A novel diagnostic system for Doppler ultrasound of the peripheral arteries to detect the presence of thoracic aortic aneurysm

Abstract

OBJECTIVE:

We aimed to develop a novel ultrasound system and examine its feasibility for noninvasively detecting thoracic aortic aneurysm (TAA) in clinical settings.

METHODS:

We developed a novel ultrasound system consisting of a modified console and data analysis algorithm. The exploratory study included 100 patients hospitalized for elective cardiovascular surgery. After admission, the arterial pulse waveform at the left carotid artery was acquired using the novel system. Based on these data, we inferred the presence of TAA based on arterial viscoelasticity and instability, which are reflected into the time-averaged trajectory of deformation of the blood vessel wall caused by disturbance of blood flow. Meanwhile, all patients underwent computed tomography as preoperative screening to confirm the presence of TAA. The sensitivity and specificity of TAA detection using the novel ultrasound system were calculated.

RESULTS:

The datasets from 37 patients were not suitable for analysis and were thus discarded. Based on computed tomography findings, 40 patients were categorized into the aneurysm group while 23 were judged not to have and aortic aneurysm. On the other hand, 44 patients were diagnosed as having TAA based on ultrasound findings obtained using the novel system. The overall sensitivity and specificity of the ultrasound system were 0.83 and 0.52, respectively.

CONCLUSION:

We successfully developed a novel system for noninvasive, ultrasound-based evaluation of the left carotid artery to detect TAA. Although improvements to the probe and diagnostic algorithm are warranted, this device has potential utility for mass screening to detect asymptomatic TAA as part of community-level healthcare programs.

Keywords

Thoracic aortic aneurysm ultrasound non-invasive detection peripheral artery

1. Background

The number of surgeries for thoracic aortic aneurysm (TAA) is steadily increasing [1], possibly in relationship with increased prevalence of atherosclerotic disease reflecting an aging population and improved diagnostic capabilities, especially following the development of novel diagnostic modalities. The availability of less invasive surgical procedures such as thoracic endovascular aneurysm repair (TEVAR) may also contribute to the increased number of surgical operations [1–3]. However, asymptomatic patients with TAA are rarely diagnosed unless they undergo comprehensive examinations such as computed tomography (CT), magnetic resonance imaging, or ultrasound echography. Despite the increase in diagnostic capability, many patients are still transferred for specialty care with undiagnosed ruptured aortic aneurysm [1–3]. According to the annual statistics of the Japanese Association for Thoracic Surgery, 8.0% of all patients referred for TAA surgery have ruptured TAA [1]. The in-hospital mortality rate associated with ruptured TAA was found to be up to 21.2% greater or 6.4-fold higher than that for unruptured TAA [1]. Therefore, we sought to develop a methodology for noninvasive diagnosis of TAA that can be implemented as part of routine health checkups or visits to the outpatient clinic.

Since the late 1990s, we have been working on the development of a device for evaluating the degree of atherosclerosis. In 2004, regulatory approval was obtained for the manufacturing of the TRY1 device, which could be used to estimate the severity of atherosclerosis based on the velocity of the blood flow (approval for the manufacturing of medical devices: 21600BZx00440000, to Taiyo Denshi Co. Ltd., Miyagi, Japan) (Supplementary File S1). Upon performing detailed analysis of data obtained using the TRY1 device, we previously found that the peripheral arterial wall exhibits specific motions in patients with abdominal aortic aneurysm, and that such motions are reflected in the degree of irregular movement of the blood vessel wall [4,5]. To facilitate TAA detection, we developed an ultrasound system consisting of a modified TRY1 console and a newly developed algorithm for data analysis. The present study was conducted to explore the potential diagnostic capability of this novel ultrasound system for TAA detection in the clinical setting.

2. Patients and methods

2.1. Theoretical background of the newly developed ultrasound system

The progression of atherosclerosis is often accompanied by structural changes in the vessel wall, with associated changes in the mechanical properties of the arterial wall. We previously proposed the I^∗-S method to estimate the degree of viscoelasticity and irregular movement of the blood vessel wall [4,6], and included entropy, S, into the Dc parameter to characterize the irregularity of the time-averaged waveform of the pulsatile blood vessel wall [5]. The principles behind these methodologies are described in detail elsewhere [4,5]. Briefly, the ultrasonic Doppler effect was used as an acceleration sensor to measure the strain rate and the elastic degradation of the vessel wall. The parameter of interest, referred to as the viscoelasticity index (I^∗), is calculated on the basis of a previously reported formula that estimates the degree of viscoelasticity in the arterial wall, which is related to the severity of atherosclerosis [4]. A negative I^∗ indicates the normal state of the vessel wall, while a positive I^∗ indicates atherosclerosis, with I^∗ > 1.0 indicating severe atherosclerosis.

On the other hand, we incidentally found that patients with aortic aneurysms exhibit a complicated two-phase peak in the output waveform only after the process of a computer analysis of data describing the motion of the blood vessel wall (Fig. 1) [4]. This waveform was not observed in patients without aortic aneurysm. However, it is difficult to detect the presence of a two-phase peak solely on the basis of I^∗, because I^∗ does not include a suitable function for determining the specific output waveform of patients with aneurysm. Therefore, using the chaos theory, we analyzed the vessel wall deformation trajectory of the attractor and calculated the entropy of such waveforms [4]. These analyses were performed according to complexity science theory and took into account the unstable behavior of the blood vessel wall to discriminate and diagnose the two-phase peak waveform indicative of aneurysm [4]. In this analysis, the trajectory of a healthy vessel wall has regular periodic elliptical cycles. However, in patients with aneurysms, characteristic circuit loops were noted in the trajectory in addition to the scattering of the trajectory, which demonstrated the irregular characteristics of the blood vessel wall (Fig. 2) [4]. This behavior occurred by wave propagation through the aortic aneurysm, generating instability in the blood flow and vessel wall motion. We also proposed that this new method coupled with a fractal concept could be used for quantitative evaluation of unstable blood vessel behavior [5]. To better understand this concept, we further proposed that the Dc parameter, defined as the increasing rate of fractal dimension based on the proposed multi fractal analysis [7], reflects the degree of irregularity of the time-averaged waveform of the pulsatile blood vessel wall [5]. Based on our previous studies, a Dc value above 0.3 is thought to reflect unstable vessel wall behavior.

Fig. 1.

Output and synthesized waveform. Output (A) and synthesized (B) waveform obtained in a subject without aortic aneurysm. The white arrow indicates a one-phase peak in the output waveform, whereas the circle indicates the corresponding one-phase peak in the synthesized waveform. Output (C) and synthesized (D) waveform in a patient with aortic aneurysm. The black arrow indicates a two-phase peak in the output waveform, while the circle indicates the corresponding two-phase peak in the synthesized waveform.

Fig. 2.

Example of a vessel wall deformation trajectory in the attractor analysis. (A) Trajectory obtained in a subject without aortic aneurysm. (B) Trajectory obtained in a patient with aortic aneurysm. The white arrow indicates characteristic circuit loops. (C) Synthesized waveform in healthy patient showed only one-phase peak (circle). (D) Synthesized waveform with aortic aneurysm indicated two-phase peak (circle).

We implemented the above-described algorithms for analysis of data collected using a modified TRY1 console, which could permit accurate data acquisition by exhibiting the real-time waveform output and precise diagnosis by discriminating the two-phase peak waveform indicative of a presence of thoracic aortic aneurysm through an application of a chaos theory generally utilized in complexity science. This newly developed ultrasound system is meant for use in the clinical setting.

2.2. Preliminary testing prior to the clinical feasibility study

Before initiating the present feasibility study, we conducted preliminary testing in a small cohort of six consecutive patients with TAA to explore the potential for detecting TAA using the novel ultrasound system. During preliminary testing, TAA detection tended to have more sensitivity for measurements performed at the left carotid artery than for those performed at other measurement points such as the right carotid or bilateral femoral arteries. Furthermore, computational fluid dynamics analysis of experimental measurements suggested that TAA diagnosis would be more sensitive for measurements performed at more upstream sites than for those performed at more downstream sites. Therefore, in the clinical feasibility study, we performed all measurements at the left carotid artery.

2.3. Patients and study design

Patients who were admitted to our department for elective cardiovascular surgery were recruited for this study. Patients admitted to the emergency department, those aged 18 years or younger, and those with previous history of carotid artery surgery were excluded. For this initial exploratory study, we sought to obtain data pertaining to 100 consecutive patients. Based on these data, we analyzed the sensitivity and specificity for TAA detection, aiming to determine how the newly developed ultrasound system can be adjusted to improve TAA detection accuracy.

The study protocol was approved by the ethical committee of Tohoku University Hospital. Informed consent was obtained from all participants prior to performing any measurements. Using the novel ultrasound system, the pulse waves were recorded at the left carotid artery for 15 seconds. The data were acquired according to the same protocol, by the same technician, who was blinded to patient data. Afterward, the de-identified data were sent to the engineering laboratory for analysis to detect TAA. These ultrasound data underwent comprehensive analysis to determine whether the predicted TAA patterns are present, and a decision was issued by the laboratory staff.

Meanwhile, all patients underwent whole-body CT as preoperative screening. The de-identified CT data were sent to the Division of Cardiovascular Surgery, where a staff member who was also blinded to patient data independently evaluated the CT scans and, in concordance with reports from radiologists, diagnosed the presence of aortic aneurysms. Thereafter, participants were divided into two groups according to the presence of TAA diagnosed based on preoperative CT findings. Finally, the sensitivity and specificity of the newly developed ultrasound system were calculated.

3. Results

A total of 100 patients were evaluated in this feasibility study. All participants underwent enhanced CT preoperatively. Based on preoperative enhanced CT findings, 64 patients were diagnosed as having aortic aneurysm (AN group), while 36 patients were not (non-AN group). However, upon evaluation by the engineering laboratory, 37 patients were excluded from the study because the ultrasound data were not suitable for analysis. Therefore, 40 patients in the AN group and 23 patients in the non-AN group were included in the final analysis of this study. A summary of preoperative patient characteristics is shown in Table 1. Age was significantly higher in the AN group than in the non-AN group (median [quartile], years: 72 [66, 80] vs 61 [48, 67] years; P < 0.001), as were the frequency of previous cardiovascular surgery (20/40, 50.0% vs 3/23, 13.0%; P = 0.01) and the frequency of use of thoracic vascular grafts (13/40, 32.5% vs 1/23, 4.6%; P = 0.01). Regarding data collected using the ultrasound device, the Dc value was significantly higher in the AN group than in the non-AN group (mean ± standard deviation: 0.37 ± 0.06 vs 0.34 ± 0.06; P < 0.05), whereas the I^∗ value did not differ significantly between the two groups. Regarding TAA, 60.0% of AN patients had fusiform type, while 40.0% had saccular type; by etiology, true aneurysm accounted for 60.0% of cases, while dissecting aneurysm and pseudoaneurysm each accounted for 20.0% of cases.

Table 1
Patient characteristics. Age is shown as median [interquartile range]. Height, weight, body mass index (BMI), ejection fraction (EF), and device data are shown as average ± two standard deviations. The other variables are given as % (N). Categorical data were compared using the chi-square test. Age was compared using the Wilcoxon test, while the other continuous variables were compared using Students t-test

Non-AN group AN group p value

(N = 23) (N = 40)

Age 61[48,67] 72[66,80] <0.001

Height (cm) 163.3 ± 10.6 162.5 ± 7.7 0.72

Weight (kg) 61.5 ± 12.4 60.2 ± 10.6 0.66

Body Mass Index 23.0 ± 0.7 22.7 ± 0.5 0.74

Diabetes Mellitus 34.7% (8) 20.0% (8) 0.23

Hyperlipidemia 56.5% (13) 67.5% (27) 0.42

CRF 21.7% (5) 45.0% (18) 0.10

Hemodialysis 4.4% (1) 7.5% (3) 1.00

Hypertension 69.6% (16) 87.5% (35) 0.10

COPD (>Mild) 30.4% (7) 40.0% (16) 0.59

CVD 8.7% (2) 15.0% (6) 0.70

Previous Operation 13.0% (3) 50.0% (20) 0.01

Thracic Vascular Graft 4.6% (1) 32.5% (13) 0.01

EF (%) 61.0 ± 17.8 65.8 ± 9.4 0.83

Atrial fibrillation 8.7% (2) 15.0% (6) 0.70

Device data

Dc value 0.34 ± 0.06 0.37 ± 0.06 0.05

I^∗ 0.14 ± 0.28 0.22 ± 0.30 0.82

Morphology

Fusiform 60.0% (24)

Saccular 40.0% (16)

Type of Aneurysm

True 60.0% (24)

Dissection 20.0% (8)

Pseudoaneurysm 20.0% (8)

	Non-AN group	AN group	p value
Age	61[48,67]	72[66,80]	<0.001
Height (cm)	163.3 ± 10.6	162.5 ± 7.7	0.72
Weight (kg)	61.5 ± 12.4	60.2 ± 10.6	0.66
Body Mass Index	23.0 ± 0.7	22.7 ± 0.5	0.74
Diabetes Mellitus	34.7% (8)	20.0% (8)	0.23
Hyperlipidemia	56.5% (13)	67.5% (27)	0.42
CRF	21.7% (5)	45.0% (18)	0.10
Hemodialysis	4.4% (1)	7.5% (3)	1.00
Hypertension	69.6% (16)	87.5% (35)	0.10
COPD (>Mild)	30.4% (7)	40.0% (16)	0.59
CVD	8.7% (2)	15.0% (6)	0.70
Previous Operation	13.0% (3)	50.0% (20)	0.01
Thracic Vascular Graft	4.6% (1)	32.5% (13)	0.01
EF (%)	61.0 ± 17.8	65.8 ± 9.4	0.83
Atrial fibrillation	8.7% (2)	15.0% (6)	0.70
Device data
Dc value	0.34 ± 0.06	0.37 ± 0.06	0.05
I^∗	0.14 ± 0.28	0.22 ± 0.30	0.82
Morphology
Fusiform		60.0% (24)
Saccular		40.0% (16)
Type of Aneurysm
True		60.0% (24)
Dissection		20.0% (8)
Pseudoaneurysm		20.0% (8)

After completing all data acquisition, 33 of 40 patients in the AN group and 11 of 23 patients in the non-AN group were judged as positive for TAA using the newly developed ultrasound system, while the other patients were deemed as negative for TAA. Based on these results, the sensitivity and specificity of TAA detection using this ultrasound system were calculated to be 0.83 and 0.52, respectively, while the positive predictive value and negative predictive value were calculated to be 0.75 and 0.63, respectively.

We performed a detailed analysis of patients misdiagnosed using the ultrasound system (11 patients in the non-AN group and 7 patients in the AN group; Table 2). Among the 11 patients with false-positive results in the non-AN group, 4 patients (36.3%) had aortic valve stenosis and 3 patients (24.2%) had reduced cardiac ejection fraction (below 50%). Of the seven patients with false-negative results in the AN-group, 1 patient had a history of surgery for acute aortic dissection and had received prosthetic vascular grafts to replace the supra-aortic neck vessels; 1 patient had previously undergone endovascular intervention for TAA, while the operative indication at the time of the study was minor type-1a endoleak; another 2 patients had root or ascending aneurysm with relatively straight rather than fusiform pattern.

Table 2

Patient characteristics in the group of false positives and false negatives. A total of 11 patients were incorrectly diagnosed as having aneurysm, while seven patients with aneurysm could not be detected using the developed device in its current design. Variables potentially associated with misjudgment of the data collected using the device are also listed

Case no	Age	Sex	Height (cm)	Weight (kg)	Operative indication	AV peak velocity (m/sec)	EF(%)
Psedo-Positive cases (N = 11)
1	72	F	151.4	43.1	Aortic valve stenosis	5.4	73
2	64	F	158.7	55.2	Aortic valve stenosis	4.5	64
3	61	F	147.5	65.9	Aortic valve stenosis	5.6	70
4	78	F	153.5	52.9	Aortic valve stenosis	3.7	82
5	42	M	175.0	63.4	Dilated Cardiomyopathy	0.7	20
6	49	M	172.0	73.0	Dilated Cardiomyopathy	1.7	22
7	72	M	166.2	68.5	Angina Pectris	1.3	46
8	66	F	147.2	38.4	Perivalvular Leakege (Mitral)	1.2	74
9	53	M	175.0	63.1	Mitral Regurgitation	1.1	71
10	59	M	161.7	53.9	Mitral Regurgitation	N/A	56
11	48	F	158.5	52.5	Right Ventricular Tumor	N/A	77
Case no	Age	Sex	Height (cm)	Weight (kg)	Location of AN	Operative indication	Type	Reoperation	Thrombosis
Pseudo-Negative cases (N = 7)
1	68	M	166.5	66.0	Root - Ascending Aorta	Aortoanular Ectasia	fusiform	no	no
2	64	F	152.0	64.2	Ascending Aorta	Aoric valve stenosis	fusiform	no	no
3	55	M	168.7	62.2	Distal Arch	Dissecting aortic aneurysm	fusiform	yes	no
4	58	M	155.5	57.1	Distal Arch	True aortic aneurysm	saccular	no	partial
5	66	M	173.2	82.2	Distal Arch	True aortic aneurysm	saccular	no	partial
6	84	M	158.8	51.4	Distal Arch	True aortic aneurysm	saccular	no	partial
7	80	M	167.2	59.9	Descending Aorta	Endoleak (type 1a)	saccular	yes	partial

4. Discussion

With respect to the detection of abdominal aortic aneurysm, ultrasound echography is a noninvasive assessment tool with a diagnostic accuracy similar to those of other modalities such as magnetic resonance imaging and CT. However, ultrasonography is generally not considered suitable for detecting TAA because of the anatomical obstacles surrounding the thoracic aorta. Nonetheless, we have developed a novel ultrasound system for noninvasive detection of TAA with high diagnostic accuracy. In the newly developed ultrasound system, the algorithm for TAA diagnosis is based on the Doppler ultrasound analysis of the arterial wall motion. This study was the first to obtain such data in patients with TAA. Data from measurements at the left carotid artery enabled us to detect a characteristic circuit loop, which, based on attractor analysis, was considered to be unique to patients with aortic aneurysm. These results suggested the possibility that such a device could be used for the detection of TAAs. In addition, we evaluated the diagnostic utility of the distribution of Dc and I^∗ values for diagnosing TAA. In previous studies, we had shown there is a correlation between the presence of abdominal aortic aneurysm and Dc ≥ 0.3 or I^∗ ≥ 0, whereas the majority of the healthy individuals examined had Dc < 0.3 and I^∗ < 0 [5]. The patients evaluated in the present study had mean Dc and I^∗ values of 0.36 ± 0.06 and 0.19 ± 0.29, respectively, which met the diagnostic criteria mentioned above. However, some patients had Dc or I^∗ values that did not fulfill the diagnostic criteria for aortic aneurism, which might reflect a heterogeneity among patients regarding the degree of atherosclerosis in the entire body, and especially at the measurement site. Furthermore, in most patients, the Dc value did not improve after the aortic operation. Therefore, we concluded that not all patients with TAA can be diagnosed solely on the basis of Dc and I^∗ values, and that the characteristic circuit loop in the attractor analysis is a key factor for this diagnostic algorithm. Although we could not obtain postoperative data in all cases, we noticed that the characteristic circuit loop in the attractor analysis was no longer present after the aortic operation (Fig. 3).

Fig. 3.

Distinctive example of the vessel wall deformation trajectory pre- and postoperatively. (A) Preoperative computed tomography scan revealed aortic arch aneurysm with 67mm in maximum diameter (black arrow). (B) Preoperative vessel wall deformation trajectory obtained using the ultrasound device. The characteristic circuit loop was depicted in the upper right of the trajectory (white arrow). (C) Postoperative computed tomography scan. The ascending aorta and aortic arch including supra-aortic neck vessels were replaced with branched prosthetic graft (dotted black arrow). (D) Postoperative vessel wall deformation trajectory obtained using the ultrasound device. The characteristic circuit loop was no longer present (dotted white arrow).

Aortic valve stenosis was often noted in patients with false-positive results (36.4%), but the groups did not differ significantly in this respect, likely because of the small sample size. Aortic valve stenosis also changes the normal behavior of the aortic wall by generating turbulence or influencing the momentum transferred to the vessel wall. On the contrary, low ejection fraction could be considered a confounding factor in diagnosing the presence of TAA using this ultrasound system because reduced cardiac function may result in inability to produce a normal distinct peak wave in the output waveform. Therefore, to improve the diagnostic accuracy of this ultrasound system, the sample size should be extended to include a sufficient number of patients with aortic valve stenosis and low ejection fraction. If we consider such conditions as positive test results reflecting aortic wall instability, the specificity of the ultrasound system improves from 0.52 to 0.73 while maintaining similar sensitivity. It can be considered that the unstable behavior of wall motion could be generated not only by an aortic aneurysm but also by other factors such as aortic stenosis or reduced cardiac function, and such instability can be detected using this ultrasound system. Therefore, the potential utility of the newly developed ultrasound system as a screening tool in outpatient clinics can be extended from TAA detection to detection of aortic valve stenosis and low ejection fraction, facilitating early treatment when necessary.

While we could not determine the risk factors for false negative results, we did notice that the distance from the left carotid artery to the beginning of the TAA was larger in patients with false-negative results than for those with positive results (mean ± two standard errors: 58.8 ± 22.5 vs 50.0 ± 8.6 mm). This parameter may be affected by the capability of the ultrasound probe to detect TAA-induced blood flow disturbances based on measurements at the left carotid artery, even though this seems to be the most sensitive site. To test this hypothesis, data from patients with abdominal aortic aneurysm should be measured at the femoral arteries (which would be closer to the abdominal aortic aneurysm) and evaluated to assess the sensitivity and specificity of the device and algorithm for the detection of abdominal aortic aneurysm.

This study has several limitations. First, the raw data obtained from a total of 37 participants were excluded because such data were not suitable for analysis, likely because of suboptimal or inconsistent positioning of the ultrasound probe during the measurement. In practice, this problem can be minimized by implementing visualization of real-time waveform data on the console display during data acquisition, which would facilitate repositioning of the ultrasound probe or adjusting the angle in order to obtain adequate data for analysis. Second, the performance of the ultrasound probe and the distance between the contact surface and the vessel wall can significantly influence data accuracy. Therefore, we should develop a new ultrasound probe with improved accuracy, which would facilitate obtaining precise data even in patients with thick adipose tissue between the contact surface and the vessel wall. Third, although we could confirm the existence of certain patterns of a characteristic circuit loop (typically manifesting at the upper right and lower left corners in the scattering elliptical cycles in the attractor analysis), the difference between these two positions in the scattering elliptical cycle is unclear. Elucidation of this phenomenon will help increase the diagnostic accuracy of the ultrasound system for detecting TAA. Finally, the cohort included in the present study had a higher prevalence of aortic aneurysm than the prevalence reported in the general population. Therefore, to obtain real data regarding the sensitivity and specificity of this ultrasound system in the clinical setting, we should perform a study in outpatient clinics or as part of health checkup surveys.

5. Conclusions

We developed a novel ultrasound system to detect TAA based only on measurements collected using an ultrasound probe applied to the left carotid artery. We conducted a preliminary feasibility study and demonstrated the utility and diagnostic accuracy of this ultrasound system. We expect that, after further improvement of the ultrasound probe and diagnostic algorithm, this novel system could serve as a valuable tool for detecting TAA in asymptomatic patients managed at the outpatient clinic or screened during routine health exams.

Footnotes

Conflict of interest

The authors have nothing to disclose in relationship with the present study.

Funding

There are no sources of funding to disclose.

Nomenclature

References

Committee for Scientific Affairs, The Japanese Association for Thoracic Surgery, M. Masuda, M. Okumura, Y. Doki, S. Endo, Y. Hirata, et al., Thoracic and cardiovascular surgery in Japan during 2014: Annual report by the Japanese Association for Thoracic Surgery, Gen Thorac Cardiovasc Surg 64(2016), 665–697.

Amano

Kuwano

and Yokomise

, Thoracic and cardiovascular surgery in Japan during 2011: Annual report by The Japanese Association for Thoracic Surgery, Gen Thorac Cardiovasc Surg 61 (2013), 578–607.

Sakata

Fujii

and Kuwano

, Thoracic and cardiovascular surgery in Japan during 2009: Annual report by the Japanese Association for Thoracic Surgery, Gen Thorac Cardiovasc Surg 59 (2011), 636–667.

Yokobori Jr

A.T.

Owa

Ichiki

Satoh

Ohtomo

Satoh

, The analysis and diagnosis of unstable behavior of the blood vessel wall with an aneurysm based on noise science, J Atheroscler Thromb 13 (2006), 163–174.

Honda

Yokobori Jr

A.T.

Ichiki

Karino

Ohtomo

and Sato

, Noninvasive estimation of blood vessel wall by ultrasonic Doppler effect based on the concept of multi-fractal, Japan Society for Mechanical Engineers 39 (2004), 178–179, (Article in Japanese).

Yokobori Jr

A.T.

Ohmi

Monma

Tomono

Inoue

Owa

, Correlation between the characteristics of acceleration and visco elasticity of artery wall under pulsatile flow conditions (physical meaning of I^∗ as a parameter of progressive behaviors of atherosclerosis and arteriosclerosis), Biomed Mater Eng 23 (2013), 75–91.

Yokobori Jr

A.T.

Sakai

Yamagiwa

and Yoshida

, Quantitative characterization of fracture surface of full lamellar TiAl under high temperature creep and fatigue conditions on the basis of fractal concept, Strength, Fracture and Complexity 1 (2003), 19–29.