The Automated system
for Ultrasound Liver

2.Liver ultrasound diagnosis
3.System configuration
4.Evaluation test
Takanishi Laboratory Home



 Medical issues in our country are aged society, doctor shortage, etc.
 Among all medical imaging modalities, echography examination offers quick, low-cost, and noninvasive diagnosis for many pathological situations. However, it is a skilled and "operator-dependent" technique.
 To solve those problems, a test platform development for automated ultrasound diagnosis by robot is proposed. By doing so, a solution to doctor shortage and examination reproducibility are assured.


 The purpose of this study is to develop a robot system which obtains diagnostic images autonomously.
 Target organ is liver, which observation is relatively easy compared with other abdominal organs, A 6-DOF industrial robot is used.
 The goal of this study is to scan all part of the liver without missing any part and obtain diagnosable images.

2.Liver ultrasound diagnosis

Liver ultrasound diagnosis

 Robotic automated scanning protocol is designed based on the sonographers' textbook.
 Probe scanning is mainly composed of longitudinal (probe is parallel to the body center) and transactional (probe is orthogonal to the body center) scanning.  Longitudinal scanning is divided into "Probe long" (around the epigastric fossa) and "Intercostal" (between two costae) scanning.  Transectional scanning is divided into "Probe wide" (around the epigastric fossa) and "Costal arch" (along the costal arch) scanning.  By performing those four sequences in turn, the whole part of the liver is inspected without missing.
 During the scanning, the pivot scanning is required to observe liver widely. Patient's body position is supine position.  To obtain clear images, doctor instructs the patient's timing of breathing. In our research, timing of breathing is instructed by doctor, not robot.

Scanning protocol

Required specifications


 Required specifications are as follows.
 (1) After initial positioning (at epigastrium),no human intervention is required.
 (2) Scanning protocol is based on the sonographers'textbook.
 (3) Robot scans all part of the liver.
 (4) Robot obtains the images clear enough to be inspected by doctor.

 Techniques to be installed into the system to achieve the above requirements are as follows.
 (a) Force control and image feedback keeping contact probe regardless to the patient body types and the movement of the body surface because of breathing.
 (b) Image processing to detect the edge of the liver

Pivot Scanning

3.System configuration


 The proposed system is composed of a 6-DOF industrial robot, a 6-axis force-torque sensor, an ultrasound diagnostic system, and PC.
 The overall view, block diagram are as the following figures.
 The four sequences in thescanning protocol and pivot scanning patterns are given by pattern generator.
 Robot detects the reaction force applied for the patient skin by the 6-axis force-torque sensor attached to edge of manipulator, and modify its trajectory to keep probe contacting to the patient skin by constant force control.
 Ultrasound images are captured by PC through video capture board. During the scanning, the probe posture is always adjusted if non-visible area appears in the ultrasound image, so that images are kept in the condition that doctor can diagnose.

The overall view of system

Block Diagram of the System

(1)Scanning protocol

 Several starting points in 6 axis are predefined according to the scanning protocol.
 The probe is moved to the starting point by interpolating the joint angles. By using joint angle command, limitation of the workspace and singularity can be avoided.
 After reaching at the starting point, the predefined trajectory with pivot scanning pattern is given to the robot in the coordinate attached to the probe face.

Scanning protocol

(2)Press force control

 Probe has to be kept contacting to the patient body regardless to the individual body shape and movement of the body surface due to breathing.
 Among the 6-axis of the trajectory given by the pattern generator, one translational axis tangential to the body surface is adjusted by the force control.
 To detect the force at probe face coordinate, the force detected by force sensor is transformed into the probe coordinate system. The constant force control is implemented by applying the virtual compliance control.

Block Diagram of force control
Block diagram of force control
Virtual compliance control
Virtual compliance control

(3)Probe angle adjustment

 If there is a gap between the left or right edge of probe and skin surface, a vertical dark zone in the ultrasound image appears in left or right part.
 To detect this, the images is divided into 5 parts, and average brightness of each region are calculated. If average brightness of 2 edge region is less than threshold, a dark gap is detected.
 Then the robot rotates the probe around the z axis towards the dark gap region.

Probe angle adjustment
Principle of the probe angle adjustment

4.Evaluation test

(1)Constant force control evaluation

 After implementing the constant force control, it has been applied for 1 healthy volunteer.  Probe was put on the costal near epigastrium. Force control is performed while the patient is breathing.  The reaction force was measured by force sensor.
 Feedback gain was adjusted so that robot does not vibrate even if the patient takes a deep breath.
 As a result, error was ±50[gf] when the reference force was 250[gf].

Constant force control

(2)Validation of the proposed system

 After implementation of the proposed automated scanning system, it has been applied for a healthy volunteer.  (After initial positioning of the probe, and start button pressed, no human intervention is allowed except emergency stop)
 The results are shown in the following i), ii).

Automated Scanning
  Click here to view the movie
  Automated Scanning


 Operation test has been applied for 1 healthy volunteer. Three clinicians (Doctor:1,Sonographer:2) were answered to the questionnaire. Questions are as follows.

1. Diagnosability: If the clinician can diagnose by the image obtained by robot
2. Scanning trajectory:If the scanning trajectory is effective
3. Image clarity:If the image contains any dark gap

 About「Diagnosability」, all have answered that if visible, diagnosis can be done by obtained image.
 About「Scanning trajectory」, Scanning trajectory of "longitudinal scan" had good score, but the other three had low scores.
 About「Image clarity」, Scanning trajectory of "longitudinal scan" had good score, but the other three had low scores.

Questionnaire results

A)Unconfirmed region

 Three clinicians were asked to specify the unconfirmed region in the liver by marking in the drawing of the liver (see figure).
 Unconfirmed regions were mainly lateral end part, right lobe of the liver, under costal part and right costal arch part.

Unconfirmed region


The clinicians confirned that diagnosis by using images which robot obtained is feasible.
However,there are unclarity image's and unconfirmed region.
To use this system for everyone, a function that recognize liver's region from image is required. Also we found that the rotational movement of the probe around z axis is required during the diagnosis. By the examinations on this test platform, required specifications to complete the automated liver scan have become clear.


We would like to express our thanks to Prof. Saito of Tokyo Women's Medical Univ.Prof., Sugawara of Himeji Dokkyo Univ., Prof. Niki of Tokyo City Univ. and Hitachi Aloka Medical,Ltd. for their daily suggestions and supports. And also thanks to SolidWorks Corp. for their support of CAD software.
 This study has been carried out under the permission of the Ethics Committee on Human Research of Waseda University.

Himeji Dokkyo University
Tokyo City University
Tokyo Women's Medical University.
Hitachi Aloka Medical,Ltd.
SolidWorks Corp.

Research Introduction Page To Home Page Copyright(C) 2011 Takanishi Laboratory
All Rights Reserved.