CN-120976112-B - Auxiliary film reading system suitable for cryptococcus pneumophilus patients

CN120976112BCN 120976112 BCN120976112 BCN 120976112BCN-120976112-B

Abstract

The invention relates to the technical field of auxiliary film reading of lung image marks, in particular to an auxiliary film reading system suitable for cryptococcus pneumonecatrix patients. The method comprises the steps of firstly obtaining continuous CT slice images of the lung, extracting a front Jing Liantong domain of a lung parenchymal region as a communicating domain to be analyzed, further obtaining the vascular probability of the target communicating domain according to the monotonicity of the width change of the target communicating domain, further obtaining the vascular probability of the target communicating domain according to the monotonicity of the area change of the communicating domain to be analyzed at the same position, obtaining the quasi-circular degree of the target communicating domain, and finally fusing the vascular probability and the vascular probability according to the quasi-circular degree and marking the target communicating domain. According to the invention, the probability that the target connected domain belongs to the blood vessel is measured from the angles of width change and area change, and the connected domain is accurately marked by fusing the blood vessel probability and the blood vessel penetration probability by means of the quasi-circular degree, so that the film reading of related personnel is assisted.

Inventors

WEN WEN
YU GUOQING
You Yanjing
YU ZONGYANG
Liu Leihan
ZHENG JINYANG
WANG HUIJUAN
WANG MENG
Zhuang Jihong

Assignees

中国人民解放军联勤保障部队第九〇〇医院

Dates

Publication Date: 20260505
Application Date: 20250718

Claims (8)

1. An assisted reading system for a cryptococcus pneumococcal patient, the system comprising: The image acquisition module is used for acquiring continuous CT slice images of the lung of a current patient, extracting a lung parenchymal region in each CT slice image through image segmentation, extracting a front Jing Liantong domain of the lung parenchymal region as a communication domain to be analyzed, and matching all the communication domains to be analyzed at the same position in the CT slice images; The image analysis module is used for selecting any connected domain to be analyzed as a target connected domain, acquiring the blood vessel probability of the target connected domain according to the monotonicity of the width change of the target connected domain, acquiring the blood vessel penetration probability of the target connected domain according to the monotonicity of the area change of the connected domain to be analyzed, which is matched with the target connected domain, and acquiring the quasi-circular degree of the target connected domain; the image labeling module fuses the blood vessel possibility and the blood vessel penetration possibility according to the quasi-circular degree and marks the target connected domain; the method for acquiring the blood vessel possibility comprises the following steps: The method comprises the steps of taking a connecting line of two farthest pixel points in a target connected domain as a reference line, taking a single pixel point as a step length, moving from one end of the reference line to the other end of the reference line, acquiring the pixel width of the nearest target connected domain in the vertical direction of the reference line after each movement, and constructing a width sequence of the target connected domain; acquiring the vessel probability of the target connected domain according to the monotonicity of the width sequence; The method for acquiring the transvascular likelihood comprises the following steps: acquiring the area of the connected domain to be analyzed, which is matched with the target connected domain, and sequencing according to the acquisition sequence of the CT slice images to obtain an area sequence; and acquiring the possibility of penetrating blood vessels of the target connected domain according to the monotonicity of the area sequence.
2. An assisted reading system for a cryptococcosis patient according to claim 1, wherein the method of obtaining the vessel probability of the target connected domain according to the monotonicity of the width sequence comprises: And acquiring the blood vessel probability of the target connected domain according to the correlation between the width sequence and a sequence number sequence formed by element sequence numbers in the width sequence, wherein the correlation between the width sequence and the sequence number sequence is positively correlated with the blood vessel probability.
3. An assisted interpretation system for cryptococcosis patients as claimed in claim 1, wherein the method for obtaining the transvascular potential of the target connected domain based on monotonicity of the area sequence comprises: And acquiring the possibility of penetrating the blood vessel of the target connected domain according to the number of extreme points at non-two ends in the area sequence, wherein the number of extreme points at the non-two ends is inversely related to the possibility of penetrating the blood vessel.
4. The assisted reading system of claim 1, wherein the method of marking the target connected domain comprises: fusing the degree of quasi-circularity and the blood vessel likelihood to obtain a first normal likelihood, wherein the degree of quasi-circularity is inversely related to the first normal likelihood; Fusing the degree of quasi-circularity and the possibility of penetrating blood vessels to obtain a second normal possibility, wherein the degree of quasi-circularity and the possibility of penetrating blood vessels are positively correlated with the second normal possibility; fusing the first normal probability and the second normal probability to obtain the normal tissue probability of the target communicating domain, wherein the first normal probability and the second normal probability are positively correlated with the normal tissue probability; And marking the target connected domain when the normal tissue possibility is lower than a preset normal threshold value.
5. The assisted reading system of claim 1, wherein the degree of circularity is the circularity of the outermost pixel outline of the communicating region of interest.
6. The assisted reading system of claim 1, wherein said method of matching said connected components to be analyzed at the same location in all said CT slice images comprises: And in the acquisition of two adjacent CT slice images, when the overlapping proportion of the pixel points in the to-be-analyzed connected domain in one frame image and the pixel points in the to-be-analyzed connected domain in the other frame image is larger than a preset threshold value, judging that the corresponding two to-be-analyzed connected domains are matched.
7. An assisted reading system for a cryptococcosis patient according to claim 1, wherein the CT slice image of the pulmonary parenchymal region is extracted by a pre-trained neural network.
8. The assisted reading system of claim 1, wherein the anterior Jing Liantong domain acquisition method comprises: And selecting a region growing algorithm, namely randomly selecting seed points from pixel points with gray values larger than the gray average value of the lung parenchymal region, and obtaining the front Jing Liantong domain.

Description

Auxiliary film reading system suitable for cryptococcus pneumophilus patients Technical Field The invention relates to the technical field of auxiliary film reading of lung image marks, in particular to an auxiliary film reading system suitable for cryptococcus pneumonecatrix patients. Background The lung infection caused by cryptococcus usually has the characteristic of nodular mass, while the characteristic is easy to be identified, but is easy to be misjudged as tumor or tuberculosis, while under the condition of the immune function inhibition of the patient, the lung infection caused by cryptococcus possibly has various satellite type focuses, the morphology of the focuses is changeable, the common diffuse, disseminated-like infiltration, actual transformation and other characteristics, and similar whitish characteristics can be generated due to the generally higher density of lesion tissues and vascular tissues, so that lesion pixels are easy to overlap with the gray scale of pulmonary vascular pixels, and are mutually interfered, and the micro lesions are difficult to distinguish. Disclosure of Invention In order to solve the technical problem that the pathological tissues and vascular tissues of the cryptococcus pneumocandidus are difficult to distinguish, the invention aims to provide an auxiliary film reading system suitable for cryptococcus pneumocandidus patients, and the adopted technical scheme is as follows: The image acquisition module is used for acquiring continuous CT slice images of the lung of a current patient, extracting a lung parenchymal region in each CT slice image through image segmentation, extracting a front Jing Liantong domain of the lung parenchymal region as a communication domain to be analyzed, and matching all the communication domains to be analyzed at the same position in the CT slice images; The image analysis module is used for selecting any connected domain to be analyzed as a target connected domain, acquiring the blood vessel probability of the target connected domain according to the monotonicity of the width change of the target connected domain, acquiring the blood vessel penetration probability of the target connected domain according to the monotonicity of the area change of the connected domain to be analyzed, which is matched with the target connected domain, and acquiring the quasi-circular degree of the target connected domain; And the image labeling module fuses the blood vessel possibility and the blood vessel penetration possibility according to the quasi-circular degree and marks the target connected domain. Further, the method for acquiring the blood vessel possibility comprises the following steps: The method comprises the steps of taking a connecting line of two farthest pixel points in a target connected domain as a reference line, taking a single pixel point as a step length, moving from one end of the reference line to the other end of the reference line, acquiring the pixel width of the nearest target connected domain in the vertical direction of the reference line after each movement, and constructing a width sequence of the target connected domain; And acquiring the vessel probability of the target connected domain according to the monotonicity of the width sequence. Further, the method for acquiring the vessel probability of the target connected domain according to the monotonicity of the width sequence comprises the following steps: And acquiring the blood vessel probability of the target connected domain according to the correlation between the width sequence and a sequence number sequence formed by element sequence numbers in the width sequence, wherein the correlation between the width sequence and the sequence number sequence is positively correlated with the blood vessel probability. Further, the method for obtaining the transvascular likelihood comprises the following steps: acquiring the area of the connected domain to be analyzed, which is matched with the target connected domain, and sequencing according to the acquisition sequence of the CT slice images to obtain an area sequence; and acquiring the possibility of penetrating blood vessels of the target connected domain according to the monotonicity of the area sequence. Further, the method for acquiring the transvascular probability of the target connected domain according to the monotonicity of the area sequence comprises the following steps: And acquiring the possibility of penetrating the blood vessel of the target connected domain according to the number of extreme points at non-two ends in the area sequence, wherein the number of extreme points at the non-two ends is inversely related to the possibility of penetrating the blood vessel. Further, the method for labeling the target connected domain includes: fusing the degree of quasi-circularity and the blood vessel likelihood to obtain a first normal likelihood, wherein the degree of quasi-circularity is inversely related to th