Development of Real-Time CT and CT Fluoroscopy

1. Introduction
The history of computed tomography (CT) can be viewed as a reduction in calculation times. In 1971, G. Hounsfield developed a prototype CT scanner that required 4.5 minutes for image acquisition and 20 minutes for image reconstruction of a single slice after the raw data was brought back to the laboratory1). CT technology subsequently underwent rapid progress, permitting the scanning time to be reduced to 1 second over the next 20 years. Images can now be displayed a few seconds after scanning is completed. However, to our knowledge, no attempts had been made to achieve true real-time observation of CT images. One interventional radiology textbook stated that "CT images cannot be observed in real time, unlike ultrasound images."2) No one tried to improve this situation because everyone assumed that a long time would always be required for the scanning and reconstruction of CT images. In practice, various processing steps such as scanning, preprocessing, data transfer, coordinate transformation, convolution, back projection, image display, and data storage must be performed before CT images can be displayed on the monitor screen. In order to achieve the real-time observation of CT images, all these steps must be completed within 1 second at the most, which is difficult to achieve even with the latest CT technology.
We were able to realize the goal of real-time CT, which had never been attempted before, by assembling a team comprising both engineers and clinicians. In this paper, the process of development of real-time CT will be reported from a viewpoint of a member of the development team. The future prospects of real-time CT will also be discussed.

2. Background of the development of real-time CT
Initially, we were by no means certain that we would be able to develop a practical real-time CT scanner from the clinical viewpoint. In 1986, a slip-ring CT scanner (TCT-900S, Toshiba, Tokyo) was installed at our institution. It was the installation of this new CT scanner, which permitted us to perform scanning with no interscan delay, that motivated us to develop real-time CT. We explored a number of new possibilities, including the development of various dynamic CT techniques to most fully exploit the advantages of continuous scanning. Not only contrast studies such as dynamic subtraction CT of the central nervous system diseases(DSC-CTA) 3), but also dynamic CT studies of body movements such as respiration (dynamic respiratory scanning , DRS), external ocular muscle movements (EOM-CT scanning) have been developed and clinically applied 4,5). Among these studies, DRS was particularly important in demonstrating the potential of helical scanning. We were assured that satisfactory image quality could be obtained even though scanning was performed while the patient couch was moved, since acceptable pulmonary images were obtained by DRS despite the vertical movements of the lung during respiration. In 1987, we performed our first helical scan study. Throughout our investigations using dynamic studies, we eagerly hoped to observe these vital movements in real time.
In parallel, the engineering group continued with their technical investigations into making the best use of the advantages of continuous-rotation slip-ring CT. Basic studies on helical scanning were performed in 1982, and patent applications were submitted.6) A reconstruction technique to improve the apparent temporal resolution in dynamic scanning was developed, which was applied in the real-time CT reconstruction afterwards. Efforts to minimize the image reconstruction time bore fruit with the development of the fast pipeline reconstruction unit (real-time reconstruction unit: RTRU) in 1990, which was incorporated into the CT reconstruction system. Ohashi developed a new algorithm, for which patent applications have been submitted, in which raw data is divided into 60-degree segments, permitting six images to be generated from the data obtained in one rotation.7) The engineering team led by Yutaka Shibata shared the dream of developing a CT system in which images could be observed in real time. Real-time CT was thus made possible by combining the dreams of both engineers and clinicians.

3. Development and presentation of CT fluoroscopy
August 28, 1993, is a memorable date for us because that was the day that the basic development goals of real-time CT were first established. On that day, four people, three responsible members of the CT development team and I, held the first of many meetings on the development of real-time CT. Representatives of the engineering group initially felt that such a development project was premature. In contrast, we clinicians were quite enthusiastic about the project, spurred on by our earlier experience with helical scanning: we had started our experiments so early, had obtained patents, and had almost reached the point of production of a helical CT scanner, but the German group beat us to the punch in presenting the application of helical scanning.8) Vowing not to repeat the same mistake, all of the participants agreed to undertake the development of real-time CT.
We immediately set clear goals for several basic characteristics of real-time CT. Clinically, the first objective was to develop real-time CT-guided biopsy. Since that time, the name "CT fluoroscopy" has been used to describe one of the representative applications of real-time CT. The target date for the development project was set as mid October of the same year, in order to meet the deadline for the Hot Topics session of the Annual Meeting of the Radiological Society of North America (RSNA). We also established the following goals: a delay time between scanning and image display of 1 second or less, the provision of an additional monitor and control panel in the CT scanning room, and the development of a needle holder.
The major technological hurdle was image storage. It took at least 1 second to store images on hard disk. I insisted that images should be stored on videotape rather than on hard disk, due to the unacceptable delay time. Finally, we all agreed that images should be stored in large-capacity memory.
In September, production of a prototype real-time CT scanner was started based on an existing CT scanner, and the first real-time CT system was completed in October thanks to our familiarity with the basic technology of real-time CT and the engineers' unflagging efforts. Phantom experiments were then conducted at the factory. In these experiments, an acrylic rod was moved slowly in the vertical direction during scanning. The rod movement was displayed in real time, and the results of these experiments were considered successful (Fig.1).
The first CT fluoroscopy system was installed at our hospital in October 1993 by adding a new image reconstruction subsystem and software to an existing CT scanner. On the day of installation, I participated in the first trial as a volunteer, and waved my hand within the gantry (Fig.2). We confirmed that CT images were displayed in real time at a rate of three images per second.
Before starting the clinical application of CT fluoroscopy, we eagerly discussed the wide range of applications of this new method. We decided to focus on lesions in the chest and head, where ultrasound examination is difficult, rather than the abdomen, where ultrasound-guided biopsy had already established itself as a useful method free of the disadvantage of X-ray exposure. The first clinical application of CT fluoroscopy was the drainage of intracerebral hematoma. During stereotaxic puncturing of the hematoma, the needle was observed on the monitor in real time (Fig.3).
We presented the results of our studies at the Hot Topics session of RSNA '93, held in November of that year.9) Immediately after our presentation, several interventional radiologists from a variety of the country expressed interest in our work. Today, real-time CT has proved its usefulness as a tool for monitoring or navigation in biopsy procedures, drainage procedures, and so on.10,11)
4. Other applications of real-time CT
CT fluoroscopy is perhaps the most exciting application of real-time CT. However, this is not a universal application intended for use at every hospital. We have developed the following two functions to further expand the usefulness of real-time CT.
The first function is real-time helical scan. Using this function, conventional helical scan data can be reconstructed and displayed in real time, so that the slice that is being scanned can be viewed immediately. This permits the optimal scanning range and scan timing to be determined in real time during the examination, thus minimizing the number of failed examinations. Since the end-point of helical scanning can be confirmed in real time, both patient exposure and the X-ray tube cooling time can be reduced. In addition, the operator does not need to wait for all of the images to be reconstructed after scanning, and the patient can leave the examination room immediately after scanning is completed, minimizing patient discomfort and maximizing patient throughput.33)
The second function is SureStart, which was developed by Dr. Hirofumi Anno. SureStart makes it possible to determine the optimal phase in contrast CT studies.12) As is well known, the time required for intravenously injected contrast medium to reach the target region is affected by various factors such as the patient's cardiac function and so on. As a result, the optimal scan timing differs from patient to patient. When SureStart is used, the scan timing can be determined by directly observing the contrast enhancement of the target blood vessel after the injection of contrast medium. As a result, the optimal phase can be precisely and easily obtained in all patients.
These two functions are general-use applications that have been developed not for special cases at specialized institutions, but for the benefit of a large number of patients. The use of these functions enhances the quality of examination, minimizes patient discomfort during scanning, and increases patient throughput, bringing benefits to radiologists, radiological technologists, patients, and hospitals. After experiencing real-time CT firsthand, users immediately realize that performing a conventional CT study is largely based on clinical experience and subjective judgment, and may therefore find it difficult to return to conventional CT examinations without real-time reconstruction. These two functions will surely be considered indispensable in the future.

5. Future prospects for real-time CT
Since the development of real-time CT, Toshiba's engineers at the Nasu factory and the staff of Fujita Health University have held periodic meetings to discuss the future development of this exciting new technology. More than one hundred items and ideas have been discussed in these meetings. Steady advances are being made in real-time CT. Among them, the most obvious is improved temporal resolution. In the prototype produced in 1993, three images were displayed per second, and in September 1994, this was increased to six images per second. Today, the latest model permits twelve images to be displayed per second with a delay time of 0.3 s, paralleling a reduction in scan time to 0.5 second per rotation. When the delay time is 0.5 s or less, images appear to the observer to be displayed in real time. Therefore, the delay in the latest CT scanners is almost unnoticeable.
What are the future prospects for real-time CT? The answer to this question is closely related to the advances that are currently being made in multi-slice CT. Multi-slice helical CT is expected to lead to dramatic changes in future CT diagnosis, and real-time CT is expected to play an important role in fully exploiting the advantages of this new technology. When multi-slice helical scanning is performed, presetting of the contrast timing is more difficult than in conventional CT scanning, since scanning is completed within a few seconds. This makes it even more important to determine the contrast timing accurately using real-time CT (SureStart). In addition, if multiple slices (in this case, three slices) can be displayed simultaneously in real time, it is possible to immediately detect any deviation of the puncturing needle from the center slice in CT fluoroscopic biopsy. Furthermore, if longitudinal images could be reconstructed from several tens of slices in real time with fast image processing, biopsy procedures could be performed while observing sagittal or coronal images, as is possible in MRI-guided biopsy.
The most outstanding feature of real-time CT, as represented by CT fluoroscopy, is that various CT examinations and procedures that were previously performed "blind", relying on the radiologist's clinical experience, can now be performed with real-time visualization. It is expected that real-time CT will be incorporated into future CT scanners as a standard feature.

Acknowledgments
The author thank Hirofumi Anno, Ryoichi Kato, Gen Takeshita, Yuko Ogura, Sukehiko Koga, Katsumi Tsujioka, and Yoshihiro Ida whose support and cooperation has been invaluable. The author also acknowledge the staff of Toshiba Corporation, including Mr. Yutaka Shibata, Mr. Yoshiki Hirao, Mr. Mitsuru Yahata, Mr. Toshihiro Rifu, Mr. Masahiro Ozaki, and other members of the CT group at Toshiba's Nasu factory, who have devoted themselves to the development of real-time CT.

References

1. Hounsfield, G.N.: Computerized transverse axial scanning (tomography). 1. Description of system. Br J Radiol, 46: 1016-1022, 1973.
2. Seibel RMM, Gronemeyer DHW : Interventional Computed Tomography. Boston, Blackwell Scientific Publications, 1990.
3. Takeshita, G.: Dynamic subtraction cine CT angiography in the central nervous system. In K. Kimura and S. Koga (Eds): Basic principles and clinical applications of helical scan, pp16-36. Tokyo, Iryokagakusya, 1993.
4. Anno, H., Koga S., Ikuta K. et al.: Development and Usefulness of Dynamic Respiratory Scanning in Fast CT. Radiology, 173(suppl.): 456, 1989.
5. Tomita, K., Katada, K., Koga, S. et al.: Detection of abnormal external ocular muscle movement with dynamic CT. Radiology, 177(suppl.): 348, 1990.
6. Mori, I.: Computerized tomographic apparatus utilizing a radiation source. U.S. Patent: 4630202, 1986.
7. Ohashi, A.: US patent: 4495649, 1985.
8. Kalender W, Seissler W, Klotz E, et al.: Spiral volumetric CT with single-breath-hold technique, continuous transport and continuous scanner rotation. Radiology 176:181-183, 1990.
9. Katada K, Anno H, Koga S, et al.: Initial trial with CT fluoroscopy . Radiology, 190(suppl):662, 1993.
10. Katada K, Anno H, Ogura Y, Takeshita G et al.: Development and early trials of real-time CT fluoroscopy, Neuroradiology 37(suppl.):587-588, 1995.
11. Katada K, Kato R, Anno H, Ogura Y, et al.: Guidance with real-time CT fluoroscopy: Early clinical experience. Radiology, 200: 851-856,1996.
12. Anno, H., Katada, K., Kato, R. et al.: Scan timing control in contrast helical CT studies using the real-time reconstruction technique - Development of the SureStart function. Medical Review, 60: 5-12, 1997.

Figure legend

Fig. 1. First phantom study for CT fluoroscopy.
a. An acrylic phantom with moving acrylic rod in center.
b. Movement of the acrylic rod simulating puncturing needle was successfully displayed at a speed of 3 images per second.

Fig. 2. First volunteer study for CT fluoroscopy.
Motion of volunteer's hand was displayed in real-time on an additional monitor.

Fig.3. First clinical application.
Stereotaxic puncture of intracerebral hematoma. An arrival of needle tip to the hematoma was monitored with the aid of CT fluoroscopy.

- go to the top of this page