Convolutional neural network which is one of the deep learning methods has dramatically improved the state-of-the art in the field of computer-aided diagnosis. Therefore, it may change the styles of radiologists’ diagnoses dramatically.

抄録全体を表示

Artificial Intelligence, which is gathering attention in the AlphaGo or the field of automatic driving vehicles, without exception, also has a major impact on the medical imaging field. In the medical imaging field, research on computer-aided diagnosis has been actively conducted for a long time, much of which utilizes technology classified as artificial intelligence. Along with the development of artificial intelligence technology in recent years, it is expected that the performance of computer-aided diagnosis will also improve, and there may be a few radiologists who are concerned about the future. However, I anticipate that artificial intelligence will not replace radiologists but will be developed as an auxiliary tool for diagnosis. In this paper, prior to the advent of the era of artificial intelligence, we will explain the principles of machine learning and show some examples in the medical imaging field.

抄録全体を表示

Computed tomography (CT) fluoroscopy-guided interventions such as ablation and biopsy are performed by inserting a specific needle into the lesion under CT fluoroscopy-guidance While those have various advantages including less invasive procedure, short procedure time, and low cost, a major concern to physicians is intraprocedural radiation exposure. In order to address this concern, we hypothesized that a remotely controllable robot could free physicians from radiation exposure. Since 2012, therefore, we have developed such a robot (Zerobot^®), the tasks of which are to hold, target, and insert a needle under CT-guidance by a physician’s remotely manipulating a controller. The robot is floor-mounted and specially designed for a sliding-gantry CT scanner for CT-guided interventions. Phantom experiments showed that robotic needle insertion was equivalent in accuracy without exposing the physician to radiation, in comparison to manual insertion. Animal experiments showed that it was feasible, safe, and accurate in an in vivo procedure. In this article, we provide an overview of the robot, including future perspectives.

抄録全体を表示

We developed a smartphone application named SmartPuncture to assist conventional CT-guided puncture without CT fluoroscopy, in this article we demonstrate the use and advantages of this application.
A puncture guideline is displayed by entering the angle into the application. Regardless of the angle at which the device is being held, the motion sensor ensures that the guideline is displayed at the appropriate angle with respect to gravity. The angle of the smartphone’s liquid crystal display (LCD) is also detected, preventing needle deflection from the CT slice image. Physicians can perform the puncture procedure by advancing the needle using the guideline while the smartphone is placed adjacent to the patient.
In an experimental puncture test using a sponge as a target, the target was punctured at 30°, 50°, and 70° when the device was tilted to 0°, 15°, 30°, and 45°, respectively. The punctured target was then imaged with a CT scan, and the puncture error was measured. The mean puncture error in the plane parallel to the LCD was less than 2°, irrespective of device tilt. The mean puncture error in the sagittal plane was less than 3° with no device tilt. However, the mean puncture error tended to increase when the tilt was increased.
This application can transform a smartphone into a valuable tool that is capable of objectively and accurately assisting CT-guided puncture procedures.

抄録全体を表示

There are various types of electrodes & generators for radiofrequency ablation (RFA). An internally cooled electrode is frequently used because it is simple and easy to use. However, the longitudinal major axis of the ablation zone is larger than the horizontal minor axis. Therefore, it is sometimes technically difficult to achieve a sufficient ablative margin without normal tissue damages.
A new RFA generator system with length-adjustable electrodes was approved for insurance purposes in March 2015. With this electrode, which can the change ablation volume with a single needle, multiple tumors in diameter of different sizes can be treated all at once without unnecessary organ damage. A total of 243 patients with 981 malignant tumors were treated with this new RFA generator system with the electrodes in our clinic from April 2015 to December 2017.
We discuss the usefulness of RFA for malignant tumors with length-adjustable electrodes.

抄録全体を表示