US-12616436-B2 - Pulmonary embolism diagnosis support apparatus, pulmonary embolism diagnosis support method, and storage medium

US12616436B2US 12616436 B2US12616436 B2US 12616436B2US-12616436-B2

Abstract

A pulmonary embolism diagnosis support apparatus includes a hardware processor that: obtains a dynamic image of a chest of an examinee captured through radiographic dynamic imaging; analyzes blood flow in the dynamic image to generate blood flow information; generates background lungs information regarding background lungs of the examinee; automatically generates diagnosis support information regarding pulmonary embolism, based on the blood flow information and the background lungs information; and outputs the diagnosis support information regarding pulmonary embolism.

Inventors

Takenori Fukumoto
NORITSUGU MATSUTANI
Yuzo Yamasaki
Kohtaro Abe
Kazuya Hosokawa
Takeshi Kamitani
Tomoyuki Hida
Kousei Ishigami

Assignees

Konica Minolta, Inc.

Dates

Publication Date: 20260505
Application Date: 20230215
Priority Date: 20220218

Claims (14)

1 . A pulmonary embolism diagnosis support apparatus comprising a hardware processor that: obtains a dynamic image of a chest of an examinee captured through radiographic dynamic imaging; analyzes blood flow in the dynamic image to generate blood flow information based on signal values of pixels of regions related to pulmonary blood flow in the dynamic image; generates background lungs information regarding background lungs of the examinee based on the dynamic image or based on an image captured by a modality different from dynamic imaging; wherein the blood flow information includes information for determining whether an abnormality exists in the blood flow, and the background lungs information includes information for determining whether an abnormality other than the blood flow exists in the background lungs; automatically generates diagnosis support information regarding pulmonary embolism, based on the blood flow information and the background lungs information; and outputs the diagnosis support information regarding pulmonary embolism.
2 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein the hardware processor automatically determines whether the blood flow is abnormal, based on the blood flow information, and based on the background lungs information and the determination on whether the blood flow is abnormal, the hardware processor automatically generates the diagnosis support information regarding pulmonary embolism.
3 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein the hardware processor automatically determines whether the background lungs are abnormal, based on the background lungs information, and based on the blood flow information and the determination on whether the background lungs are abnormal, the hardware processor automatically generates the diagnosis support information regarding pulmonary embolism.
4 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein the hardware processor automatically determines whether the blood flow is abnormal, based on the blood flow information, the hardware processor automatically determines whether the background lungs are abnormal, based on the background lungs information, and based on the determination on whether the blood flow is abnormal and the determination on whether the background lungs are abnormal, the hardware processor automatically generates the diagnosis support information regarding pulmonary embolism.
5 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein when the blood flow is abnormal but the background lungs are normal, the hardware processor generates the diagnosis support information that indicates a possibility of pulmonary embolism.
6 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein when (i) the blood flow is abnormal and (ii) a region of the background lungs corresponding to a region of the abnormal blood flow is abnormal, the hardware processor generates the diagnosis support information that indicates a possibility of a disease different from pulmonary embolism.
7 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein when the blood flow is normal, the hardware processor generates the diagnosis support information that indicates no possibility of pulmonary embolism.
8 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein the hardware processor generates the background lungs information, based on one or more frame images included in the dynamic image.
9 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein as the background lungs information, the hardware processor receives (i) an image captured by a modality that performs imaging different from dynamic imaging or (ii) information generated based on the image captured by the modality.
10 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein the hardware processor obtains the background lungs information regarding background lungs of the examinee based on a plain X-ray image captured by a modality different from dynamic imaging.
11 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein wherein the diagnosis support information indicates a possibility of pulmonary embolism when the blood flow information indicates that the blood flow is abnormal and the background lungs information indicates that the background lungs are normal, the diagnosis support information indicates a possibility of a disease different from pulmonary embolism when the blood flow information indicates that the blood flow is abnormal and the background lungs information indicates that a region of the background lungs corresponding to a region of the abnormal blood flow is abnormal, and the diagnosis support information indicates no possibility of pulmonary embolism when the blood flow information indicates that the blood flow is normal.
12 . The pulmonary embolism diagnosis support apparatus according to claim 1 , wherein wherein the hardware processor determines the blood flow is abnormal based on a decrease of blood flow in a region or based on a comparison of values of the blood flow with a predetermined threshold or predetermined range, and wherein the hardware processor determines the background lungs are abnormal based on a finding of a morphological abnormality or based on a comparison of a change amount of a specific component of the background lungs with a predetermined threshold or predetermined range.
13 . A pulmonary embolism diagnosis support method comprising: obtaining a dynamic image of a chest of an examinee captured through radiographic dynamic imaging; analyzing blood flow in the dynamic image to generate blood flow information based on signal values of pixels of regions related to pulmonary blood flow in the dynamic image; generating background lungs information regarding background lungs of the examinee based on the dynamic image or based on an image captured by a modality different from dynamic imaging; wherein the blood flow information includes information for determining whether an abnormality exists in the blood flow, and the background lungs information includes information for determining whether an abnormality other than the blood flow exists in the background lungs; automatically generating diagnosis support information regarding pulmonary embolism, based on the blood flow information and the background lungs information; and outputting the diagnosis support information regarding pulmonary embolism.
14 . A nontransitory computer-readable storage medium storing a program that causes a compute to: obtain a dynamic image of a chest of an examinee captured through radiographic dynamic imaging; analyzing blood flow in the dynamic image to generate blood flow information based on signal values of pixels of regions related to pulmonary blood flow in the dynamic image; generate background lungs information regarding background lungs of the examinee based on the dynamic image or based on an image captured by a modality different from dynamic imaging; wherein the blood flow information includes information for determining whether an abnormality exists in the blood flow, and the background lungs information includes information for determining whether an abnormality other than the blood flow exists in the background lungs; automatically generate diagnosis support information regarding pulmonary embolism, based on the blood flow information and the background lungs information; and output the diagnosis support information regarding pulmonary embolism.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS The entire disclosure of Japanese Patent Applications No. 2022-023408 filed on Feb. 18, 2022 and No. 2022-078594 filed on May 12, 2022 is incorporated herein by reference in its entirety. TECHNICAL FIELD The present disclosure relates to a pulmonary embolism diagnosis support apparatus, a pulmonary embolism diagnosis support method, and a storage medium. DESCRIPTION OF THE RELATED ART Diagnosis of pulmonary embolism involves checking presence of an embolus (e.g., a thrombus) and decrease in pulmonary blood flow caused by the embolus. Known methods and modalities used for diagnosing pulmonary embolism include pulmonary arteriography, contrast-enhanced computed tomography (CT), and scintigraphy of pulmonary ventilation and blood flow. Among these, contrast-enhanced CT has become a standard method. However, contrast-enhanced CT may cause contrast media allergy and radiation exposure (in particular, for expectant mothers and infants). Scintigraphy of pulmonary ventilation and blood flow does not cause contrast media allergy but requires administration of radiopharmaceuticals, by which a patient may be exposed to radiation. As a criterion for diagnosing pulmonary embolism in scintigraphy, a patient is diagnosed with pulmonary embolism if the result of pulmonary ventilation scintigraphy shows no abnormalities but the result of pulmonary blood flow scintigraphy shows a segmental defect, for example. A patient is also diagnosed with pulmonary embolism if the background lungs show no abnormalities in plain X-ray images or in chest CT but the result of pulmonary blood flow scintigraphy shows a segmental defect. JP2020-54580A discloses an apparatus that identifies a thrombus region based on medical images obtained by plain CT. In order to deal with a difficulty of obtaining a large amount of data showing correct regions of a disease, the apparatus learns, as training data, thrombus regions/infarct regions identified in brain CT images of patients (examinees) having a cerebral thrombosis/cerebral infarction. Thus, the apparatus enhances accuracy of identifying a thrombus. Dynamic state diagnosis has attracted attention as a new diagnosis method. Dynamic state diagnosis is advantageous in that (i) it requires lower radiation exposure than CT, (ii) it is as quick as plain X-ray imaging, and (iii) it provides more information than plain X-ray imaging. For example, WO2014091977A1 describes performing blood flow analysis based on a dynamic image. SUMMARY OF THE INVENTION However, researches of the dynamic state for diagnosing thrombi have not progressed sufficiently. WO2014091977A1 only indicates the applicability of dynamic imaging for diagnosing thrombi. In order to increase accuracy in identifying thrombi in dynamic state diagnosis, artificial intelligence (AI) may be applied, as disclosed in JP2020-54580A. However, the blood flow analysis based on a dynamic image does not detect a thrombus itself but shows an abnormality of blood flow at peripheral regions rather than at the thrombi. Therefore, when the result of blood flow analysis indicates a poor blood flow region, it is difficult to identify the cause of the poor blood flow (whether the poor blood flow is caused by a thrombus or by any other disease, such as chronic obstructive pulmonary disease (COPD), pneumothorax, bulla (a disease that causes bloated bubbles of pulmonary alveoli), or interstitial pneumonia). In other words, blood flow analysis may show the same result even if causes (diseases) are different. Even if an AI learns a set of data having correct answers regarding thrombi based on the result of the dynamic state analysis, the AI cannot avoid the possibility of wrongly determining that a patient has a thrombus, based on the blood flow analysis result of the patient having a disease different from a thrombus. It is therefore difficult to increase accuracy in diagnosing a thrombus based only on the blood flow analysis of a dynamic image. As described above, the known methods for diagnosing pulmonary embolism require administration of contrast media/radiopharmaceuticals to patients, which is a burden on the patients. The known methods also require labor-consuming and time-consuming preparation. Further, modalities used in the known methods are expensive. In terms of cost and invasiveness to a patient, such modalities may not be repetitively used for imaging for the purpose of diagnosis. The present invention has been conceived in view of the above issues. Objects of the present invention include enabling diagnosis of pulmonary embolism that lessens burdens on patients, that is quick with a low cost, and that can be performed repetitively. To achieve at least one of the above objects, according to an aspect of the present invention, a pulmonary embolism diagnosis support apparatus includes a hardware processor that: obtains a dynamic image of a chest of an examinee captured through radiographic dynamic imaging; analyz