Summary: Your image-enabled personal healthcare records belong on your smart phone, the instant that they have been created.
Imagine you are a patient, and as you climb off the CT table, return to the dressing room and prepare to leave, your smart phone or similar device, which is always turned on, has already authenticated itself on your behalf to the PACS, and proceeded to download the complete set of images just acquired (which come out of the reconstruction pipeline almost instantly nowadays).
Then, whilst sitting in your doctor's waiting room (or perhaps whilst stuck in traffic on the way there; naughty, no reading images whilst driving) you could scroll through the images and decide what questions you wanted to ask about them, perhaps with the benefit of the previous images and reports that you already had, or even with the fresh report that just popped up as the radiologist (finally!) got around to reporting the study half an hour after it was acquired (probably the delay was negotiating for the lowest bidder to report it in India or China).
Technical challenges of bandwidth, storage, compression and so on aside, this is entirely within the realm of possibility right now, much less in a few years, when we will fondly recall how limiting 4G was and how quaint 2GHz quad-core processors were. As applications and services providing seamless synchronization between an individual's myriad devices proliferate, together with remote "backup" (on "cloud" providers), the sort of device-independent availability of information that users already expect for mail, contacts and documents, will become available for healthcare records too.
The imaging use case is more challenging, given the large volume of data, but it is qualitatively if not quantitatively similar.
As for security, authentication and access control, most healthcare facilities seem more than capable of finding ways to identify the patient in order to bill them or or their insurance company. It seems likely that just as one typically enables personal devices like phones to automatically authenticate to commercial service providers for shopping, music and video access, a patient that establishes a relationship with a medical facility ("registers") could at that time (or subsequently) register their device, just as one does with any other online "portal", and the device would cache the credentials (or not, depending on the paranoia and personal policy of its owner).
Subsequently, the "patient health care record app" on the device would take care of automatic synchronization with the most recent record information, at every provider that the patient visited, seamlessly, automatically and silently. It could merrily beep as it received new information, make more arousing noises to notify when more critical information arrived (like an out of range lab test or abnormal radiology result). As the longitudinal record grew it could flush from its cache older information (on a rule driven basis), with the patient secure in the knowledge that the multiple "cloud" servers on which everything was backed up and the entire record stored would allow them to retrieve older stuff on demand as necessary. Geographic proximity to the provider would be unnecessary, and the device would, as they typically do already, even for voice, adapt to the fastest accessible network (Wifi if local, 4G or less if not).
Then, when visiting a new provider and registering with them, the reverse could occur. The patient's own device could synchronize whatever the patient wanted to share with the providers systems immediately on registration, or communicate the unique identifiers of each piece of information to the provider's system so that it could synchronize it from the cloud servers if not locally cached.
With nearly half of all US phone subscribers acquiring a new phone now choosing a smart device, it seems inevitable that the penetration of these devices will asymptotically approach ubiquity.
The single device that a person already carries offers the potential to eliminate the need for paper tokens like airplane boarding passes or entertainment tickets, and for payment, instead of cash and credit cards, and already serves many of us as the primary interface for online information gathering, decision making, navigation and shopping (see it, snap a picture or bar code, buy it online). It seems inevitable that it will also become the primary conduit for all personal information access and storage, so why not healthcare information, including diagnostic images, too?
The idea of a PHR on a smart phone is incredibly obvious, and already various PHR groups are starting to provide such apps and services with varying degrees of sophistication and integration (see, to name only a few, motionPHR, MyMedicalRecords.com, mobileStorm) and it is inevitable that the sophistication of all aspects of "mHealth", as it is now referred to, will grow at a phenomenal pace. Extending it to include diagnostic imaging beyond the current crop of simple "viewers" (like Mobile MIM) to include seamless synchronization as part of the integrated personal record seems like a no-brainer.
Perhaps by the time we get to Stage 3 of Meaningful Use, the neanderthals who colluded to cripple Stage 2 will have been outpaced by more disruptive innovators, and it will be timely to reimburse specifically for seamless patient-controlled synchronization of images and everything else, not only in a general "download and transmit" sense, but with a specific certification requirement to permit this via their mobile devices.