Subject: Re: PATNEWS: How to bust a spectroscopy software patent
From: Alan Engel
Date: Tue, 10 Dec 1996 07:32:02 -0500
Greg,
Oh yee of such boundless technical knowledge. Have you ever run an
infrared
spectrum or used one in research?
The non-patent prior art as you present it does nothing to 'bust' this
claim.
In my mind, it actually appears to do the opposite.
To establish a reference date, when I was actively using infrared for
chemical research in 1986, using IR for identifying tissues
was embryonic. An invention for 1) using infrared spectroscopy to 2)
identify tissues by 3) using pattern recognition techniques was probably
not possible at that time.
It is now 10 years later. What has happened according to your title
search?
First, the use of IR to characterize tissue types (not the chemicals in
them)
had its early rush of research in the two to three years prior to the
July 1994
filing.
Second, the use of pattern recognition to identify chemical compositions
had its
research rush in the late 1980's. In looking at your titles on
automated
IR identification, I would bet that they all deal with identifying
chemical
components (aldehydes and the like) and not one mentions IR of tissues.
In my mind (that of a B.S. in chemistry and a Ph.D. in biochemistry),
the leap
from identifying chemicals to identifying tissues is not at all
obvious. The
prior art, as you present it, does not suggest to me that the automated
identification of tissues based on IR has become obvious in the last 10
years
even with issuing of the patent in question.
Gregory Aharonian wrote:
>
> Recently a patent issued that was bad enough to be easy enough to bust
> that I wouldn't mind doing so over the news service. It deals with infrared
> spectroscopy, which is apropos since an article appeared earlier this year
> where a whole bunch of professors complained about the really bad spectroscopy
> patents being issued. This patent shows the dysfunctionality of the patent
> system at its worse.
> The details. The patent issued this July to the National Research Council
> of Canada (shame on my NRC readers :-), number 5,539,207, and with the benign
> title of "Method of identifying tissue", with the following first claim which
> pretty much is also the abstract:
>
> 1. A method of identifying tissue comprising the steps of
> determining the infrared spectrum of a tissue sample over
> a wide range of frequencies in at least one frequency band,
> and comparing the infrared spectrum of said sample with a
> library of stored infrared spectra of known tissue types
> by visual comparison or using pattern recognition techniques
> to find the closest match.
>
> The patent was filed in July of 1994, and cited only one non-patent prior art
> item: "Human Colon Adenocarcinoma Cell Lines Display Infrared Spectroscopic
> Features of Malignant Colon Tissues." Cancer Research, vol. 52 (Jan. 1, 1992)
> pp. 84-88. The picture in the Official Gazette includes a block depicting
> a FT-IR spectrometer. That's it, and thus the outrage - this first exemplary
> claim should never have issued - it is way tooooooooooooo broad to border on
> the banally trivial. Now maybe something in the rest of the claims is novel
> and unobviousness as taught by the specification, but not this first claim.
> And if so, the examiner should have demanded that the novelty appear higher
> in the claims, as high as the exemplary claim. Assuming the examiner was
> allowed to do any sort of non-patent searching to establish a context.
>
> To show this, I did a search the hard way, by querying ISI Science Citation
> Index, which only has titles of journal articles, no abstracts, by Boolean
> ANDing two keywords, "infrared" and "tissue". Here's what I came up with:
>
> Distinctive infrared spectral features in liver-tumor tissue of mice -
> evidence of structural modifications at the molecular level
> Experimental and Molecular Pathology, vol 55 no 3, 1991, pg. 269
> (which not only covers the first claim, but probably
> at least one of the dependent claims dealing with
> cancer, guessing by the one reference cited)
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Infrared microscopy of human tissue
> Applied Spectroscopy, vol 43 no 6, 1989, pg. 1095
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Characteristics and imaging properties of near-infrared radiation
> in tissue
> British Journal of Radiology, vol 59, 1986, pg. 836
>
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Internal reflection spectroscopy applied to quantitiation of tissue
> absorption features in the infrared
> Lasers in Surgery and Medicine, vol 8 no 2, 1988, pg. 142
>
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Fat, protein and moisture analysis of fish tissue by mid-infrared
> transmission spectroscopy
> Canadian Institute of Food Science and Technology Journ., 1988, pg. 364
>
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Near-infrared spectroscopy - a new technique for the non-invasive
> monitoring of tissue and blood oxygenation in vivo
> Biochemical Society Transactions, vol 16 no 6, 1988, pg. 978
>
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Visible and infrared spectrophotometry of tissue or organs - an
> approach to microcirculation and cellular metabolism
> Alcohol and Alcoholism, vol 23 no 3, 1988, pg. A28
>
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Rapid near-infrared raman spectroscopy of human tissue with a
> spectrograph and CCD detector
> Applied Spectroscopy, vol 46 no 2, 1992, pg. 187
>
Where in the above title does it even suggest using IR to identify
tissue,
let alone using pattern recognition to do so?
> Fourier transform infrared (FT-IR) microscopic identification of
> foreign materials in tissue sections
> Laboratory Investigation, vol 66 no 1, 1992, pg. 123
>
Where in the above title does it even suggest using IR to identify
tissue,
(the title says "foreign materials in tissue" - not the same thing)
let alone using pattern recognition to do so?
> Near infrared radiation for diagnosis of tissue
> Biomedizinische Technik, vol 38 no 7-8, 1993, pg. 162
>
Where in the above title does it suggest using IR,
with pattern recognition to identify tissue?
> Infrared and raman microscopy of cellular proteins in human tissue
> specimens
> Biophysical Journal, vol 64 no 2, 1993, pg. A128
>
Where in the above title does it even suggest using IR to identify
tissue,
(the title specifies protein components in tissue, not tissue)
let alone using pattern recognition to do so?
> Tissue proteogylcans studied by infrared spectroscopy
> Bull. of Experimental Biology and Medicine, vol 114 no 9, 1992, pg 1245
>
Where in the above title does it even suggest using IR to identify
tissue,
(the title specifies protein components in tissue, not tissue)
let alone using pattern recognition to do so?
> Admittedly, not the most thorough of searches, but enough to seriously
> question the validity of the first claim. These papers clearly suggest,
> especially to one skilled in the art, that you can use infrared spectroscopy
> to analyze tissues and their contents, even for complicated conditions such
> as cancer tumors - i.e., the first claim. Given thousands of papers on
> tissue analysis using other forms of radiation,
Other forms are not infrared. The technical principles of analysis are
completely different for other forms of radiation.
> this first claim is utterly
> obvious (as a claims chart using the above papers could show).
>
> Now you might argue that what is being claimed is not infrared spectroscopy
> for tissue analysis, but rather some advanced pattern recognition technique,
> again as suggested by the first claim. Nice try, but again extremely obvious
> to one skilled in the art, at least in light of the following articles:
>
> ESSESA - an expert system for elucidation of structures from spectra:
> knowledge base of infrared spectra and analysis and intrepretation
> program
> J. Chemical Information and Computer Sciences, vol 30, 1990, pg. 203
Any mention of tissues? Most likely only chemical structures.
> Artificial intelligence used for the interpretation of combined
> spectral data: automated generation of interpretation rules for
> infrared spectral data
> J. Chemical Information and Computer Sciences, vol 27, 1987, pg. 203
>
Any mention of tissues? Most likely only chemical structures.
> Database development and search algorithms for automated infrared
> spectral identification
> J. Chemical Information and Computer Sciences, vol 25, 1985, pg. 235
>
Any mention of tissues? Most likely only chemical structures.
> Performance analysis of a simple infrared library search system
> J. Chemical Information and Computer Sciences, vol 25, 1985, pg. 241
>
Any mention of tissues? Most likely only chemical structures.
> Automatic testing and presentation of interpretation rules used by
> PAIRS (Program for Analysis of IR Spectra)
> Applied Spectroscopy, vol 39, 1985, pg. 331
>
Any mention of tissues? Most likely only chemical structures.
> Automated rule generation for the Program for Analysis of IR Spectra
> Anal. Chim. Acta, vol 162, 1984, pg. 227
>
Any mention of tissues? Most likely only chemical structures.
> and for the golden oldie:
>
> Infrared spectra of biological macromolecules and related systems
> in Structure and Stability of Biological Macromolecules, Marcel
> Dekker New York, 1969, pg.57
Any mention of tissues? Most likely only chemical structures.
Identifying
chemical structures is not the same as identifying tissues.
> This first claim should not have issued. But it did because of the basic
> dysfunctionality of the system. A simple MEDLINE search would have pulled
> all of these papers, plus more, and yet was not done. Why? Because (the
> chorus sings out) examiners do not have enough time, money and resources to
> even do something as simple as search through MEDLINE, then order the papers
> and then have time to read them and assess novelty and obviousness of such
> claims. This is why the Software Patent Institute and a whole bunch of
> other Patent Office supported and/or funded projects are an utter joke: while
> examiners work without regular access to already organized databases, money is
> being wasted building speciality databases, a deliberate waste when existing
> databases could contribute so much, 11,000,000 times so much.
>
> Now you might object, "Greg - this really bad spectroscopy patent is
> probably the exception, not the rule".
Bad conclusions based on bad searches - Not limited to the patent
office.
Subject: Re: PATNEWS: How to bust a spectroscopy software patent
From: werdna@gate.net (Andrew C. Greenberg)
Date: Wed, 11 Dec 1996 08:42:56 -0400
NOTE: THIS POSTING TAKES TO TASK A VERY DETAILED ARGUMENT BY GREG, SUPPORTED
BY HIM WITH SUBSTANTIAL EVIDENCE AND CITATIONS TO SPECIFIC PRIOR ART, AND I
THANK HIM FOR HIS ENERGIES TO THAT END. THE LAST TWO PARAGRAPHS MAKE THIS
POINT, AND LEST I BE ACCUSED OF "BURYING" MY FAINT PRAISE, I QUOTE MY LAST
PARAGRAPH HERE AT THE LEAD:
> Thus, I honor Greg's effort to lay out a strong argument with sound facts,
> and thank him for adding much light to the debate. I hope my response did
> the same, and pointed out the complexity of these issues, even in the face
> of what at first blush appeared to be very strong facts.
In article , srctran@world.std.com (Gregory
Aharonian) wrote:
> 1. A method of identifying tissue comprising the steps of
> determining the infrared spectrum of a tissue sample over
> a wide range of frequencies in at least one frequency band,
> and comparing the infrared spectrum of said sample with a
> library of stored infrared spectra of known tissue types
> by visual comparison or using pattern recognition techniques
> to find the closest match.
The subject appears to cover a method of __identifying tissue__, with
the following elements:
(1) determining the spectrum; and
(2) comparing the spectrum with a library of
tissue types to find a match
The articles cited, while they do use spectroscopy to determine the
chemical composition of matter (first year physics), do not appear to
involve tissue typing, whatever that is. Ordinarily, an examiner is
required to cite precisely to that portion of the article addressing each
element of the claim to make a prima facie case of invalidity. Greg is
not bound by that rule, but his failure to do so puts his analysis in
question. I suspect that the words "tissue type" are critical to the
issue of patentability.
Did you go beyond the abstract and claims, and actually look at the
patent prior art to see if it raised similar issues, Greg? Did you
check the file wrapper to see if those issues were addressed during
the examination? I have not, of course, but I imagine that there is
a fair chance that patent were cited raising precisely the issues raised
by Greg's non-patent art, and that the obviousness and 102 issues were
explicitly addressed therein.
By the way, Greg, what was the effective filing date of that application?
(You don't say in your report.) Are you certain that all of the articles
cited would even be prior art?
> Now you might object, "Greg - this really bad spectroscopy patent is
> probably the exception, not the rule". You might, but you would be wrong,
> especially if you don't pay much attention to ISSUED patents (as opposed to
> issued court cases).
Fair enough, but I spend a great deal of time analyzing file wrappers and
the actual application of the Patent Act with respect to issues such as'
you describe. BOTH is necessary to justify your conclusions concerning
obviousness and that the non-patent art was overlooked.
Put another way, when I do a search, I pick zillions of patents out of
a shoe, skim through them, selecting those that appear close to material.
When I analyze a patent, I discard the dozen that raise an issue of
patentability, picking the closest one, and discuss only that one.
It is quite possible that the non-patent art does not raise new issues of
patentability over the patent art that was cited. This is an essential
missing part of Greg's claim that failure to cite non-patent art is such
an evil. In many art areas I dare say, (probably not software, yet) the
patent art is at least as well-developed as the non-patent art.
> "It is our belief that there is more than sufficient evidence of
> prior art to seriously call into question recent patent applications
> that make claim to intellectual property that has, in fact, been
> available to researchers for decades".
>
> "Recently, however, these very tenets have appeared in new patent
> applications, particularly those dealing with the near-infrared(IR)
> and mid-IR regions of electromagnetic radiation, in descriptions
> that suggest that no prior art or prior teaching exists."
>
> "The use of specific infrared absorptions for qualitative and
> quantitative analysis does not represent new intellectual property
> and should not now be (nor have been in the past 30 years) patentable".
>
> W.G. Farley, Chemistry Professor at Kansas State University and
> former editor of Applied Spectroscopy
> B.R. Kowalski, Chemistry Professor at University of Washington
> P.R. Griffiths, Chemistry Professor at Univesity of Idaho
> H.W. Siesler, Chemistry Professor at University of Essen
There will always be those who feel this way. Always. Riley likes to do
this as well (cite to lists of professors making general assertions and
immediately taking them as gospel, not to be overridden by mere logical
argument based on evidence). These are quotes of professors leaping to
legal conclusions. Again, their remarks are not addressed specifically to
the subject patent, nor are they addressed to the question of tissue
typing. While they give weight to the conclusion they state, they are
not by themselves dispositive.
> And I don't want to hear the standard lawyer's rebuttal to these postings,
> ".... why worry, most of these bad patents never get asserted and those that
> do the courts take care of." The integrity of a monopoly intellectual
> property rights system is undermined by these patents. Switch to a patent
> registration and fine, let such stuff issue.
Certainly, Greg is not suggesting we switch to a registration system.
What we have is clearly far better than mere registration, which would
exacerbate, not lessen the problem he is raising. And from my experience,
no amount of searching will reveal all of the art that is out there -- you
reach a point of diminishing returns very, very quickly. Who will pay for
these mega-searches Greg would have the PTO perform? Aren't patents
expensive enough to obtain?
As to the argument in the first sentence, I think that much would have
been resolved had the amalgam patent act passed, permitting those
professors to make the patentee go through the paces of distinguishing
prior art without resort to a trial in a meaningful review; that is, if
they really believed what they had written. The new reexamination rules
were designed PRECISELY to address the issue of the bad patents that
inevitably issue from a finite-resource patent search.
> Software, biotech, optics, telecommunications - increasingly the trade press
> is filled with complaints from scientists and engineers about the growing
> number of overly broad and unnovel patents, while the patent bar community
> continues to focus solely on the statutorial aspects. Each side sticks to
> what it is comfortable with, while the problem gets worse.
This is news? Non-patentees have ALWAYS whined about issued patents. The
argument is always much the same as greg's: a general discussion of the
patent, a note that the subject matter of the patent is generally the
subject matter of some specific research for which there is a lot, a lot
of published works; and a leaped-to conclusion that the prior art is read
upon by the claims of the patent (stated in the form "so it must be
invalid"). Sometimes they have been right, but this is becoming less and
less true. Patents are surviving elaborate scrutiny more and more each
day. This suggests that better examination and better analysis by the
courts has bolstered, not diminished the integrity of the patent system,
and not the reverse.
> APS belongs on the Internet. MEDLINE belongs on the Internet. NTIS belongs
> on the Internet. Anything less makes a mockery of technology transfer and
> novelty/obviousness analysis in light of jokes like the SPI.
Who shall pay for this? Why should it be given away? None of these
databases are inaccessible per se: its a question of market supply and
demand.
Finally, and in fairness to Greg, I want to thank him for raising the issue
in the manner he has accomplished here. In the past, I had YELLED at him for
merely quoting abstracts and leaping to conclusions. This particular posting
is far, far more detailed, containing a specific claim and citing particular
prior art. It is deserving of attention and, despite all of my arguments,
he may even be right on this issue at the end of the day (I certainly have
not read either the patent, cited prior art, Greg's prior art or the file
wrapper). My point is merely that these issues are very, very complex, and
that there are few slam-dunk issues. I think Greg's example is an excellent
illustration that even apparent slam-dunk invalidity issues can be much less
clear than they seem at first blush.
Thus, I honor Greg's effort to lay out a strong argument with sound facts,
and thank him for adding much light to the debate. I hope my response did
the same, and pointed out the complexity of these issues, even in the face
of what at first blush appeared to be very strong facts.
--
just another view,
Andy Greenberg (werdna@gate.net)
