I was very disappointed in the title and context of "Potential for Fraudulent Use of Digital Radiographs" by Drs. Andrew Tsang, David Sweet and Robert E. Wood, which appeared in the September issue. After having heard a number of my peers (students) and instructors (dentists) quote the article as reasons for "keeping film" or "why we can’t go digital yet" or "why insurance companies won’t accept digital from us," I realized that clarifications are required.
First of all, fraud of this type has nothing to do with digital radiography, or DR. The authors did not use DR. They used a digitizing process to commit fraud, and picked "analog" radiographs as the vehicle. They could have picked the chart, the diagnosis or other information sent to insurance companies as the material to digitize and modify, then reprint-as-new, thus committing fraud. Remember, fraud became easier with the invention of the pencil eraser, and has become even easier to commit and harder to detect with the advent of all sorts of technology—copiers, color copiers, digitizers and so forth.
Second, in more than one case, the authors maintain that modifying DR is easier. I would claim not. The easiest method is as they did it—take a normal film, digitize it with a relatively cheap scanner, and complete the process as they did. Good scanners are now in the low $100s. If you want to do this to DR, you have to spend $10,000 to get the DR system, and then you have to understand the file system and file format to dig out the "pictures" in the computer and commit the fraud. Understanding the file system is not something the authors explained (see the next reason).
Finally, there seems to be an assumption that DR contains no authentication and that digital information can’t be guaranteed to be pure. These are dangerous thought trends. When you understand cyclic redundancy checks, error-correcting code memory, digital watermarks, embedded encryption and even basic encryption techniques, you begin to realize that digital data can be far more authentic than other mediums. It is far easier to retouch a negative or photograph than it is to modify a digital picture if any of the above techniques are embedded in the data. Unless you have access to the developers’ proprietary software, you should never assume you can identify and/or successfully defeat authentication schemes—even then it may be impossible unless you have the key. It is true that we should require these schemes in our DR systems, but the fact remains that the potential for fraud is actually, and will forever be, on films, as Dr. Tsang and colleagues so aptly demonstrated.
