The Costs of Computer Based Medical Technology

My take on Arnold Kling's article on the "Myth of Massive Health Care Waste" has generated quite a bit of discussion on Kling's EconBlog, here. What generated the most interest was the question of whether the cost of computer based medical technology was too high. (The example used was the CT scanner.)
Some of the main conceptual points raised were:

• One needs to separate the costs of the equipment and its maintenance from the other costs incurred by hospitals or offices in doing CT scans, and these, in turn, need to be separated from the actual price of CT scans
• Hardware costs need to be adjusted for inflation, but more important, the increasing capabilities of the hardware need to be taken into account

Some striking opinions were:

• "I think you'll find that in real dollars, the price of CT equipment and technology as well as the cost per procedure has dropped substantially...."
• "The PC hardware is a small part of the technology."
• "GE Medical dominates the market and can charge pretty much what it wants."
• "as quality/performance rises, 'prices' don't necessarily fall continually...."
• "If high fixed costs are the problem, more volume (more patients, more procedures) will lower the cost per procedure."
• "I was part of a design team for a CAT scanner.... I proposed many approaches that would have saved quite a bit in both hardware and software costs without reducing functionality, reliability, and performance. I was frequently met with the answer, 'We want it to be expensive.' The medical technology marketplace is a perverse marketplace. There are too many docs involved in the process and they don't respect anything unless it's expensive."

To make this discussion a bit more data driven, I was able to find some about the economics of CT scans over time. [Stockburger WT. CT imaging, then and now. Radiology Management, November/December, 2004. On the web here.]
From 1974 to 2004, the list price of a CT scanner has gone up faster than inflation (from $385K to $2200K, 471% increase, versus 342% increase in consumer price index). So has the acquisition cost ($385K to $1600K, 416%). However, revenue obtained per CT scan has gone up even faster (the Medicare payment for a CT scan of the head without contrast increased 700%.) Finally, and perhaps most importantly, the amount to be made by CT scanning facilities has risen even faster, because, as noted above, it is now possible to do more CT scans per unit time. In 1974, by my calculations, the typical profit of a single-scanner facility would have been $56700 in 1974 dollars. In 2004, it would have been $577,900 in 2004 dollars! Even adjusted for inflation, that's a big increase (1030% unadjusted, 298% adjusted for inflation)!
So, to summarize, my hypothesis that the costs of CT scanners over the last 30 years have become exaggerated in the face of the vast increase in computer performance/ price, given that the "C" in CT stands for computer, remains unproven. However, there is anecdotal evidence (from the comments above) that suggests that these prices may have been artificially elevated by a less than competitive marketplace plus the industries interest in wanting it to be expensive.
More importantly, the data from the article above suggests the larger problem. Despite the rise of managed care and of government efforts to contain costs, reimbursement for CT has risen faster than costs and inflation, and given that CT scans can be done more and more quickly, which raises volume while it further reduces the fixed costs associated with each scan, the amount of money made by CT scan facilities has risen even faster.
So the big question remains. If managed care was supposed to restrain costs by using a more business-like approach, and more attention to the economic issues, why has it so massively failed to restrain these sorts of costs?