The fact that someone is dying whose brain may be relatively intact gives you a lot of options. Only in the situation where the cancer was observed to have metastasised to her brain and created an abundance of tumors would I consider going the route you seem to be considering (saving the memories in a digital video/audio/written approximation).
If the cancer has not/does not metastasise to the brain the *best* thing to do is to consider whole body, or in the case of disseminated cancer, "head" (brain) cryonic suspension. I will assert that until the cancer, or you indirectly through cremation or burial, has caused a significant disassembly of the cells and synapses which contain the memories of an individual that individual IS NOT IRREVOCABLY DEAD. They are simply beyond the reach of current technologies to present the appearance of what we consider to be "alive". There is a range of states from alive to "disassembled" and unfortunately society and most backgrounds lock us into a two-state mentality. I have told and will continue to tell people in facing this situation that considering cyronics is the only rational thing to do. You may choose not to exercise that option, and there may be legitimate reasons for doing so, financial, not wanting loved ones to have to deal with semi-alive/dead states, etc. but not doing so is irrational.
I personally know the people such as Greg Fahy, Ralph Merkle, Robert Freitas and others who have worked on improving the methods of cryonic suspension and methods for eventual reanimation. These are very serious (and very bright) individuals whose commitment to this area isn't going to go away and is highly likely to eventually yield positive results (IMO).
There is also the prospect that cures to cure her cancer may be developed in the not so distant future. If it were me personally and I didn't want to take the suspension route (which involved legally "dying") I would consider a "suspended animation" route which would involve a low temperature coma state or a H2S induced "suspension state" (which would be considered "risky" to "experimental" in conventional medical circles). I would then make sure I got a set of cancer biopsy samples to a lab capable of "extreme cutting edge" genomic medicine [1] to answer the precise question of "what genetic mutations have occurred in this/these cancer(s) and what specific drugs are available to target those pathways?" [2]. Some pathway specific drugs are available now. More will be under development. Until such time as one can produce a "cocktail" of drugs, perhaps combined with site-specific nano-targeting methodology, one should pursue a strategy which slows down the individuals "rate-of-living", thus "suspended animation", as much as possible. It may be that this would require an individual to be kept barely alive for 5-10 or more years but this is not outside the realm of medical capabilities at this point (though it would likely be quite expensive). You would also have to locate a team of individuals willing to be this "experimental" with a human life.
1. For example one would want to be able to "match" the patient with one or more of the cancers being studied in the various "Cancer Genome Atlas/Anatomy Projects" which will eventually have analysed and classified tens of thousands of cancers.
2. If you are not dealing with an oncologist or oncologist team which can tell you specifically *what* oncogenes and tumor suppressor genes are broken in the current cancer you are dealing with "snake-oil" physicians. Cancer genomics is *way* beyond the stage of surgery/radiation/toxic-drugs which have worked in the past yet most oncologists have probably not moved beyond those as the standard therapies.