Abstract In the last three years, many research laboratories and commercial firms have been very actively interested in computer-assisted tomography (CAT) and the very rapid expansion of the field makes whatever is written today out-of-date before it appears in print. In our laboratory we have been investigating basic principles of whole-body CAT scanning. We are building a 256 xenon detector array in a fan geometry, having beforehand carried out extensive measurements on a 10-channel prototype array, operating at a pressure of 5 atm of xenon and obtained detailed data on the saturation characteristics of this prototype. For beams from diagnostic X-ray machines, saturation is difficult to achieve. With about 1000 V across a 1.5-mm gap in the chamber, about 95% of the ions can be collected. Chambers with wider gaps are probably usable. Designing a chamber which is linear over a wide range of exposure rates is difficult; by attenuating the X-ray beam with layers of aluminum, we have been able to test the detector for linearity. The absorption of radiation in one chamber introduces about 1.5% “cross-talk” radiation into adjacent ones, which is probably not a serious problem. Scatter of radiation from the whole fan into the detectors can be minimized by radial collimators and reduced to the level where it is not a problem. The emission of “off-focus” X-rays, a more serious problem, can be removed only in part with suitable diaphragms. Improved design of X-ray targets may be required. One of our two experimental X-ray sources produces 30 pulses per sec, 9.3 ms long. The pulsing, done in the primary of the generator, is very difficult to stabilize. Sequential pulses change in size as different parts of the rotating target are bombarded by the electron beam, which suggests that the emission from even a new target is dependent on its angle. We are comparing this source to a highly stabilized DC X-ray source. Electronic circuits that measure the currents in xenon ion chambers are needed to present them in usable form to the PDP-11T55 computer. One particular circuit is able to handle a large dynamic range, and measure to a precision of about 0.2%. Several types of computers are ideally suited to image reconstruction. Use of the convolution algorithm to construct an image from fan geometry causes artifacts that are considerably different from those produced in a “translate-rotate” geometry. In theory, liquid xenon chambers should be about five times more sensitive than gaseous chambers; liquid xenon might be no more complicated to handle than high-pressure gas. CAT scanners may shortly become indispensible for radiation therapy planning, especially where high-LET radiation is used and corrections for inhomogeneities become very important. In principle, we now have sufficient information to calculate precise isodose distributions in patients.
Read full abstract