Method for processing images in interventional radioscopy to detect guiding instrumentation equipment

A method for processing images produced by a medical imaging device, the method comprising:

• determining a family of two-dimensional directional filters (27) to apply to a raw image representative of an internal structure of a patient's body, wherein the raw image is obtained from a radiation detector;
• determining for each filter of the family of two-dimensional filters an orientation θi within a range of orientations θi defined beforehand,
• wherein i belongs to an interval [1, n],
• wherein n is equal to a number of filters of the family of two-dimensional filters;
• wherein for each orientation θi, the method further comprises:
• separating a corresponding filter into two mono-dimensional filters:
• a mono-dimensional constant filter along one direction, and
• a mon-dimensional non constant filter along another direction;
• formulating the mono-dimensional constant filter into a discrete mono-dimensional constant filter;
• formulating the discrete mono-dimensional constant filter in a recursive manner;
• successively applying to the raw image the recursive discrete constant filter and the non constant filter of said filter;
• combining all of the responses obtained by an application to the raw image of the filters of the family of two-dimensional filters;
• and
• producing a confidence map and an orientation image from the combined responses of the filters of the family of two-dimensional filters.

The method of claim 14, wherein combining the responses of two selected half-filters into one pixel of the image further comprises:

• determining a discrete line of orientation θi going through a pixel of the image;
• determining a position of a central pixel of one of the two selected half-filters of orientation θi on the discrete line of orientation θi, said central pixel being situated at L/2 pixels before the pixel of the image on the discrete line;
• selecting, at this position of the central pixel on the discrete line, a response G of said corresponding half-filter among the responses stored in the memory;
• determining the discrete line of orientation θj going through the pixel of the image;
• determining a position of a central pixel of the other of the two selected half-filters of orientation θj on the discrete line of orientation θj, said central pixel being situated at L/2 pixels after the pixel of the image on the discrete line;
• selecting, at this position of the central pixel on the discrete line, a response H of said corresponding half-filter among the responses stored in the memory;
• and
• calculating a response of the divided filter to the pixel of the image, by combining the responses G and H according to a function f defined as follows:
• f(G ,H)=(G +H)×α(G+H,G−H) where a is a function defined beforehand.