The present disclosure relates to the diagnosis and treatment
planning of patients based on the automatic detection of landmarks belonging to
human organs in diagnostic images (two-dimensional and three-dimensional).
BACKGROUND
[0002] The present invention discloses a method of automatic detection
of the landmarks in medical images for patient’s diagnosis and treatment planning.
The landmark refers to the already defined anatomical location which can be
identified easily on any patient to perform medical analysis. The analysis of
medical images requires landmarks which leads to segmentation of a particular
organ, computation of size of the organ, computation of volume of the organ,
computation of major axis, computation of minor axis, computation of a specific
nodules etc.
[0003] This invention is related to the patient’s diagnosis and treatment
planning. The patient’s diagnosis and treatment planning requires to perform
specific measurements as well as segmentation in two dimensional diagnostic
images such as X-ray, ultrasound, etc. and volumetric segmentation in threedimensional diagnostic images such as CT (Computational Tomography) and MRI
(Magnetic Resonance Imaging) etc. Both types of medical segmentation require
landmark detection for precise segmentation procedure. This invention discloses
the method of automatic landmark detection in body area images.
OBJECTIVE
[0004] A method for detection of landmarks automatically in medical
images.
3
SUMMARY OF THE INVENTION
[0005] According to the some embodiment, the present invention
automatically detect the landmarks on patient’s image based on the sample image
of the same organ.
[0006] Yet according to the some embodiment, the present invention
automatically calibrate and scale the patient’s image and corresponding sample
images.
[0007] Still according to some embodiments, disclosed method is
applicable for the two-dimensional as well as three-dimensional images of the
body organs.
BRIEF DESCRIPTION OF DRAWINGS
[0008] Figure 1 illustrates automatic landmark detection on patient’s
diagnostic image.
[0009] Figure 2 illustrates a system for automatic landmark detection on
patient’s diagnostic image
DETAILED DESCRIPTION
[0010] The present invention provides a method of landmark localization
on the human body parts automatically which are essential for the patient’s
diagnosis and treatment planning. The automatic placement of the landmarks
makes easy the process of patient’s diagnostic image analysis. The automatic
placement of the landmarks also saves the time and efforts of the expert and
radiologist.
[0011] The invention discloses a method of automatic landmark detection
on diagnostic medical images. The invention is presented in Fig. 2 which comprises
4
of a system of detecting landmarks automatically on the medical images. At Step
202, patient’s medical image is given as input through the input device. This input
image can be X-ray image, CT image, MRI, ultrasound etc.
[0012] At step 204, at least one sample image of the input image is already
stored which does not belongs to the same patient. However, this image convers
the same human body area as of the input patient’s image through step 202.
[0013] At step 206, the processing of the detection of landmarks takes
place automatically. The computation of the detection of landmarks is presented
in Fig. 1.
[0014] Atstep 101, the patient’s image along with sample image belonging
to the same region is acquired. The sample image is the already stored image with
known anatomy of the same anatomical region. This image is used for the
transformation of the same information on the real patient’s diagnostic image.
The sample image is the knowledge-base of the system which already has the
same landmarks as needed to detect automatically on the patient’s image.
[0015] At step 103, the calibration of the image is performed over the
sample image. For CT and MRI, calibration can be performed using the dicom tag
values. While X-ray and ultrasound images can be calibrated by giving a value of
DPI of the image.
[0016] At step 105, the scaling of the sample image is performed with
respect to the patient’s image. The scaling of sample image should be performed
in such a way that the resolution of the patient image must be the same as the
corresponding sample image.
[0017] At step 107, the registration of the sample image is performed over
its corresponding patient’s image. The registration is a type of deformable
registration whereas the landmarks of the sample image are transform on the
patient’s image. This transformation is based on the deformation of the sample
5
image into patient image. Contours can be detected automatically on both
patient’s image and sample image. These contours can be considered as the
prominent boundaries of the desired region. As both patient’s image and sample
image have similar region, therefore the contours will be approximately same.
[0018] At step 109, the virtual framework of the contours along with
already available landmarks on the sample images are registered on the patient’s
images. These transformed landmarks can be considered as the landmarks of the
patient’s image.
[0019] The detected landmarks can be used for the further segmentation
of the human body organ and any other analysis of the human body which leads
to the diagnosis and treatment planning.
[0020] At step 208, the landmarks can be visualized on the display device.
The segmentation of the organ can also be performed using the landmarks. The
diagnosis and treatment planning can be performed by visualizing the landmarks
along with the patient’s image. The same landmarks can be stored in the computer
memory for further use such as printing of reports, preparation of datasets and
reporting purposes etc.
[0021] The present invention is disclosed in reference to the medical
images. However, this is an exemplary to explain the invention and must not be
considered as a limitations of medical application. The same disclosure can be
used in other applications.

We claim:
1. A method for detecting landmarks on medical images of a patient for diagnosis
and treatment planning, the method comprising:
acquiring patient’s image and corresponding sample image of the same
body organ;
calibration of the patient’s image;
scaling of the sample image;
sample image registration over patient’s image;
transformation of landmarks from sample image to patient’s image.
2. The method as claimed in claim 1, where scaling of sample image is performed
to make the same calibration factor of sample and patient image.
3. The method as claimed in claim 1, the registration of patient’s image and
sample image uses the contours of anatomical boundaries of the human organ
visible in the image.
4. The method as claimed in claim 1, the transformation of the landmarks is
performed after registration of the anatomical boundaries of the sample and
patient’s image.
5. The method as claimed in claim 1 is applicable to all types of image including
two-dimensional and three-dimensional medical and non-medical images.
6. A system for detecting landmarks on medical images of a patient for diagnosis
and treatment planning, the system comprises:
input device for taking input as sample image and patient’s image;
storage unit to store at least one sample image;
processing unit to perform the computations;
display device to perform the diagnosis and treatment planning;
acquiring patient’s image and corresponding sample image of the same
body organ;
calibration of the patient’s image;
scaling of the sample image;
7
sample image registration over patient’s image;
transformation of landmarks from sample image to patient’s image.
7. The system as claimed in claim 1, where scaling of sample image is performed
to make the same calibration factor of sample and patient image.
8. The system as claimed in claim 1, the registration of patient’s image and
sample image uses the contours of anatomical boundaries of the human organ
visible in the image.
9. The system as claimed in claim 1, the transformation of the landmarks is
performed after registration of the anatomical boundaries of the sample and
patient’s image.
10. The system as claimed in claim 1 is applicable to all types of image including
two-dimensional and three-dimensional medical and non-medical images.