
• (d) locating a second protein spot on a second gel image using a molecular block-matching method, said second protein spot represents the same protein as said first protein spot, and said molecular block-matching method comprises:
• i. centering a first block on said first gel image at the center coordinates of said first protein spot,
• ii. centering a second block on said second gel image at an initial location corresponding to the location of said first protein spot on said first gel image,
• iii. shifting said second block in single pixel increments in a widening spiral around said initial location up to a maximum displacement from said initial location,
• iv. comparing the images within said first and second blocks prior to each incremental shift of said second block using Pearson's correlation as a block comparison criterion, in order to determine the closest matching second block to said first block, and
• v. assigning the center location of said closest matching second block as the approximate coordinates of said second protein spot;
• (e) identifying the perimeter of said second protein spot using a wave search based on the pixel-intensity of the area surrounding said approximate coordinates of said second protein spot;
• and
• (f) identifying the center coordinates of said second protein spot using a Levenberg-Marquardt fitting algorithm to model pixel intensity within said perimeter of said second protein spot, said Levenberg-Marquardt fitting algorithm based on a Gaussian function,
• whereby the magnitude and direction of displacement of said second protein spot relative to said first protein spot can be ascertained.