EP3332354A1 - Dispositif et procédé de détection automatique d'un outil chirurgical sur une image fournie par un système d'imagerie médicale - Google Patents
Dispositif et procédé de détection automatique d'un outil chirurgical sur une image fournie par un système d'imagerie médicaleInfo
- Publication number
- EP3332354A1 EP3332354A1 EP16757322.9A EP16757322A EP3332354A1 EP 3332354 A1 EP3332354 A1 EP 3332354A1 EP 16757322 A EP16757322 A EP 16757322A EP 3332354 A1 EP3332354 A1 EP 3332354A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- image
- window
- surgical tool
- pixels
- determining
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional [3D] objects
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
- G06F18/2113—Selection of the most significant subset of features by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
- G06V2201/034—Recognition of patterns in medical or anatomical images of medical instruments
Definitions
- the present application relates to computer-assisted surgical procedures in which the surgeon performs the procedure with the assistance of images provided by a medical imaging system.
- the present invention more particularly relates to a device and a method for automatic detection of surgical tools on the images provided by the medical imaging system.
- the abdominal cavity of a patient is inflated with carbon dioxide.
- Small incisions are made on the abdominal wall and a trocar is inserted into each incision.
- the number of incisions depends on the type of intervention envisaged and generally varies from 2 to 5.
- Each trocar corresponds to a hollow and sealed tube in which a surgical tool is slidable.
- An endoscope and surgical tools are introduced into the abdominal cavity via the trocars.
- the image provided by the endoscope is captured by a camera and displayed on a display screen.
- an assistant moves the endoscope according to the instructions given by the surgeon.
- the surgeon manipulates the surgical tools in the abdominal cavity according to the intervention to be performed.
- Positioning systems can be used to move the endoscope in place of an assistant.
- These are, for example, the system marketed by Computer Motion under the name Aesop, the system marketed by Armstrong Healthcare under the name EndoAssist or the system marketed by Intuitive Surgical under the name da Vinci Surgical System. .
- US 8391571 discloses a method for determining the position of a surgical tool on video images.
- the positioning system of the endoscope can then be controlled automatically from the determined positions of the surgical tools and, for example, an initial command of the surgeon.
- a disadvantage of the method described in US8391571 is that it requires the prior determination of the position of the incision point of the surgical tool. This can especially be achieved by a probe whose position is identified by a location device or by triangulation from two images provided by the endoscope. This prior step can be complex to implement.
- the present application aims to overcome all or part of the disadvantages of automatic detection methods of surgical tools on images provided by a medical imaging system.
- Another object of an embodiment is not to require the prior determination of the position of the incision of the surgical tool by a localization device.
- the detection of surgical tools is carried out in real time.
- an embodiment provides a method of detecting a surgical tool on a first image comprising the following steps:
- the surgical tool has an elongated shape, the lateral edges of the surgical tool in the third image corresponding to straight lines.
- candidate rights classified according to a ranking order are determined by applying a Hough transformation to the portion of the third image contained in the second window and the side edges are determined from the candidate lines from the shape parameters.
- the second window is rectangular and divided into a first sub-window and a second sub-window
- a first lateral edge corresponds to the candidate right of ranking in the first sub-window and inclined by relative to the major axis of symmetry of the second window is smaller than a first threshold
- a second lateral edge corresponds to the candidate right of best ranking in the second sub-window, whose inclination relative to the major axis of symmetry of the second window is less than the first threshold and whose distance with the first lateral edge is between a second threshold and a third threshold which depend on the width of the second window.
- the first processing comprises binarizing the first image and determining the second image from the determination of the distance map of the binarized image.
- the second processing comprises applying the Frangi algorithm to the first image with a search parameter, representative of the number of pixels of the apparent diameter of tubular structures sought by the Frangi algorithm, less than 10 pixels.
- the first image is a grayscale image obtained from the 'a' or 'b' component of a fourth image acquired by a camera and coded in the CIE Lab color space.
- the first windows are rectangular and the second window is selected by retaining only the first windows whose length is greater than half the width.
- the first image is obtained from a fourth image acquired by a camera, the method further comprising the following steps:
- the method further comprises the following steps:
- the method comprises the following steps:
- Another embodiment provides a device for detecting a surgical tool on a first image, comprising:
- FIG. 1 schematically represents an embodiment of a device for controlling an endoscope positioning system implementing the method for detecting surgical tools
- Figure 2 illustrates, in the form of a block diagram, an embodiment of a method for detecting surgical tools
- FIGS. 3 to 7 are examples of images determined at different stages during the implementation of the embodiment of the method for detecting surgical tools illustrated in FIG. 2;
- Figure 8 illustrates an embodiment of a method of determining the edges of a surgical tool in an image
- FIG. 9 illustrates the relationship between the position of a point of space and the projection of this point on the image plane of a camera; and Figures 10 and 11 illustrate an embodiment of a method for determining the central axis of a surgical tool.
- FIG. 1 represents an embodiment of a device 5 for controlling a positioning system 10 of an endoscope 12 implementing the method for detecting surgical tools according to the invention.
- the positioning system 10 is placed on the abdomen 14 of a patient whose abdominal cavity has been filled with gas.
- the positioning system 10 maintains a trocar 15 containing the endoscope 12 and penetrating into the abdominal cavity through an incision 16.
- a camera 17 is attached to the end of the endoscope 12 outside the endoscope 12. abdominal cavity.
- the positioning system 10 is of the type allowing a displacement of the trocar 15 and the endoscope 12 according to a certain number of degrees of freedom, for example a degree of freedom of translation and two degrees of freedom of rotation.
- the positioning system can directly hold the endoscope 12, and move the trocar 15 through the endoscope 12.
- the images captured by the camera 17 are transmitted to an image acquisition system 20 adapted to display the video images on a display screen 22.
- the acquisition system 20 is connected to an acquisition card 24 by intermediate of a link 26, for example an S-VIDEO cable, or a digital cable, for example an HDMI cable or a USB cable.
- the acquisition card 24 is connected to a processing module 28, via a link 29, for example a Fire-Wire cable (IEEE cable 1394). Alternatively, the acquisition card can be directly integrated in the processing module 28.
- the processing module 28 may comprise a processor, a microprocessor or microcontroller adapted to execute the instructions of a computer program stored in a memory .
- the acquisition card 24 performs a pre-processing of the video images which are transmitted to the processing module 28.
- the processing module 28 is adapted, as is explained in more detail below, to analyze the video images to detect the presence of video. surgical tools on the images.
- the processing module 28 is, moreover, adapted to transmit displacement commands to a control box 30 via a link 32.
- the control unit 30 is adapted to translate the displacement commands transmitted on the link 32 to control signals of the positioning system 10 and to transmit the control signals to the positioning system 10 via a link 34.
- the surgeon can activate or deactivate the control box 30 via a
- the surgeon may provide instructions to the treatment module 28 via a man / machine interface 38 which may include a voice-activated system, a system for detecting the movements of the surgeon's head, and / or a foot control system.
- the positioning system 10 corresponds, for example, to the positioning system described in the patent US8591397.
- the endoscope 12 may have the appearance of a cylindrical tube of axis ⁇ about forty centimeters in length and a few centimeters in diameter.
- the axis ⁇ corresponds, for example, to the optical axis of the endoscope 12 and the camera 17.
- the positioning system 10 is adapted to modify the inclination of the trocar 15 with respect to the center 0, that is, that is to say to modify the inclination of the axis ⁇ relative to the center 0 and to slide the endoscope 12 in the trocar 15 along the axis ⁇ .
- the positioning system 10 previously described makes it possible to move the endoscope 12 according to two degrees of freedom of rotation and a degree of freedom of translation from control signals provided by the control box 30.
- Additional trocars 40, 41 are disposed at incisions 42, 43 of small dimensions made in the abdominal wall 14 of the patient.
- the additional trocars 40, 41 allow the introduction of surgical tools 44, 45, partially shown in Figure 2, in the abdominal cavity.
- the trocars 40, 41 may not be present and the surgical tools 44, 45 may be inserted directly through the incisions 42, 43.
- R Q (0, x, y, Z) a marker, e.g. orthonormal, whose origin 0 corresponds for example to a "fixed point" of the positioning system 10, for example substantially the intersection between the axis ⁇ and the incision 16.
- the axes (Ox) and (Oy) correspond to two axes perpendicular to each other and perpendicular to the axis (Oz) which is perpendicular to the abdominal wall 14 at the level of the incision 16.
- the mark R Q is considered to be fixed relative to the patient during the surgical procedure.
- the camera 17 and the endoscope 12 are, in operation, fixed relative to each other.
- C a point fixed with respect to the camera 17, located on the axis ⁇ , at a distance r from the point 0.
- the angle ⁇ corresponds to the angle between the axis ⁇ and the axis (Oz) and the angle ⁇ corresponds to the angle between the projection of the axis ⁇ in the plane (Oxy) and the axis (Ox).
- the coordinates of point C in the RO frame are (rsinBcoscp, rsinBsincp, rcosB).
- R Q C, ⁇ B, ⁇ g, ⁇ ⁇
- the reference R Q is a fixed reference with respect to the camera 17.
- (X, Y, Z) the coordinates of any point P expressed in the reference
- FIG. 2 illustrates the steps of an embodiment of a method for detecting surgical tools on an image supplied by the camera 17 implemented by the processing module 28 and Figures 3 to 6 show images determined at certain stages of this detection method.
- the method is implemented for each new image IQ acquired by the acquisition system 20 and transmitted to the processing module 28.
- the image IQ is formed in the image plane PI of the camera 17.
- the image IQ is composed of a matrix of pixels.
- step S1 the processing module 28 performs IQ image processing to change how the color data is encoded for each pixel in the image.
- the color data of each image pixel can be encoded in the Red Green Blue or RGB color space.
- the processing module 28 may determine a new image I] _ for which the color data of each pixel are stored in a color space suitable for human perception, e.g. HSV color spaces, CIE Lab, or CIE Luv.
- the processing module 28 converts the image into an image Ig I] _ coded in an independent color space the type of camera 7 used.
- the brightness of the image is coded separately from the actual color of the image.
- the color data is preferably stored as three components including a component that is representative of the luminance of the pixel and two components that are representative of the chrominance of the pixel.
- the components that are representative of the chrominance are therefore independent of the illumination of the scene. The process continues in step 52.
- FIG. 3 represents, in grayscale, the component 'a' of an exemplary image I ] _ in the CIE Lab color space.
- Continuous line 46 corresponds to the physical limit of image I ] _.
- image I ] _ shown in Figure 3 there is a gray area 47, darker than the rest of the image, which corresponds to a surgical tool.
- FIG. 3 shows the reference R PI (Q, ⁇ , ⁇ ).
- step S2 the processing module 28 determines a new image ⁇ 2 by binarization of the image I ] _.
- the components representative of the chrominance of the image I ] _ are used to perform the binarization, for example the components a and b of the CIE Lab color space or the components u and v of the CIE Luv color space.
- only one of the components representative of the chrominance of the image I ] _ is used to perform the binarization.
- An example of a binarization process is Otsu's method. The Otsu method is, for example, described in the publication entitled "A threshold selection method from gray level histograms" by N. Otsu, IEEE Trans.
- the binarization can be performed by determining, for all the pixels of the image I ] _, from the representative chrominance components, a histogram of a representative H function the color of the pixel, according to a number of levels, or color classes, which depends on the precision used for the coding of the function H.
- a threshold is then defined according to the method of Otsu. Pixels for which the function H is below the threshold are then assigned a first color class, for example the lowest color class of the histogram, and the pixels for which the function H is greater than the threshold are assigned a second class.
- the image I2 obtained after the binarization step comprises two pixel classes that can be represented by black pixels and white pixels.
- the surgical tools belong to the first class of pixels while the soft tissues belong to the second class of pixels.
- the white color can be assigned to the pixels of the first class and the black color can be assigned to the pixels of the second class.
- the pixels of the first class form regions in the image that can potentially correspond to surgical tools. These regions are called potential regions afterwards.
- the threshold can be determined from the histogram of the selected color component of the image I ] _.
- FIG. 4 represents an exemplary image ⁇ 2 obtained from the image I ] _ of FIG. 3 after the implementation of steps S1 and S2.
- the image ⁇ 2 includes several white regions 48, some of which are interconnected and only one of which corresponds to a surgical tool. As shown in FIG. 4, the image ⁇ 2 obtained is a very noisy image. The process continues in step S3.
- the module 28 determines a new image I3 by image processing ⁇ 2 so as to remove some regions 48 of the image ⁇ 2 of dimensions too small and which can not correspond to a surgical tool and to isolate the regions 48 which could correspond to a surgical tool.
- the image processing ⁇ 2 may include determining a distance map, also called a distance transform, which associates with each pixel of the image ⁇ 2 the distance to the nearest obstacle point, and the determination of an I3 grayscale image from the distance map.
- the obstacle points are the black pixels.
- the distance may correspond to the Euclidean distance, the distance from Manhattan or the distance from Chebyshev.
- FIG. 5 represents an exemplary image I 3 obtained from the image I2 after the implementation of step S3.
- the image I 3 comprises regions 50 comprising white or grayscale pixels and substantially corresponding to the regions 48 of the image I2 having large areas. In particular, the regions 48 of small dimensions have disappeared, the corresponding pixels now being black.
- the edge 51 of each region 50 of the image I 3 comprises a continuous transition in gray levels from white to black. The process continues in step S4.
- step S4 the module 28 determines, for each region
- each pixel of region 50 which is in gray levels, can be considered a white pixel.
- the outline of each region 51 can be searched in the form of a broken line.
- the module 28 determines, for each region 50 of the image I 3, a window, for example rectangular having the minimum area and whose sides are in contact with the contour of the region 50.
- the module 28 determines and stores data relating to the window F, for example the position of the center of the window in the image I3, the width of the window F, the length of the window F, the area of the window and the orientation of the major axis of symmetry of the window F.
- the module 28 can make a first selection among the windows F thus determined keeping only the windows that meet certain criteria. These criteria are defined in particular according to the expected shape of the surgical tools on the I3 image.
- An example of a criterion is that the ratio between the length and the width of the window F must be greater than two, which corresponds to the fact that the surgical tool has an elongated shape.
- Another example of a criterion is that the area of the window F must be greater than a threshold. This allows you to exclude windows that are too small to match the potential presence of an instrument. For example, the minimum threshold for the area of the window may be 200 pixels.
- the processing module 28 has determined windows of interest F in each of which a surgical tool may be present. The process continues in step S5.
- step S5 the module 18 can implement a second selection among the windows of interest F selected after the first selection.
- the second selection can be made from information from the detection of surgical tools on a previous image.
- Step S5 may not be implemented during the acquisition of the initial images by the acquisition system 20 and may be implemented only when surgical tools are detected on the images provided by In one embodiment, this information may include the position of the insertion point of the surgical tool in the plane of the image and / or the position of the edges of the surgical tool in the implant. image plane which are determined at later stages of the method described hereinafter.
- One criterion may be that the major axis of symmetry of a window of interest F containing a tool The surgical site must pass close to the insertion point of this surgical tool.
- the windows of interest F preserved at the end of step S5 are called selected windows Fg.
- step S5 may not be present. The process continues in step S7.
- the method further comprises a step S6 which can be performed independently and in parallel with the steps S2 to S5 described above.
- the processing module 28 determines a new image I4 by applying to the image I ] _ a treatment which highlights the contours of the elements present in the image I ] _, and in particular the contours of the surgical tools.
- the processing implements the Frangi algorithm.
- Frangi's algorithm is described in particular in the publication by Frangi et al entitled "Multiscale vessel ennhancement filtering" MICCAI '98 Lecture Notes in Computer Science, vol. 1496, PP. 130-137.
- the Frangi algorithm is used to make tubular elements, in particular blood vessels, appear more clearly on an image.
- the Frangi algorithm is not used conventionally for contour detection.
- the Frangi algorithm uses a search parameter that is representative of the number of pixels of the image corresponding to the apparent diameter of the tubular structures sought on the image.
- the inventors have demonstrated that by using the Frangi algorithm with a search parameter corresponding to less than 10 pixels, preferably less than 5 pixels, in particular about 2 pixels, that is to say at a value lower than those conventionally used and less than the apparent diameter of the surgical tools in the image I4, the contours of the elements of the image I3 appear more clearly in the image I4.
- FIG. 7 represents an exemplary image I4 obtained from the image I ] _ after the implementation of step S6.
- step S7 the module 28 searches, for each selected window Fg in step S5, the edges D ] _, D2 of a possible surgical tool in the part of the image I4 determined in step S6 contained in the selected window Fg.
- the detection method is based on the fact that each surgical tool has a shape known in advance which is generally at least partly cylindrical or frustoconical. In this case, the edges of the surgical tool in the I4 image have a rectilinear shape.
- each selected window Fg is divided, according to the major axis of symmetry of the window, into a first subregion and a second subregion. When a surgical tool is present in the portion of the image I4 contained in the window Fg, each sub-window must contain a single edge of the surgical tool.
- Figure 8 schematically shows a window Fg divided into two subwindows F ] _ and F2.
- the angle ⁇ corresponds to the orientation of the window Fg, that is to say the angle between the long side of the window Fg and a preferred direction, for example the axis (Q, ⁇ ⁇ ) of the reference Rp j .
- the width of the small side of the window Fg is denoted L.
- the processing module 28 implements a pattern recognition method in each sub-window F] _ and F2.
- the treatment module 28 implements a method for recognizing lines in each sub-window F ] _ and F2.
- the processing module 28 implements the Hough transform described for example in the book "Use of the Hough Transformation to Detect Lines and Curves in Pictures” by Richard 0. Duda and Peter E. Hart , Communications of the ACM, Vol. 15 (1), pp. 11-15, 1972.
- the Hough transform provides for each pane F] _ and F2 several straight candidates ranked the best candidate to candidate worst.
- the processing module 28 realizes a selection of the candidate lines provided by the Hough transform for the sub-window F ] _.
- Each candidate line of the sub-window F ] _ is defined by an angle a which is the angle between the candidate line of the sub-window F ] _ and the preferred direction.
- the processing mode compares the angle to candidate lines with the angle ⁇ .
- the edge D ] _ of the surgical tool in the sub-window F ] _ corresponds to the candidate line having the best classification and whose angle a satisfies the following relation (1):
- the processing module 28 then performs a selection of the candidate lines provided by the Hough transform for the sub-window F2.
- Each candidate right pane F2 is defined by an angle p which is the angle between the candidate of the right pane F2 and the preferred direction and the mean distance d with the edge D] _ of the sub- window F ] _.
- the processing mode compares the angle p of the candidate lines with the angle ⁇ and compares the distance d with the width L, expressed in numbers of pixels.
- the edge D2 of the surgical tool in the sub-window F2 corresponds to the candidate straight line having the best classification satisfying the following relationships (2):
- the processing module 28 concludes that a surgical tool is present in the selected window.
- the selected windows Fg in which surgical tools are detected are called instrument windows F j .
- the processing module 28 stores data such as the position and orientation of the instrument F j corresponding window and the two edges D] _, D2 of the surgical tool.
- the position and orientation of each instrument window F j can be used by the module 28 in step S5 when processing the next image acquired by the camera 17. The process continues at step S8.
- step S8 the processing module 28 determines the axis of symmetry of revolution A ⁇ D, also called central axis, of each surgical tool in the reference frame RQ.
- the processing module 28 determines the axis of symmetry of revolution A ⁇ D in the reference RQ.
- it is planned, before the surgical procedure, a step of calibration or calibration of the positioning system 10 which consists of determining the passage matrix for obtaining the coordinates (1, m, n) of a point P expressed in the reference RQ from the coordinates (X, Y, Z) expressed in the reference RQ and the relationship between the displacement commands provided by the processing module 28 to the control box 30 and the evolution of the parameters r , ⁇ and ⁇ .
- the processing module 28 is then adapted to determine at any time the position of the reference RQ relative to the reference RQ from the movement commands supplied to the control box 30.
- the reference marks RQ and RQ can be defined differently from what has been previously described. The only condition is that the reference RQ can be considered as being fixed relative to the patient during the intervention, that the reference RQ can be considered as being fixed with respect to the camera during the intervention and that one can determine at any time the position of the reference RQ relative to the reference RQ.
- FIG. 9 diagrammatically shows the marks RQ and RQ whose origins 0 and C are situated on the optical axis ⁇ of the endoscope 12.
- the camera 17 operates according to a pinhole camera model.
- a camera model is described, for example, in the book entitled “Three Dimensional Computer Vision - A Geometry Viewpoint” by Olivier Faugeras, series: Artificial Intelligence, the MIT Press, Cambridge Massachussets, ISBN 0-262-6158- 9 (chapter 3).
- the camera 17 can then be represented by a projection center F and the image plane PI.
- the image plane PI corresponds to the plane in which the images captured by the camera 17 and transmitted to the acquisition system 20 are formed.
- the image plane PI and the projection center F are fixed with respect to the reference mark RQ. Any point P of the space is projected in the image plane PI to form a point I.
- the two-dimensional reference frame Rp j (Q, ⁇ ⁇ , ⁇ B) associated with the image plane PI is represented.
- another camera model than the pinhole model can be used, for example a fisheye camera model.
- any camera model which specifies the transformation associating with each point P of any space the corresponding point I in the PI image plane, can be used.
- a step of calibrating or calibrating the camera 17 is provided. This consists in defining the passage matrix that makes it possible to obtain the coordinates (u, v) of the point I expressed in the reference Rp j to from the coordinates (1, m, n) of the point P expressed in the reference RQ.
- the point I corresponds to the point of intersection between the image plane PI and the straight line passing through the point P and the projection center F.
- the transit matrix is obtained from the coordinates of the point F and the equation of the image plane PI expressed in the reference R Q and which depend on the technical characteristics of the camera 17 used.
- Other camera models can be used.
- the calibration step of the camera 17 can be performed during the surgical procedure by a calibration process called "online" from the analysis of images acquired by the camera 17 to during the surgical operation.
- the processing module 28 can determine, for any coordinate point P (X, Y, Z) in the fixed coordinate system R Q , the coordinates (u, v) of the point I, projection of the point P in the image plane PI, regardless of the position of the camera 17 in the frame R Q.
- FIG. 10 schematically shows the surgical tool 44, the projection center F, the optical axis ⁇ and the image plane PI of the camera 17.
- the edges D are represented. _, D2 of the surgical tool determined in step S7.
- the processing module 28 determines the equation in the reference frame R Q of the plane ⁇ ] _ passing through the edge D] _ and the projection center F and the equation in the frame R Q of the plane ⁇ 2 passing through the edge D2 and the projection center F.
- rfj * the vector normal to the plane
- the surgical tool 44 having a cylindrical shape, the planes ⁇ ] _ and ⁇ 2 are tangent to the tool 44.
- the intersection of the planes ⁇ ] _ and ⁇ 2 is a line D of vector director v and passing through the center of projection F.
- the line D is parallel to the central axis A3p.
- Figure 11 is a sectional view of the surgical tool 40 in a plane perpendicular to the vectors 3 ⁇ 4 and 3 ⁇ 4.
- the tool 44 corresponds to a circle CI tangent to the planes ⁇ ] _ and ⁇ 2 and whose radius r is known.
- the module can then determine the position in the reference RQ of the center of the circle CI which corresponds to the passage point M.
- the central axis A ⁇ D is defined by the waypoint M and the director vector v.
- the processing module 28 determines the projection A2D of the central axis A ⁇ D in the image plane PI of the camera 17.
- the line A2D obtained corresponds to the central axis of the surgical tool expressed in the reference Rp j .
- the axis A2D corresponds to the axis of symmetry of the edges of the surgical tool. It should be noted that the axis A2D thus determined does not correspond, except in special cases, to the projection of the central axis A ⁇ D in the image plane PI of the camera since the central projection does not maintain the proportions between the distances .
- step S9 the processing module 28 determines, for each surgical tool 44, 45, the position in the reference frame RQ of a point Pj ⁇ O 'called insertion point, located substantially in the "center" of the incision 42, 43 through which the surgical tool passes.
- the processing module 28 uses the positions of the central axis A ⁇ D in the reference frame RQ determined in step S8 for at least two different orientations of the surgical tool.
- the insertion point P13D corresponds to the point of intersection of the lines corresponding to the different positions of the axis A ⁇ D thus determined.
- the position of the insertion point is obtained from more than two different orientations of the surgical tool.
- the insertion points Pj ⁇ O of the surgical tools can be used in step S5 when selecting the windows Fg, a window can be selected only if the large axis of symmetry of the window passes close to one of the insertion points.
- the insertion points P 1 of the surgical tools may be used in steps S7 and S8 for the detection of edges D 1, D 2.
- edge detection D 1, D 2 is considered poor if the distance between the associated insertion point P13D and the central axis A D of the surgical tool is greater than a threshold.
- step S9 is not present.
- the method further comprises a step S10 for determining, for each surgical tool, the position Pp2D of the tip of the surgical instrument in the reference Rp j of the image plane PI of the camera 17 from the filtered image I4 determined in step S 6 and A2D projection of the central axis in the image plane PI determined in step S8.
- the axis A2D can, according to one embodiment, the processing module 28 search, among the pixels of the axis A2D, a group of successive pixels, comprising for example at least three successive pixels, having a maximum gray level. The tip of the surgical tool corresponds to the group of pixels thus determined.
- the search for groups of pixels can only be carried out on the parts of the A2D axis that are close, for example to plus or minus 100 pixels, of the sides of the selected window Fg, determined in FIG. step S5, and containing the A2 axis surgical tool ) .
- the processing module 28 implements a method for selecting pixels based on the Otsu method. It may be necessary, in order to more finely determine the position of the tip of the tool, to consider in more detail the pixels of the central axis around the tip of the tool and to apply color constraints to them to determine if indeed, they belong or not to the tool.
- the processing module 28 therefore determined for each surgical tool, the central axis, the edges and the tip. The process continues in step SU.
- step SU the module 28 determines, for each surgical tool, the insertion depth E of the surgical tool relative to the associated passage point M from the position Pp2D of the tip of the surgical tool in the reference Rpj determined in step S10, the position of the point of passage M in the reference frame RQ and the director vector v of the central axis A313.
- the treatment module 28 can then determine, for each surgical tool, the position Pp3D of the tip of the surgical tool in the reference frame RQ from the position of the point of passage M, the depression E and the director vector v according to the following relation (3):
- the vector u which passes through the center C of the camera and the point Pp2D of the image I4, expressed in the reference frame RQ, is determined at from the calibration parameters of the camera 17. It is then determined the intersection between the line passing through the point C and having the vector u as a direction vector and the line passing through the waypoint M and having the vector v as vector director.
- the present invention allows the implementation of complex displacement commands by the processing module 28 for the displacement of the positioning system 10 of the camera 17.
- the processing module 28 can control the positioning system 10 to move the endoscope 12 until the tip of a particular surgical tool is at the center of the image provided by the camera 17.
- the surgeon can provide a control of the "center on identification "to the processing module 28 via the voice control system 38.
- the processing module 28 can then detect the tools present on the image provided by the camera 17. If the tip of the tool corresponding to the identifier provided by the surgeon is present on the image, the processing module 28 then controls the system of positioning 10 to move the endoscope 12 so that the tip of the tool is at a predefined central region of the image.
- the processing module 28 is adapted to control the system. positioning 10 to move the endoscope 12 to the tip of the tool since the position of the insertion point associated with the tool is known. If the tool corresponding to the identifier provided by the surgeon is not present on the image, the control module 28 can control the positioning system 10 to orient the endoscope 12 towards the insertion point associated with the the desired tool. The tool must then be on the image provided by the camera 17 and the centering on the tip of the tool is performed as described above.
- control is to move the endoscope 12 so that the tip of a surgical tool is permanently at the central region of the image provided by the camera 17.
- the surgeon can provide a command of the "follow identifier" type to the processing module 28 via the voice control system 38.
- the processing module 28 controls the positioning system 10 as previously described to center the image provided by the camera 17 on the tip of the tool corresponding to the identifier provided by the surgeon. Subsequently, for each new image received by the processing module 28, the module 28 controls the positioning system 10 to refocus, if necessary, the image on the tip of the tool.
- control is to command an "enlargement" of the image at the tip of a particular tool.
- the surgeon can provide a command of the "zoom on identifier" type to the processing module 28 via the voice control system 38.
- the processing module 28 controls the positioning system 10 as has been described. previously to center the supplied image by the camera 17 on the tip of the tool corresponding to the identifier provided by the surgeon.
- the processing module 28 can then determine the distance between the edge lines associated with the tool considered at the tip of the tool and deduce a value representative of the enlargement of the image.
- the treatment module 28 can then control the positioning system 10 to move the endoscope 12 to increase or reduce such magnification to achieve a predetermined magnification or magnification provided by the surgeon.
- the camera 17 may include an enlargement function (zoom).
- the processing module 28 is then adapted to control the enlargement function of the camera 17 from the detection of the distance between the lines.
- the duration of the processing of an image by the processing module 28 may vary from a few tens of milliseconds to a hundred milliseconds.
- the processing module 28 can therefore receive new images to be processed at a high frequency.
- the method of detecting surgical tools according to the invention can therefore be implemented in real time.
- the present invention thus makes it possible, during a surgical procedure, to relieve the surgeon who does not have to systematically control each movement of the endoscope 12.
- the commands previously described enable the surgeon to devote all his attention to the surgical operation, the processing module 28 controlling parallel movements of the endoscope 12 automatically according to the initial command provided by the surgeon.
- the present invention can also be implemented when the surgical tools do not have a cylindrical shape.
- the preceding embodiment can be directly applied to a tool having an elongated shape and whose projection in a plane comprises straight edges. This is the case for a conical, prismatic, pyramidal shape, etc. More generally, even for tools with more complex shapes, characteristic geometrical parameters of the tools can be memorized at the level of the processing module 28. The steps of the method for detecting the previously described tools can then be adapted according to the characteristic geometrical parameters of each surgical tool.
- the detection method described above allows the detection of the position of the surgical tool, more specifically the determination of the central axis A ⁇ D of a surgical tool, in the reference R Q.
- the surgical tool can be carried by a robot.
- the position of the surgical tool relative to a reference RR linked to the robot is generally known.
- the processing module 28 can thus determine the relation of passage between the marks R Q and RR.
- the robot carrying the surgical tool can thus be separated from the system.
- the geometric model of the robot, from which the position of the tool Surgical can be expressed in the reference R Q can be imprecise.
- the detection of the position of the surgical tool in the reference frame R Q may then advantageously make it possible to compensate at least in part for the inaccuracy of the geometric model of the robot.
- the detection of the position of the surgical tool in the reference mark R Q can to improve the calibration of the arms relative to each other.
- the detection of the position of the surgical tool in the frame R Q can be used to record the movements of the surgical tool directed by the surgeon during the procedure, especially for learning purposes or for the analysis of the surgical procedure.
- the present invention is susceptible of various variations and modifications which will be apparent to those skilled in the art.
- the present invention has been described for surgical procedures, for which the video images are provided by a camera attached to an endoscope, the present invention can be applied to any type of image provided by a camera system.
- medical imaging include, for example, ultrasound images, fluoroscopic images or scanner images.
- the present invention has been described for a particular example of a positioning system, it can be applied to any type of positioning system.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Endoscopes (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR1557508A FR3039910B1 (fr) | 2015-08-04 | 2015-08-04 | Dispositif et procede de detection automatique d'un outil chirurgical sur une image fournie par un systeme d'imagerie medicale |
| PCT/FR2016/052018 WO2017021654A1 (fr) | 2015-08-04 | 2016-08-02 | Dispositif et procédé de détection automatique d'un outil chirurgical sur une image fournie par un système d'imagerie médicale |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP3332354A1 true EP3332354A1 (fr) | 2018-06-13 |
Family
ID=54545276
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP16757322.9A Withdrawn EP3332354A1 (fr) | 2015-08-04 | 2016-08-02 | Dispositif et procédé de détection automatique d'un outil chirurgical sur une image fournie par un système d'imagerie médicale |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US10671872B2 (fr) |
| EP (1) | EP3332354A1 (fr) |
| FR (1) | FR3039910B1 (fr) |
| WO (1) | WO2017021654A1 (fr) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112070789B (zh) * | 2020-08-27 | 2022-06-21 | 电子科技大学 | 一种密集纤维细胞的轮廓估算方法 |
| CN114399550B (zh) * | 2022-01-18 | 2024-06-07 | 中冶赛迪信息技术(重庆)有限公司 | 一种基于三维激光扫描的汽车鞍座提取方法及系统 |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2897703B1 (fr) * | 2006-02-20 | 2008-04-25 | Univ Grenoble 1 | Detection automatique d'un outil chirurgical sur une image fournie par un systeme d'imagerie medicale |
| US9317920B2 (en) * | 2011-11-30 | 2016-04-19 | Rush University Medical Center | System and methods for identification of implanted medical devices and/or detection of retained surgical foreign objects from medical images |
| US9111355B1 (en) * | 2013-03-13 | 2015-08-18 | Hrl Laboratories, Llc | Selective color processing for vision systems that enables optimal detection and recognition |
| US20150170381A1 (en) * | 2013-12-16 | 2015-06-18 | Sony Corporation | Tool localization system with image enhancement and method of operation thereof |
| KR20150077184A (ko) * | 2013-12-27 | 2015-07-07 | 삼성전자주식회사 | 의료 영상의 병변 유사도 판단 장치 및 방법 |
| US9547940B1 (en) * | 2014-09-12 | 2017-01-17 | University Of South Florida | Systems and methods for providing augmented reality in minimally invasive surgery |
| US9905000B2 (en) * | 2015-02-19 | 2018-02-27 | Sony Corporation | Method and system for surgical tool localization during anatomical surgery |
| GB201506842D0 (en) * | 2015-04-22 | 2015-06-03 | Ucl Business Plc And Schooling Steven | Locally rigid vessel based registration for laparoscopic liver surgery |
-
2015
- 2015-08-04 FR FR1557508A patent/FR3039910B1/fr not_active Expired - Fee Related
-
2016
- 2016-08-02 EP EP16757322.9A patent/EP3332354A1/fr not_active Withdrawn
- 2016-08-02 US US15/750,199 patent/US10671872B2/en not_active Expired - Fee Related
- 2016-08-02 WO PCT/FR2016/052018 patent/WO2017021654A1/fr not_active Ceased
Also Published As
| Publication number | Publication date |
|---|---|
| US20180225538A1 (en) | 2018-08-09 |
| US10671872B2 (en) | 2020-06-02 |
| FR3039910A1 (fr) | 2017-02-10 |
| FR3039910B1 (fr) | 2018-08-24 |
| WO2017021654A1 (fr) | 2017-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1987473B1 (fr) | Detection automatique d'un outil chirurgical sur une image fournie par un systeme d'imagerie medicale | |
| Shvets et al. | Automatic instrument segmentation in robot-assisted surgery using deep learning | |
| US20220125280A1 (en) | Apparatuses and methods involving multi-modal imaging of a sample | |
| US9597009B2 (en) | Marker-based tool tracking | |
| Crihalmeanu et al. | Enhancement and registration schemes for matching conjunctival vasculature | |
| AU2010202258B2 (en) | System and method for classifying follicular units | |
| US9601025B2 (en) | Method and system for automatic tool position determination for minimally-invasive surgery training | |
| Saint-Pierre et al. | Detection and correction of specular reflections for automatic surgical tool segmentation in thoracoscopic images | |
| Doignon et al. | Real-time segmentation of surgical instruments inside the abdominal cavity using a joint hue saturation color feature | |
| Mahmood et al. | DSRD-Net: Dual-stream residual dense network for semantic segmentation of instruments in robot-assisted surgery | |
| JP2014166298A (ja) | 内視鏡用画像処理装置、内視鏡装置、画像処理方法及び画像処理プログラム | |
| US9721191B2 (en) | Method and system for image recognition of an instrument | |
| CA2933684A1 (fr) | Suivi sans marqueur d'outils chirurgicaux robotises | |
| Pakhomov et al. | Towards unsupervised learning for instrument segmentation in robotic surgery with cycle-consistent adversarial networks | |
| CN115363750B (zh) | 一种手术机器人速度限制方法及系统 | |
| WO2019028021A1 (fr) | Système et méthode de localisation basée sur la vision artificielle et un matériel hybride | |
| Nosrati et al. | Simultaneous multi-structure segmentation and 3D nonrigid pose estimation in image-guided robotic surgery | |
| Zhou et al. | Visual tracking of laparoscopic instruments | |
| Speidel et al. | Automatic classification of minimally invasive instruments based on endoscopic image sequences | |
| EP3332354A1 (fr) | Dispositif et procédé de détection automatique d'un outil chirurgical sur une image fournie par un système d'imagerie médicale | |
| Doignon et al. | Detection of grey regions in color images: application to the segmentation of a surgical instrument in robotized laparoscopy | |
| US20040247183A1 (en) | Method for image analysis | |
| WO2022153673A1 (fr) | Dispositif dépilatoire et procédé d'épilation | |
| Sogabe et al. | Detection of Instruments Inserted into Eye in Cataract Surgery Using Single-shot Multibox Detector. | |
| WO2007110982A1 (fr) | Analyse d'image et programme pour une image stéréo de fond de l'œil |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20180206 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| AX | Request for extension of the european patent |
Extension state: BA ME |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| 17Q | First examination report despatched |
Effective date: 20200827 |
|
| RAP1 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE Owner name: INSTITUT NATIONAL DE LA SANTE ET DE LA RECHERCHE MEDICALE Owner name: UNIVERSITE GRENOBLE ALPES |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
| 18D | Application deemed to be withdrawn |
Effective date: 20240618 |