Computer program, information processing device, and information processing method

Abstract

This computer program causes a computer to execute processes of: acquiring a reflected wave of an ultrasound wave emitted onto the body surface of a subject via an ultrasound probe that can be attached to and detached from the body surface of the subject and is pasted on the body surface; generating an ultrasound image on the basis of the acquired reflected wave; and generating diagnosis assistance information relating to a neuromuscular disease of the subject on the basis of the generated ultrasound image.

Description

Computer program, information processing device, and information processing method

[0001] The present invention relates to a computer program, an information processing device, and an information processing method.

[0002] Amyotrophic lateral sclerosis (ALS) is a neuromuscular disease that involves motor neurons in the brain and spinal cord. It is said that patients will require the use of a ventilator or die within 2 to 5 years of onset, so early diagnosis is essential. Furthermore, clinical diagnosis of ALS is often symptomatic, making a definitive diagnosis difficult.

[0003] Fascicles are considered important diagnostic indicators of lower motor neuron disorders. In recent years, the gold standard for diagnosing ALS has been needle electromyography (EEG), which detects fasciculation potentials, positive sharp waves (PSWs), or fibrillations (Fibs). However, needle EMG can only provide information from 2-3 mm around the needle. Furthermore, needle EMG is painful and highly invasive.

[0004] Patent Document 1 discloses an ultrasound analysis device that receives ultrasound waves transmitted from multiple positions on the surface of a subject and reflected internally, generates an ultrasound image based on the received signals, and calculates muscle-related indices.

[0005] The use of muscle echo as in Patent Document 1 has attracted attention because it allows for wide and deep observation of fasciculation and has high sensitivity for detecting fasciculation.

[0006] International Publication No. 2020 / 039796

[0007] However, conventional muscle ultrasound is limited by time constraints, such as the fact that the examination staff must manually hold the ultrasound probe while observing, meaning that observations can only be made for about 30 to 60 seconds per muscle. This means that evaluation is limited to the presence or absence of fasciculation, making quantitative evaluation difficult. Furthermore, it can only be performed at medical institutions with staff capable of performing the procedure, which hinders its use in diagnosing neuromuscular diseases.

[0008] The present invention has been made in view of the above circumstances, and aims to provide a computer program, an information processing device, and an information processing method that enable ultrasound examination to be utilized in the diagnosis of neuromuscular diseases.

[0009] The present application includes multiple means for solving the above-mentioned problems. As an example, a computer program causes a computer to execute the steps of acquiring reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto a body surface of a subject via an ultrasound probe detachable from the body surface of the subject and attached to the body surface of the subject; generating an ultrasound image based on the acquired reflected wave data; and generating diagnostic support information regarding the neuromuscular disease of the subject based on the generated ultrasound image.

[0010] According to the present invention, ultrasound examination can be utilized for the evaluation of neuromuscular diseases in a timely manner with fewer limitations due to human factors.

[0011] FIG. 1 is a diagram showing an example of the configuration of an information processing device of this embodiment. FIG. 2 is a diagram showing a first example of generation of diagnostic support information using a learning model. FIG. 3 is a diagram showing a second example of generation of diagnostic support information using a learning model. FIG. 4 is a diagram showing a fourth example of generation of diagnostic support information using a learning model. FIG. 5 is a diagram showing an example of display of diagnostic support information. FIG. 6 is a diagram showing an example of evaluation of a candidate drug for clinical trial. FIG. 7 is a diagram showing an example of processing procedures by an information processing device.

[0012] An embodiment of the present invention will now be described with reference to the drawings. Fig. 1 is a diagram showing an example of the configuration of an information processing device 50 according to this embodiment. The information processing device 50 may be configured, for example, as a personal computer or a tablet terminal. An ultrasound probe 10 can be connected to the information processing device 50.

[0013] The ultrasonic probe 10 is configured to be detachably attached to a human body surface. The ultrasonic probe 10 is, for example, sheet-shaped, thin, and flexible, and can be attached to the body surface of a subject, such as a patient suspected of having amyotrophic lateral sclerosis (ALS). The ultrasonic probe 10 includes, for example, an ultrasonic array sensor 11 configured by arranging a plurality of ultrasonic transducers in a row or matrix. Specifically, thin members such as synthetic resin, metal plate, conductive resin, flexible substrate, and silicone resin are laminated to cover the ultrasonic array sensor 11. For example, piezoelectric ceramic (PZT: lead zirconate titanate) can be used as the ultrasonic transducer.

[0014] In needle electromyography, only information from 2 to 3 mm around the needle can be obtained, but the ultrasound probe 10 of this embodiment can be attached while being in close contact with the body surface of the subject, and can irradiate ultrasound over a wider area than with a needle, making it possible to observe multiple muscles simultaneously. Note that the shape and structure of the ultrasound probe 10 illustrated in Figure 1 are merely an example, and are not limited to the example in Figure 1 as long as it can be attached to the body surface of the subject and the area of ​​the part in close contact with the body surface is sufficiently large.

[0015] The ultrasonic probe 10 may have any shape as long as it can be attached to and detached from the human body surface, and may be inflexible and have a main body containing an ultrasonic transducer and a fixing part for fixing the main body to the human body surface. The fixing part may be, for example, a sheet including a hook-and-loop fastener that can be connected to the main body, or an adapter with a seal that can be attached to the body surface while the main body is housed therein.

[0016] The ultrasound probe 10 irradiates ultrasound from the body surface of the subject, receives ultrasound reflected from the inside of the body surface, and outputs an electrical signal (echo signal) representing the received ultrasound to the information processing device 50. The ultrasound probe 10 may also perform analog-to-digital conversion of the electrical signal and output the converted signal to the information processing device 50.

[0017] In this embodiment, for example, a doctor using the information processing device 50 can attach the ultrasound probe 10 to the body surface of the subject (for example, the skin on the surface of the biceps brachii) and observe the condition of the subject's muscles for a relatively long period of time (for example, about 10 minutes to 1 hour).

[0018] The information processing device 50 includes a control unit 51 that controls the entire device, an interface unit 52 , an ultrasound image generating unit 53 , a display unit 54 , an operation unit 55 , a memory 56 , a storage unit 57 , and a communication unit 58 .

[0019] The control unit 51 may be configured by incorporating a required number of central processing units (CPUs), micro-processing units (MPUs), graphics processing units (GPUs), etc. The control unit 51 may also be configured by combining digital signal processors (DSPs), field-programmable gate arrays (FPGAs), etc.

[0020] The interface unit 52 provides an interface function with the ultrasound probe 10. The interface unit 52 acquires electrical signals (echo signals) from the ultrasound probe 10. That is, the interface unit 52 acquires reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto the body surface of the subject via the sheet-like ultrasound probe 10 attached to the body surface of the subject. The interface unit 52 outputs the acquired electrical signals to the ultrasound image generation unit 53.

[0021] The ultrasound image generator 53 generates an ultrasound image based on the acquired electrical signal. The ultrasound image represents the shade of pixels on the image according to the intensity of the electrical signal (the reflected ultrasound wave). The ultrasound image is a relatively long video (for example, about 10 minutes to 1 hour). The ultrasound image may be a two-dimensional image or a three-dimensional image.

[0022] The display unit 54 can be configured with a liquid crystal panel, an organic EL (Electro Luminescence) display, etc. The information processing device 50 may not have the display unit 54, and an external display device of the information processing device 50 may function as the display unit 54. The display unit 54 can display the ultrasound image generated by the ultrasound image generation unit 53.

[0023] The operation unit 55 is configured with a touch panel or the like, and can perform operations such as selecting an icon displayed on the display unit 54, moving a cursor, and inputting characters, etc. The operation unit 55 may be a mouse or a keyboard.

[0024] The communication unit 58 includes a communication module and has the function of communicating with external devices via a communication network.

[0025] The storage unit 57 may be configured with a semiconductor memory, a hard disk, or the like, and stores a computer program (program product) 61, a learning model 62, and required information. The required information includes, for example, processing results by the information processing device 50.

[0026] The computer program 61 can be stored in the storage unit 57 by reading the computer program 61 recorded on a recording medium (e.g., an optically readable disk storage medium such as a CD-ROM) M using a recording medium reading unit (not shown). The computer program 61 may be read from a recording medium such as a storage device (semiconductor memory such as a solid-state drive (SSD)) connected by a standard for connecting to a computer (e.g., USB (Universal Serial Bus) or other standard) and stored in the storage unit 57. The computer program 61 may also be downloaded from an external device via the communication unit 58 and stored in the storage unit 57. Details of the learning model 62 will be described later.

[0027] The memory 56 can be configured with semiconductor memory such as SRAM (Static Random Access Memory), DRAM (Dynamic Random Access Memory), flash memory, etc. The control unit 51 executes a computer program 61 to perform various functions. The control unit 51 can execute processes defined by the computer program 61.

[0028] The control unit 51 acquires reflected waves of ultrasound irradiated onto the body surface of the subject via the sheet-like ultrasound probe 10 attached to the body surface, and generates an ultrasound image based on the acquired reflected waves. The control unit 51 can generate diagnostic support information for the neuromuscular disease of the subject based on the generated ultrasound image. Note that in the process of generating diagnostic support information, the subject is assumed to be a patient suspected of having ALS. Furthermore, as described below, in the machine learning process for generating the learning model 62, the subject includes not only patients suspected of having ALS, but also patients definitively diagnosed with ALS, patients confirmed to be non-ALS, and healthy individuals.

[0029] As described above, according to this embodiment, compared to needle electromyography, a physician using the information processing device 50 can observe the muscle condition of a subject over a period of time (e.g., approximately 10 minutes to 1 hour), eliminating time constraints. While typical muscle echography only allows observation for approximately 30 to 60 seconds, a period of time refers to a longer period than the observation time required for muscle echography. Furthermore, because the muscle condition can be observed simply by attaching the ultrasound probe 10 to the subject's body surface, a user can observe the subject's muscles even without skilled staff, eliminating limitations that hinder its use in diagnosing neuromuscular diseases. This allows physicians to utilize ultrasound examinations in diagnosing neuromuscular diseases without being restricted by human factors. Furthermore, by simultaneously attaching multiple ultrasound patches, ultrasound examinations can be used in diagnosing neuromuscular diseases in a relatively short period of time.

[0030] Fascicles observed on ultrasound images are so-called muscle twitches, and are relatively small, involuntary muscle contractions and relaxations that can be observed locally under the skin. Fascicles are not only observed in patients suspected of ALS, but also benign ones that occur in healthy people and in neuromuscular diseases other than ALS.

[0031] The fasciculation seen in non-ALS patients is relatively transient, exhibits a relatively uniform pattern, occurs infrequently, and is observed in muscles nonspecific to ALS, such as the biceps brachii. In contrast, the fasciculation seen in suspected ALS patients is relatively persistent, exhibits a more complex and irregular pattern, and occurs more frequently. Furthermore, the fasciculation seen in suspected ALS patients occurs in multiple specific muscles, such as the tongue, sternocleidomastoid, paraspinal muscles, and trapezius.

[0032] When the control unit 51 acquires an ultrasound image from the ultrasound image generation unit 53, it can detect fasciculations (muscle fasciculations) based on the ultrasound image and generate diagnostic support information for the subject's neuromuscular disease based on the detected fasciculations. For example, the control unit 51 compares movement pattern data indicating the muscle movement pattern shown in the ultrasound image with one or more reference pattern data indicating muscle fasciculations, and when it detects reference pattern data whose similarity to the movement pattern data is equal to or greater than a threshold, it generates diagnostic support information indicating the possibility of a neuromuscular disease.

[0033] The reference pattern data is data obtained by analyzing the movement pattern of a muscle in which fasciculations have occurred using optical flow, the data having been generated in advance and stored in memory. The movement pattern data is data obtained by analyzing ultrasound images using optical flow. The similarity between the movement pattern data and the reference pattern data is expressed, for example, by the correlation value between these data. The algorithm is not limited to an abnormality detection method such as optical flow, and this method is merely one example of a detection method.

[0034] The control unit 51 may use a machine learning model to detect fasciculations. Specifically, the control unit 51 can input the ultrasound image to a learning model 62 that generates diagnostic support information for neuromuscular diseases, thereby generating diagnostic support information for the subject's neuromuscular disease. More specifically, the control unit 51 can detect fasciculations based on the ultrasound image and generate diagnostic support information for the subject's neuromuscular disease based on the detected fasciculations. In this embodiment, the learning model 62 can detect fasciculations based on the ultrasound image and generate diagnostic support information for the neuromuscular disease based on the detected fasciculations. The learning model 62 can extract fasciculations as features from the ultrasound image and determine whether or not ALS is suspected based on the temporal changes in the extracted features and the regions (muscles) from which the features were extracted.

[0035] According to this embodiment, fasciculation is detected based on ultrasound images observed over a relatively long period of time using the ultrasound probe 10. This makes it possible to monitor the occurrence of fasciculation in a subject over a long period of time, and quantitatively evaluate the number of fasciculations detected over a relatively long period of time, etc. This makes it possible to generate diagnostic support information including quantitative evaluations for neuromuscular diseases such as ALS.

[0036] Next, the generation of diagnostic support information by the learning model 62 will be described.

[0037] 2 is a diagram showing a first example of generation of diagnostic assistance information by the learning model 62. As shown in Fig. 2, when the control unit 51 inputs an ultrasound image generated by the ultrasound image generation unit 53 to the learning model 62, the learning model 62 generates diagnostic assistance information and outputs the generated diagnostic assistance information. In the first example illustrated in Fig. 2, the diagnostic assistance information is information regarding the possibility of ALS, and for example, at least one of "possibility of ALS" and "low possibility of ALS" can be output.

[0038] The learning model 62 can use algorithms such as CNN (Convolutional Neural Network), LSTM (Long Short-Term Memory), CNN-LSTM, and RNN (Recurrent Neural Network).

[0039] The generation of the learning model 62 of the first example, i.e., machine learning, can be performed, for example, as follows: First, the control unit 51 collects first training data including ultrasound images of multiple patients who have been definitively diagnosed as having a possibility of ALS and the diagnosis result "possible ALS." Next, the control unit 51 collects second training data including ultrasound images of multiple patients who have been definitively diagnosed as having a low possibility of ALS and the diagnosis result "low possibility of ALS."

[0040] When the control unit 51 inputs an ultrasound image of a patient who has been definitively diagnosed as having a possibility of ALS based on the first training data into the learning model 62, the control unit 51 can update the parameters of the learning model 62 so that the learning model 62 outputs a diagnosis result of "possible ALS." Furthermore, when the control unit 51 inputs an ultrasound image of a patient who has been definitively diagnosed as having a low possibility of ALS based on the second training data into the learning model 62, the control unit 51 can update the parameters of the learning model 62 so that the learning model 62 outputs a diagnosis result of "low possibility of ALS."

[0041] As described above, when the control unit 51 acquires an ultrasound image from the ultrasound image generation unit 53, the control unit 51 can input the ultrasound image to the learning model 62 that generates diagnostic support information including information about the possibility of ALS, and generate diagnostic support information including information about the possibility of ALS in the subject. The diagnosis of ALS depends largely on symptoms, and a definitive diagnosis is not easy, but the information processing device 50 can provide diagnostic support information as to whether or not a patient is likely to have ALS when a doctor or the like makes a definitive diagnosis, thereby assisting the doctor in making a definitive diagnosis.

[0042] 3 is a diagram showing a second example of generation of diagnostic assistance information by the learning model 62. In the second example illustrated in FIG. 3, the diagnostic assistance information is the number of times fasciculation is detected, and for example, the learning model 62 can output the number of times fasciculation is detected in a predetermined time (e.g., 10 minutes, 20 minutes, 30 minutes, 1 hour, etc.). Note that instead of the number of times fasciculation is detected, the frequency of fasciculation may also be used. When fasciculation is detected multiple times within a predetermined time, the frequency can be calculated as {frequency = number of detections / predetermined time}.

[0043] The learning model 62 of the second example, i.e., machine learning, can be generated, for example, as follows. First, the control unit 51 collects first training data including ultrasound images of multiple patients who have been definitively diagnosed with ALS and the number of times fasciculation has been detected in the ultrasound images (teacher data). Next, the control unit 51 collects second training data including ultrasound images of multiple patients who have been definitively diagnosed with non-ALS and the number of times fasciculation has been detected in the ultrasound images (teacher data).

[0044] When an ultrasound image of a patient who has been definitively diagnosed with ALS is input to the learning model 62 based on the first training data, the control unit 51 can update the parameters of the learning model 62 so that the learning model 62 outputs the "number of times fasciculation was detected" to generate the learning model 62. Furthermore, when an ultrasound image of a patient who has been definitively diagnosed with non-ALS is input to the learning model 62 based on the second training data, the control unit 51 can update the parameters of the learning model 62 so that the learning model 62 outputs the "number of times fasciculation was detected" to generate the learning model 62.

[0045] As described above, when an ultrasound image is input, the control unit 51 inputs the ultrasound image generated by the ultrasound image generation unit 53 into the learning model 62, which generates diagnostic support information including the number of times fasciculation was detected, and can generate diagnostic support information including the number of times fasciculation was detected in the subject. Diagnosis of ALS depends largely on symptoms, and a definitive diagnosis is not easy, but the information processing device 50 can provide diagnostic support information including the number of times fasciculation was detected in the patient when a doctor or the like makes a definitive diagnosis, thereby assisting the doctor in making a definitive diagnosis.

[0046] 4 is a diagram showing a third example of generation of diagnostic support information by the learning model 62. In the third example illustrated in FIG. 4, the diagnostic support information is the number of times fasciculation is detected for each of multiple muscles, and the learning model 62 can output, for example, the number of times fasciculation is detected for each of multiple muscles over a predetermined time period (e.g., 10 minutes, 20 minutes, 30 minutes, 1 hour, etc.). Note that instead of the number of times fasciculation is detected, the frequency of fasciculation may also be used. When fasciculation is detected multiple times within a predetermined time period, the frequency can be calculated as {frequency = number of detections / predetermined time}.

[0047] The learning model 62 of the third example, i.e., machine learning, can be generated, for example, as follows. First, the control unit 51 collects first training data including ultrasound images of multiple patients who have been definitively diagnosed with ALS and the number of times fasciculation has been detected for each of multiple muscles in the ultrasound images (teacher data). Next, the control unit 51 collects second training data including ultrasound images of multiple patients who have been definitively diagnosed with non-ALS and the number of times fasciculation has been detected for each of multiple muscles in the ultrasound images (teacher data).

[0048] When an ultrasound image of a patient who has been definitively diagnosed with ALS is input to the learning model 62 based on the first training data, the control unit 51 can update the parameters of the learning model 62 so that the learning model 62 outputs the "number of times fasciculation was detected for each of multiple muscles" to generate the learning model 62. Furthermore, when an ultrasound image of a patient who has been definitively diagnosed with non-ALS is input to the learning model 62 based on the second training data, the control unit 51 can update the parameters of the learning model 62 so that the learning model 62 outputs the "number of times fasciculation was detected for each of multiple muscles" to generate the learning model 62.

[0049] As described above, when the control unit 51 acquires an ultrasound image from the ultrasound image generation unit 53, the control unit 51 inputs the ultrasound image into the learning model 62, which generates diagnostic support information including the number of times fasciculation is detected for each of the subject's multiple muscles, to generate diagnostic support information including the number of times fasciculation is detected for each of the subject's multiple muscles. Diagnosis of ALS is often symptom-dependent, making a definitive diagnosis difficult. However, the information processing device 50 can provide diagnostic support information including the number of times fasciculation is detected for each of the patient's multiple muscles when a doctor or other physician makes a definitive diagnosis, thereby assisting the doctor in making a definitive diagnosis. Furthermore, the ultrasound probe 10 of this embodiment can generate ultrasound images of multiple muscles by attaching one probe to the subject's body surface. Furthermore, ultrasound images of even more muscles can be generated by simultaneously attaching multiple ultrasound probes 10 to the subject's body surface.

[0050] Although not shown, the control unit 51 may include time-series data on the detection time points of fasciculation for each of multiple muscles in the diagnostic assistance information. For example, the control unit 51 may set the detection time points of fasciculation for muscle AAA as t11, t12, ..., t1n, the detection time points of fasciculation for muscle BBB as t21, t22, ..., t2m, and the detection time points of fasciculation for muscle CCC as t31, t32, ..., t3p. In this case, when the control unit 51 acquires ultrasound images from the ultrasound image generation unit 53, the control unit 51 inputs the ultrasound images to a learning model 62 that generates diagnostic assistance information including time-series data on the detection time points of fasciculation for each of multiple muscles, thereby generating diagnostic assistance information including time-series data on the detection time points of fasciculation for each of the subject's multiple muscles. The information processing device 50 can provide diagnostic assistance information indicating the timing at which fasciculation is detected in which muscles of the subject when a doctor or other physician makes a definitive diagnosis, thereby providing reference for a definitive diagnosis of ALS.

[0051] Fig. 5 is a diagram showing a fourth example of generation of diagnostic support information by the learning model 62. In the fourth example illustrated in Fig. 5, the diagnostic support information is the accuracy of ALS, and a numerical value within the range of 0 to 100%, for example, can be output. However, since a definitive diagnosis of ALS is extremely difficult, it is difficult for the learning model 62 to output an accuracy of 0% or 100%, and an output of, for example, an accuracy of about 10% to 90% is expected.

[0052] The generation of the learning model 62 of the fourth example, i.e., machine learning, can be performed, for example, as follows. First, the control unit 51 collects first training data including ultrasound images of multiple patients who have been definitively diagnosed with ALS and the diagnosis result "ALS." Next, the control unit 51 collects second training data including ultrasound images of multiple patients who have been definitively diagnosed with non-ALS and the diagnosis result "non-ALS." Based on the first training data, the control unit 51 can generate the learning model 62 by updating the parameters of the learning model 62 so that the learning model 62 outputs "ALS" when ultrasound images of patients who have been definitively diagnosed with ALS are input to the learning model 62. Furthermore, based on the second training data, the control unit 51 can generate the learning model 62 by updating the parameters of the learning model 62 so that the learning model 62 outputs "non-ALS" when ultrasound images of patients who have been definitively diagnosed with non-ALS are input to the learning model 62.

[0053] As described above, when the control unit 51 acquires an ultrasound image from the ultrasound image generation unit 53, the control unit 51 can input the ultrasound image to the learning model 62 that generates diagnostic support information including the probability of ALS, and generate diagnostic support information including the probability of ALS for the subject. This allows the information processing device 50 to provide diagnostic support information including the probability of ALS for the patient when a doctor or the like makes a definitive diagnosis, thereby supporting the doctor in making a definitive diagnosis.

[0054] 2 to 5, the learning model 62 is configured to individually output the presence or absence of ALS, the number of fasciculations detected, the number of fasciculations detected for each of the multiple muscles, and the accuracy of ALS, but is not limited to this. The learning model 62 may be configured to output the presence or absence of ALS, the number of fasciculations detected, the number of fasciculations detected for each of the multiple muscles, and the accuracy of ALS together, or may be configured to output two or more pieces of information from the presence or absence of ALS, the number of fasciculations detected, the number of fasciculations detected for each of the multiple muscles, and the accuracy of ALS.

[0055] FIG. 6 is a diagram showing an example of the display of diagnostic support information. The ALS diagnostic support information shown in FIG. 6 is displayed on the display unit 54 and can be visually confirmed by a doctor or the like. The ALS diagnostic support information includes a selection field for selecting a patient ID from a patient list, a field displaying the possibility of ALS, a field displaying the accuracy of ALS, and a field displaying the number of fasciculations detected. In the example of FIG. 6, the possibility of ALS is "yes," the accuracy of ALS is "60%, and the number of fasciculations detected is displayed for each muscle (muscle) "aaa," "bbb," and "ccc." Specific examples of muscles include, but are not limited to, the tongue muscle, sternocleidomastoid muscle, sclerospinous muscle, and trapezius muscle.

[0056] By referring to the ALS diagnostic support information as shown in FIG. 6, doctors can receive diagnostic information that can complement or replace electromyography.

[0057] ALS is said to require the use of a ventilator or lead to death within two to five years of onset, making early treatment essential. In Japan, various clinical trials have been conducted on drugs such as HGF (hepatocyte growth factor), perampanel, methylcobalamin, ropinirole, bosutinib, and tofersen. Drug screening using iPS cells has also increased the number of investigational drug candidates. However, the number of patients eligible for clinical trials is currently limited, and despite the large number of investigational drug candidates, few ALS patients are able to participate in clinical trials. Below, we explain how to select an investigational drug from among the candidate drugs.

[0058] As an example, a doctor uses the ultrasound probe 10 of this embodiment to identify the number or frequency of fasciculations detected by the information processing device 50 before and after administering an investigational drug candidate (candidate drug) to multiple ALS patients. The administration period of the investigational drug candidate may vary depending on the investigational drug candidate, but is typically about one to two weeks. If the number of fasciculations detected after administration is lower than the number before administration, or if the frequency of fasciculations after administration is lower than the frequency before administration, the doctor can decide to select the investigational drug candidate as the investigational drug.

[0059] As described above, the control unit 51 generates first diagnostic support information including the number of fasciculations detected in the subject within a predetermined time period before administration of the candidate drug, and generates second diagnostic support information including the number of fasciculations detected in the subject within a predetermined time period after administration of the candidate drug. The control unit 51 can determine whether or not to select the candidate drug as an investigational drug based on the generated first diagnostic support information and second diagnostic support information.

[0060] In clinical trials, the primary endpoint is the ALSFRS (ALS Functional Rating Scale) score, which is subjectively assigned by a physician. The ALSFRS is evaluated as the sum of scores for 12 items, including language, salivation, swallowing, writing, eating behavior, and daily activities. According to this embodiment, the ALSFRS can be used as a quantitative supplementary index for the ALSFRS.

[0061] The control unit 51 may also generate first diagnostic support information including the number of times fasciculation is detected in the subject within a predetermined time period before administration of each of the plurality of candidate drugs, and may generate second diagnostic support information including the number of times fasciculation is detected in the subject within a predetermined time period after administration of each of the plurality of candidate drugs. The control unit 51 may also determine the priority of investigational drugs from among the plurality of candidate drugs based on the generated first diagnostic support information and second diagnostic support information.

[0062] FIG. 7 shows an example of a clinical trial candidate drug evaluation. The clinical trial candidate drug evaluation shown in FIG. 7 is displayed on the display unit 54 and can be visually confirmed by a physician or other professional. The clinical trial candidate drug evaluation includes fields for displaying, for each clinical trial candidate drug, the number of fasciculations detected before and after administration and the clinical trial candidate drug's priority for clinical trials. In the example of FIG. 7 , the number of fasciculations detected before administration of the clinical trial candidate drug "xxx" is 40, the number of fasciculations detected after administration is 10, and the priority is 2. The number of fasciculations detected before administration of the clinical trial candidate drug "yyy" is 40, the number of fasciculations detected after administration is 5, and the priority is 1. The number of fasciculations detected before administration of the clinical trial candidate drug "zzz" is 40, the number of fasciculations detected after administration is 20, and the priority is 3. Note that the number of fasciculations detected is for convenience and is not limited to the example shown in FIG. 7 .

[0063] When a candidate drug for clinical trials is administered to multiple subjects, the number of times fasciculation is detected before and after administration may be a statistical value (e.g., mean value, median value, etc.) of the number of times fasciculation is detected in multiple subjects.

[0064] By referring to the evaluation of candidate investigational drugs shown in Figure 7, doctors can prioritize the limited number of investigational drug candidates that can be included in clinical trials from among the many mixed investigational drug candidates.

[0065] 8 is a diagram showing an example of a processing procedure by the information processing device 50. The ultrasound image generating unit 53 acquires reflected waves of ultrasound irradiated onto the body surface of the subject via the sheet-like ultrasound probe 10 (S11), and generates an ultrasound image based on the acquired reflected waves (S12).

[0066] The control unit 51 inputs the generated ultrasound image to the learning model 62 (S13), and acquires diagnostic assistance information generated and output by the learning model 62 (S14). The control unit 51 outputs the acquired diagnostic assistance information (S15), and ends the process.

[0067] As described above, this embodiment can be applied as a diagnosis support device that supports early diagnosis of ALS and as a drug efficacy evaluation support device that supports early treatment. The information processing device 50 may also be used as a screening device that determines whether a non-ALS patient has non-ALS. Furthermore, the information processing device 50 does not need to perform all of its functions within a single device, and may be configured to acquire ultrasound images and analyze the images at different locations. For example, the information processing device 50 may include multiple devices installed at multiple locations, with a first device acquiring ultrasound images at home or a medical institution and a second device analyzing the ultrasound images at an imaging center.

[0068] [First Modification] In the above description, the control unit 51 determines whether or not a neuromuscular disease is present based on an ultrasound image. However, even if fasciculation is detected in an ultrasound image, there may be cases in which a neuromuscular disease is not present. If diagnostic assistance information indicating the possibility of a neuromuscular disease is output when a neuromuscular disease is not present, the patient may experience unnecessary anxiety. Therefore, the control unit 51 may acquire biometric data, other than reflected waves, that indicates the subject's biological condition, and generate diagnostic assistance information regarding the subject's neuromuscular disease based on a combination of the biometric data and the ultrasound image.

[0069] As an example, the biological data is data indicating potentials obtained by needle electromyography. When an ultrasound image shows fasciculation, the control unit 51 may generate diagnostic support information indicating a possible neuromuscular disease when either a PSW or Fib potential appears, and may not generate diagnostic support information indicating a possible neuromuscular disease when the potential does not appear. By operating in this manner, the control unit 51 can prevent erroneous diagnostic support information from being output when an ultrasound image shows fasciculation but does not indicate a neuromuscular disease.

[0070] The control unit 51 may calculate the probability of a neuromuscular disease based on a combination of the likelihood that muscle movement detected in the ultrasound image is a fasciculation and the range to which the potential obtained by needle electromyography or high-density electromyography belongs, and output diagnostic assistance information including information corresponding to the calculated probability. By outputting such diagnostic assistance information, the control unit 51 allows the doctor or patient to understand the probability of a neuromuscular disease, making it easier for the doctor to provide appropriate treatment.

[0071] [Second Modification] In the above description, the ultrasound probe 10 generates an ultrasound image of one area, but the ultrasound probe 10 may have multiple ultrasound transducers and be configured to simultaneously generate ultrasound images of multiple different areas. In this case, the control unit 51 determines whether each of the multiple ultrasound images shows fasciculation. The control unit 51 may output diagnostic assistance information indicating whether or not a neuromuscular disease is likely to occur in each of the multiple areas, in association with information for identifying the multiple areas.

[0072] (Supplementary Note 1) The computer program causes a computer to execute the steps of acquiring reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto a subject's body surface via an ultrasound probe detachable from the subject's body surface and attached to the body surface of the subject; generating an ultrasound image based on the acquired reflected wave data; and generating diagnostic support information regarding the subject's neuromuscular disease based on the generated ultrasound image.

[0073] (Appendix 2) In the step of generating the diagnostic assistance information in Appendix 1, the computer inputs the generated ultrasound image into a learning model that generates information to assist in the diagnosis of neuromuscular diseases when an ultrasound image is input, and generates the diagnostic assistance information regarding the neuromuscular disease of the subject.

[0074] (Supplementary Note 3) In Supplementary Note 1 or Supplementary Note 2, the computer is further caused to execute a step of comparing movement pattern data indicating a muscle movement pattern shown in the ultrasound image with one or more reference pattern data indicating muscle fasciculations, and in the step of generating diagnostic support information, if the computer detects any of the reference pattern data whose similarity to the movement pattern data is equal to or greater than a threshold, the computer generates the diagnostic support information indicating a possibility of a neuromuscular disease.

[0075] (Supplementary Note 4) In any one of Supplementary Note 1 to Supplementary Note 3, the diagnostic assistance information includes information regarding the possibility of amyotrophic lateral sclerosis.

[0076] (Supplementary Note 5) In any one of Supplementary Notes 1 to 4, the diagnostic assistance information includes the number of times fasciculations are detected within a predetermined time period.

[0077] (Supplementary Note 6) In any one of Supplementary Notes 1 to 5, the diagnostic assistance information includes the number of times fasciculations are detected for each of a plurality of muscles.

[0078] (Supplementary Note 7) In any one of Supplementary Notes 1 to 6, the diagnostic assistance information includes a probability that the subject's neuromuscular disease is amyotrophic lateral sclerosis.

[0079] (Appendix 8) In any one of Appendices 1 to 7, the computer program causes the computer to execute the steps of generating first diagnostic support information including the number of fasciculations detected in a subject within a predetermined time period before administration of a candidate drug, generating second diagnostic support information including the number of fasciculations detected in the subject within a predetermined time period after administration of the candidate drug, and determining whether or not to select the candidate drug as an investigational drug based on the first diagnostic support information and the second diagnostic support information.

[0080] (Supplementary Note 9) In any one of Supplements 1 to 8, the computer program causes the computer to execute the steps of generating first diagnostic support information including the number of times fasciculations are detected in a subject within a predetermined time period before administration of each of a plurality of candidate drugs; generating second diagnostic support information including the number of times fasciculations are detected in the subject within a predetermined time period after administration of each of the plurality of candidate drugs; and determining the priority of investigational drugs from among the plurality of candidate drugs based on the first diagnostic support information and the second diagnostic support information.

[0081] (Appendix 10) In any one of Appendices 1 to 9, the computer program further causes the computer to execute a step of acquiring biological data indicating the biological condition of the subject, different from the reflected wave data, and in the step of generating diagnostic support information, the computer generates diagnostic support information regarding the neuromuscular disease of the subject based on a combination of the biological data and the ultrasound image.

[0082] (Supplementary Note 11) In any one of Supplementary Note 1 to Supplementary Note 10, the biological data is data indicating an electric potential obtained by needle electromyography, and in the step of generating the diagnostic support information, when the ultrasound image shows muscle fasciculations, if either a positive sharp wave or a spontaneous fiber potential appears, the computer generates the diagnostic support information indicating a possibility of ALS, and when the electric potential does not appear, the computer does not generate the diagnostic support information indicating a possibility of ALS.

[0083] (Appendix 12) The information processing device includes a control unit, which acquires reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto the body surface of the subject via an ultrasound probe detachable from the body surface of the subject and attached to the body surface of the subject, generates an ultrasound image based on the acquired reflected wave data, and generates diagnostic support information regarding the neuromuscular disease of the subject based on the generated ultrasound image.

[0084] (Appendix 13) The information processing method is a method executed by a computer, which acquires reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto a body surface of a subject via an ultrasound probe detachable from the body surface of the subject and attached to the body surface of the subject, generates an ultrasound image based on the acquired reflected wave data, and generates diagnostic support information regarding a neuromuscular disease of the subject based on the generated ultrasound image.

[0085] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. A multiple claim (multi-multi claim) that references at least one other multiple claim may also be used.

[0086] REFERENCE SIGNS LIST 10 Ultrasonic probe 11 Ultrasonic array sensor 50 Information processing device 51 Control unit 52 Interface unit 53 Ultrasonic image generation unit 54 Display unit 55 Operation unit 56 Memory 57 Storage unit 58 Communication unit 61 Computer program 62 Learning model

Claims

1. A computer program that causes a computer to execute the steps of: acquiring reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto a subject's body surface via an ultrasound probe that is detachable from the subject's body surface and attached to the subject's body surface; generating an ultrasound image based on the acquired reflected wave data; and generating diagnostic support information regarding the subject's neuromuscular disease based on the generated ultrasound image.

2. The computer program of claim 1, wherein in the step of generating diagnostic assistance information, the computer inputs the generated ultrasound image into a learning model that generates information to assist in the diagnosis of neuromuscular diseases when an ultrasound image is input, and generates the diagnostic assistance information regarding the neuromuscular diseases of the subject.

3. The computer program of claim 1, further causing the computer to execute a step of comparing movement pattern data indicating the muscle movement pattern shown in the ultrasound image with one or more reference pattern data indicating muscle fasciculations, and in the step of generating diagnostic support information, the computer generates the diagnostic support information indicating a possibility of a neuromuscular disease if it detects reference pattern data whose similarity with the movement pattern data is equal to or greater than a threshold.

4. A computer program according to any one of claims 1 to 3, wherein the diagnostic assistance information includes information relating to the possibility of amyotrophic lateral sclerosis.

5. A computer program according to any one of claims 1 to 3, wherein the diagnostic assistance information includes the number of times fasciculations are detected within a predetermined period of time.

6. A computer program according to any one of claims 1 to 3, wherein the diagnostic assistance information includes the number of times fasciculations are detected for each of a plurality of muscles.

7. A computer program according to any one of claims 1 to 3, wherein the diagnostic assistance information includes a probability that the subject's neuromuscular disease is amyotrophic lateral sclerosis.

8. A computer program as described in any one of claims 1 to 3, causing the computer to execute the following steps: generating first diagnostic support information including the number of fasciculations detected in a subject within a predetermined time period before administration of a candidate drug; generating second diagnostic support information including the number of fasciculations detected in the subject within a predetermined time period after administration of the candidate drug; and determining whether or not to select the candidate drug as an investigational drug based on the first diagnostic support information and the second diagnostic support information.

9. A computer program as described in any one of claims 1 to 3, which causes the computer to execute the following steps: generating first diagnostic support information including the number of muscle fasciculations detected in the subject over a predetermined time period before administration of each of a plurality of candidate drugs; generating second diagnostic support information including the number of muscle fasciculations detected in the subject over a predetermined time period after administration of each of the plurality of candidate drugs; and determining the priority of investigational drugs from among the plurality of candidate drugs based on the first diagnostic support information and the second diagnostic support information.

10. A computer program as described in any one of claims 1 to 3, further causing the computer to execute a step of acquiring biological data indicating the biological condition of the subject, different from the reflected wave data, and in the step of generating diagnostic support information, the computer generates diagnostic support information regarding the neuromuscular disease of the subject based on a combination of the biological data and the ultrasound image.

11. The computer program of claim 10, wherein the biological data is data indicating potentials obtained by needle electromyography, and in the step of generating the diagnostic support information, the computer generates the diagnostic support information indicating the possibility of ALS when the ultrasound image shows muscle fasciculations and either a positive sharp wave or a spontaneous fiber potential appears, and does not generate the diagnostic support information indicating the possibility of ALS when such potentials do not appear.

12. An information processing device comprising a control unit, which acquires reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto a subject's body surface via an ultrasound probe detachable from the subject's body surface and attached to the body surface of the subject, generates an ultrasound image based on the acquired reflected wave data, and generates diagnostic support information regarding neuromuscular disease of the subject based on the generated ultrasound image.

13. An information processing method executed by a computer, comprising: acquiring reflected wave data indicating the intensity of reflected waves of ultrasound irradiated onto a subject's body surface via an ultrasound probe detachable from the subject's body surface and attached to the subject's body surface; generating an ultrasound image based on the acquired reflected wave data; and generating diagnostic support information regarding neuromuscular disease of the subject based on the generated ultrasound image.

Images