Medical image processing apparatus and medical image processing method
To provide a medical image processing apparatus and a medical image processing method enabling a reduction in burden of re-imaging and re-interpretation of a medical image. The medical image processing apparatus for processing a medical image includes: a detection section that detects, from the medical image, a candidate region which is a region including a lesion candidate or target tissue, and that calculates an identification score for the candidate region; a designated region acquisition section that acquires a designated region which is a region designated by an operator; and a redetection section that calculates an identification score for the designated region based on the designated region and multiple predefined regions which are regions used in a process of detecting the candidate region.
1. A medical image processing apparatus for processing a medical image, comprising:
a processor that, when executing at least one program, configures the processor to:
detect, from the medical image, a candidate region which is a region including a lesion candidate or target tissue, and calculate an identification score for the candidate region;
acquire a designated region which is a region designated by an operator;
calculate an identification score for the designated region based on the designated region and multiple predefined regions which are regions used in a process of detecting the candidate region;
calculate a degree of similarity between the designated region and each of the multiple predefined regions, and calculate an identification score for the designated region based on the degree of similarity and an identification score for each of the multiple predefined regions; and
calculate the identification score for the designated region as Σ(W i ·s i )/Σ(W i ) where s i is an identification score estimated by convolution processing for each of the multiple predefined regions, and W i is the degree of similarity.
2. The medical image processing apparatus according to claim 1 , wherein the processor is further configured to use only the identification score s i for a predefined region with the degree of similarity W i at or above a predetermined threshold value.
3. The medical image processing apparatus according to claim 1 , wherein the designated region is generated by deforming the candidate region.
4. The medical image processing apparatus according to claim 1 , wherein the designated region has any one of a plurality of shapes including a rectangular shape, an elliptic shape, a circular shape, a polygonal shape, and a closed region by a free form curve.
5. The medical image processing apparatus according to claim 1 , wherein the identification score is an index including at least one of a degree of abnormality, a degree of malignancy, a likelihood of tissue classification, and a likelihood of lesion category.
6. A medical image processing method that is performed by a medical image processing apparatus for processing a medical image, comprising the steps of:
detecting, from the medical image, a candidate region which is a region including a lesion candidate or target tissue, and calculating an identification score for the candidate region;
acquiring a designated region which is a region designated by an operator;
calculating an identification score for the designated region based on the designated region and multiple predefined regions which are regions used in a process of detecting the candidate region;
calculating a degree of similarity between the designated region and each of the multiple predefined regions, and calculating an identification score for the designated region based on the degree of similarity and an identification score for each of the multiple predefined regions; and
calculating the identification score for the designated region as Σ(W i ·s i )/Σ(W i ) where s i is an identification score estimated by convolution processing for each of the multiple predefined regions, and W i is the degree of similarity.