CA2003535A1 - Method and apparatus for detecting and removing plaque from arteries by laser pulses - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
A system for distinguishing healthy tissue from diseased tissue for purposes of controlling a treatment laser to reopen a channel in blood vessels. The system uses predetermined reference data that the define the fluorescent optical characteristics of healthy tissue such as non atherosclerotic blood vessel wall. The tissue to be distinguished is excited to fluorescence by excitation light from a laser. The fluorescent light caused thereby is detected to determine its optical characteristics such as peak intensity, wavelength at the peak and shape of the spectrum. These optical characteristics are compared to the optical characteristics of known healthy tissue. If this comparison indicates that the tissue emitting the fluorescent light is probably diseased tissue such as plaque, one or more pulses of laser light is guided through a fiber optical waveguide to ablate the diseased tissue. Criteria for determining whether the tissue under treatment is normal may be derived from data generated from abnormal tissue samples. In a broader embodiment, the method includes steps for clinical treatment of a variety of medical problems, without restriction to a particular type, by implementing a problem solving technique derived from the particular laser angioplasty technique.

Description

: ::

METHOD AND APPARATUS FOR DETECTING AND REMOVING
PLAQUE PROM_ARTERIES BY LASER PULSES ~::
", 10 Field of the In entlon This invsntion pertains to the field of laser angioplasty, and, more particularly, to the field of systems for di tinguishing healthy tissue from plaque. ~-Specifically, the invention concerns algorithms which use optical characteristics of normal tissue to control probe :::~
and fire laser systems. More generally, the invention relates to a treatment and problem-solving method involv~
ing a hierarchy of iterative test steps.

Backqround of the Invention Angiogxaphy provides fiber optic visualization : :;
such that the position of a fiber optic probe for use in laser angioplasty or for other uses may be observed : :~.
directly by the surgeon or through the use of an optical vlewing device. This technique in~olves the use of radiopaque dyes and x-rays to visualize from x-ray at tenua~ion data the position of a blood vessel obstruction and assists a phyYician in guiding a fiber optic probe to `~
an occlusion. One probll.em with thiis method results from the use of a fluoroscope to assist the physician in guid~
ing the:optical fiber, which provides only a two~
: dimensional view of the fiber and the suspec~ed lesion or :~
other tissue. Thus, when the laser is fired, the physi- :~
cian cannot be certain in advance that the fiber is :: :'' ..

- : . ,:: . . ~ :

Z~ 535 properly aimed at the treatment site, even though this may ~ ~"
appear from the fluoroscope to be the case, thus leading to damage of healthy tissue adjacen~ the treatment area.
Using the method known as la~er angioplasty, laser pulses have been used to recanalize occluded blood vessel. (Abela, G.S. et al., Laser Anqioplasty ~ith Angioscopic ~uidance in Humans, J Am Coll Cardiol (1986) ` ;
8:184, No. 1 at pp. 184-192; Underhill, D.J. et al., Hiqh Resolution Anqioscopy: Feasibility, Limitation, and Desiqn Considerations For Laser Coronary Anqioplasty, Surg Forum (1985) 36:2g9).
Several methods for guiding medical laser systems have been described in the prior art. Methods presently in use do not, however, assure that healthy tis-sue will not be ablated by a rapidly firing high energy laser, since no method of distinguishing healthy tissue ~
from plaque is taught other than directly butting the end ` ;
of the fiber optic probe against the obstruction.
Furthermore, the physician generally has no means to make a decision adequately and with cer~ainty as to whether theprobe is pointed at plaque or healthy tissue. Since some types of plaque, such as fibrous or calcified plaque, require levels of laser energy to destroy them that are above the energy level that will destroy healthy tissue, any mistake in identifying appropriate tissue to fire the laser at can result in destruction of healthy tissue ~uch a~ a blood ve~sel wall. Perforation of the blood vessel ~ `~
wall iY noted in Underhill et al., supra, as a major clinical obstacle. -~
~ One problem with present laser angioplasty systems is that human reaction time is relatively slow, such that when the operator wishes to stop the firing of the laser, unwanted damage is caused by the continued fir~
ing of the laser even after the operator visually identi-: :~'".~'.",' :' ~:~ ' '`, ,"`'' . ~ :
'-" ~ -':`'~"'`"'' 20C~535 fies the need to stop. Thus, a reliable system for automatic laser angioplasty is needed.
Another basic problem with all laser angioplasty is how to make the distinction between healthy tissue and plaque or other unwanted obstruction before firing the laser. One approach which has been tried in the prior art to make this distinction is to utilize differences in light reflectivity, i.e., selective light absorption, between healthy tissue and plaque. Preferential absorp-tion of laser light by atherosclerotic plaque comparedwith healthy tissue at selected wavelengths assures more exten ive ablation of plaque than healthy tissue by providing better ability to make the decision when and if to fire the laser. For instance, tetracycline has been 1~ applied to atherosclerotic plaque to enhance it, for greater reliability in the destruction of atheroma by ultraviole~ laser radiation. See Murphy-Chutorian, Douglas, et al., Selective Absorption of Ultraviolet Laser Energy by Human Atherosclerotic Plaque Treated with Tetra~
cycline, Am J Cardiol 1985; 55: 1293-1297, which is in-corporated herein by reference. See also Prince, M.R. et al., Preferential Liaht Absorption in Atheromas In Vitro: - -Implication For Laser AnqioPlastY~ J Clin Invest ~1986) -~
78:295, in which it is taught that atheromas from cadavers absorb more laser light than normal aorta tissue between 420 and 530 nm. This absorbance was attributable to yel-low chromophores consisting primarily of a mix of carotenoids that are known constituents of atheromatous lesions. The reference concludes that preferential 30 ~absorption of blue laser light by carotenoids in atheromas may permit selective ablation of atheromatous obstruc- ; `;~
tions. The reference suggests that selective absorption of some wavel~ngths of laser light by plaque may be used for distin~uishing healthy tissue from unwanted obstruc-35 tions to circumvent the problem of vessel perforations, ;
". . ..
'' ', . '''.' '."'`,`:

. .: .,,,: :', ' :

_4~ 3~
',"" ~' . .
aneurisms and other forms of inadvertent laser damage to ~ -normal tissue lying adjacent to or under atheromas. The reference also suggests that increased absorption proper-ties of atheromas may be obtained by administration of chromophores such as beta carotene orally. However, this selective absorption technique doe~ not prevent ablation of healthy tissue with sufficiently high accuracy for widespread commercial use. Further, not all plaques have the same light absorption spectra, and there is some clinical evidence to suggest that the spectrum of light absorption of plaque shifts between the in vivo and in vitro states. Thus, a system tuned for one type of plaque characterized by in vitro studies may not be accurate in making fire decisions for different kinds of plaque or the same type of plaque in an in vivo setting.
Other references teach the administration of chromophores to the patient before treatment to enhance the fluorescent intensity of unwanted tissue to be ablated. For instance, two related patents owned by the `
assignee of the present invention, U.S. patent 4,641,650 and its continuation in part, U.S. patent 4,682,594, teach a probe-and-fire laser system for angioplasty, optionally using an administered atheroma-enhancing agent (such as tetracycline), which is preferentially absorbed by plaque and which causes diseased tissue to have distinguishing properties (namely, a characteristic fluorescent spectrum) to aLd the system in distinguishing normal tissue from diseased tissue for purposes of making treatment laser firing decisions. These two patents provide for automatically monitoringiand controlling the output of the laser and for terminating its operation before there is a destruction of healthy tissue around the treatment site.
Similarly, U.S. Patent No. 4,336,809 and international Patent No. PCT/US84/00840 teach that the visualization of a tumor can be enhanced with hematoporphyrin dye which - : ~

.: .. ,: : :

Z0~S35 fluoresce~ when illuminated with a particular wavelength of light.
Several investigators have noted differences in the fluorescence intensities of atherosclerotic plaque and healthy tissue (Lu, D. et al., Atherosclerotic Plaque Identification Usin~ Surface Fluorescence, Clin Res (1986);
34:630A (Abstract) (wide band xenon source excitation of cadaver thoracic atherosclerotic aortas at UV 340-380 nm, blue 450-490 nm and green 530 560 nm with fluorescent intensity scanned from 350 to 670 nm showing blue fluorescence having a peak at 540 nm and a weak shoulder at 580 nm with fluorescence intensity noted to be greater in normal tissue regions than in diseased regions); Leon, M. et al., Human Arterial Surface Fluorescence:
Atherosclerotic Plaque Identification and Effects of laser Atheroma Ablation, J Am Plaque Coll Card (1988) in press;
Sartori, M et al., Estimation of Arterial Wall thickness , --~
, :: i: : ~
and Detection of Atherosis by Laser Induced Arqon ` -~
. : : :
Fluorescence, Circ (1986) 76:IV-408). At least one ~ ~
.: . , 20 researcher has noted the difference in shape between `~
fluorescence spectra of normal tissue versus atherosclerotic tissue. (Kittrell et al., Diaqnosis of Fibrou~ Arterial Atherosclerosis Usinq Fluorescence, Ap~
pl~ed Optics, Vol. 24, No. 15, at pp. 2280-2281 1198S)).
Thresholds for tissue spectra are discussed as ratios or products of ratios published in the Abela and Vnderhill articles r~ferenced above and in the article by Deckelbaum et al., Clinical Res, Vol. 34:292A (1986). The tissue spectra on which the~e analyses are based are dissimilar from those obtained!in;vivo.
Some al~orithms which distinguish plaque from healthy tissue use ratios of intensities of fluorescent ;
emission spectra excited between 450 nm and 490 nm (Kittrell, C. et al., ~ ; Sartori, M. et al., Autofluorescence Maps of Human Arteries, J Am Coll Cardiol .: ,:::', ~ ::", ::, . ....
:' :'",, ': ':
- :.
~: , ~ 6-20~ 3S
(1986) 7:207A; Sartori, M. et al., supra; Casale, P.N. et al., Improved Criteria for Detecting Atherosclerotic Plaque by Fluorescence SpectroscopY, Circ (1987) 76:IV-524) (suggesting excitation at 337 nm with ratio taken at 460/385 nm as the best two wavelengths to use);
Deckelbaum, L.I. et al., Discrimination of Normal and Atherosclerotic Aorta By Laser Induced Fluorescence, Clin Res (1986), Vol. 34, No. 2 at 292A). A difficulty with ratioing intensities of normal tissue spectra to unwanted substance spectra at a few wavelengths is that the optical characteristics of unwanted substances can vary significantly from substance to substance and between in vitro and in vivo states. That is, different plaques have different fluorescence spectrum shapes. Algorithms based on a few ratios are not able to differentiate healthy tis-sue from all unwanted substances that may be found at a diseased site or in an arterial obstruction with suf-ficient accuracy to risk possibly damaging healthy tissue.
Thus, although fluorescence intensities can assist in the -guidance of a laser system and in the diagnosis of the tissue, they lack the necessary consistency and accuracy to be used alone for treatment laser firing decisions. ` -Other references which teach the use of laser systems within the body cavity do not distinguish between 25 healthy tissue and unwanted substances before firing the ;
treatment laser. European Patent 86302603.5 teaches the administration before treatment of a chromophore which is ~ ~
preferentially absorbed by plaque. This chromophore ~` `
absorbs laser light. Therefore, laser energy ablates ~ s -~-plaque more readily thanihealthy tissue. U.S. Patent Nos.
3,858,577 and 4,273,109 are fiber optic systems which ; ~` -deliver laser light to an internal diseased site. ;~-~ ```
Other references teach the use of laser systems which operate outside of the body cavity. U.S. Patent No. ~; u-4,438,765 teaches the use of a medical device which fires :' : ., ..,. ,, "", ,, 2(~S3S ;

a laser to fuse the retina. This device senses eye motion and only fires the laser when the eye is not moving. U.S.
Parent No. 4,316,467 teaches the uqe of a laser system which removes piqmented tissue from skin. The laser is fired only when a photodetector in the system senses the characteristic color of pigmented tissue.
U.S. patent 4,718,417 describes a method of diagnosis of the type of tissue in an artery using laser light at about 480 nm. Excitation of in vitro tissue produces spectra which have peaks at 550 nm and 600 nm, and a valley at 580 nm. These features result from absorption by hemoglobin in the tissue. Recent collabora- ;
tive studies carried out by MCM Laboratories and the National Institutes of Health show that hemoglobin absorp-tion does not occur in in vivo tissue or in vitro tissue which is more than several hours old. Therefore, the method described in U.S. patent 4,718,417 will discriminate normal tissue from a~heromatous ti~sue in in vitro samples more than several hours old, but will not work in living patients.
Although there is a substantial amount of exist~
ing technology ~or laser surgery, a method is needed which distinguishes healthy from substantially all unwanted ;~
substance~ which may be found at a particular site inside the body cavity without the necessity of administering a chromophore before treatment and with a sufficiently high degree of accuracy to minimize the risk of damage to healthy tissue. None of the technology described above is ;
capable of satisfying all of these criteria.
SummarY of th_ Invention `
It is accordingly a general object of the inven- ~ ;
tion to provide an improved method of distinguishing unwanted substances from healthy tissue within the body . . ' ' ~ . ~ .; "

20~535 :

cavity so that laser energy will be absorbed only in the unwanted substance.
It is a more specific ohject of the invention to provide a methcd of distinguishing healthy tissue from unwanted substances without requiring administration of chromophores before treatment. In addition, due to the possible occurrence of several types of unwanted substances at a ~ite, a specific ob~ect of the invention is to distinguish substantially all unwanted substances -lO whose optical signal differs from that of healthy tissue.
It is another particular object of this inven-tion to provide a method of controlling the firing of a laser for treatment of diseased areas wherein optical ~-characteristics of diseased areas are distinguished from 15 those of healthy areas, such as by distinguishing the ;~
respective fluorescent spectra or atherosclerotic plaque from the spectra of healthy arterial walls.
It is a specific objective of the invention that a computer-implemented process of distinguishing healthy from diseased tissue be used wherein the details of the form of the algorithm shall be chosen so as to give the ~ `~
best performance in distinguishing unwanted substances from healthy tissues.
It is a specific objective of the invention that 25 the calculations made by the algorithm be completed in a ;~
period of time which is shorter than the period of time in which the distal end of the fiber moves to a different position.
It is another object of the invention to describq the process o~clontrolling laser firing to safely clear occluded arteries by comparing incoming optical ~ -signals from said arteries to standard reference signals and generating a fire/no fire signal as appropriate to -said ob~ect.

: - 9 -2~ ;3S
:: .
In order to implement the invention, there is contemplated electrooptical means such that incoming opti~
cal signals (i.e., electromagnetic radiation, e.g., light to said means are converted into an outgoing electrical or -optical signal which can trigger a treatment means (e.g., a laser) to treat or avoid treating the area of interest.
It is a general feature of the invention that the incoming ;
optical signals are complex and multiform and that the electrooptical means converts said signals into simpler representations which effects appropriate and safe treat-ment of the diseased areas.
It is a further ob~ect to describe a method of determining controI signals by first determining either -standard normal curves or a range of values of specific -intensities (relative or absolute) for a specific wavelength or range of wavelsngths of electromagnetic radiation (e.g.~ light) from nondiseased substances likely to be encountered by a fiber cable means (e.g., blood, ~ ;
airl healthy internal blood vessel surfaces and vessel ;;
walls) and similarly for diseased substances ~e.g., thrombus, atheroma, etc.) by obtaining adequate numbers of ~;
specimens of each substance and then calculating threshold levels which serve as criteria to identify or distinguish these nondiseased substances and/or these diseased 25 substances and/or combinations of subs~ances from each ;
other.
It is a further object that said threshold levels are written into the control system and electrooptically connected to incoming light from said 30` substances either inherent ! to or induced form said substances (e.g., by photoexcitation with a light source to cause tissue fluorescence) in order to react ap-propriately to ~rigger or inhibit an electrical and/or optical signal to a treatment means intended to remove , :
,. , 20~
diseased substances (e.g., laser light to remove atherosclerotic plaque)~
According to the teachings of the invention a trea~ment laser and a diagnostic laser have their light outputs coupled to a fiber optic probe catheter (hereafter the catheter). A computer with associated interface circuitry controls the diagnostic laser to cause it to send one or more pulses of excitation light to the site in front of the current position of the catheter. The fluorescence spectra emitted from the site is then compared to a composite normal tissue fluorescence spectra `~
in several respects in order to make a fire decision for control of the treatment laser.
~ Three types of information are extracted by the computer in the analysis of returned light and in the comparison to the reference normal tissue spectrum. These three tests are: a comparison of the relative shapes of the return light spectrum and the reference normal tissue ;~
spectrum; a comparison of the relative wavelength loca-tions of the peak in the return light spectrum and the peak of the reference normal tissue spectrum; and the ~ `
relative intensity of the fluorescent return light versus ;~
the intensity o~ the reference normal tissue spectrum. `~
All three tests must indicate the catheter is pointed at pla~ue before the treatment laser will be fired.
The algorithm uses a signal processing technique which considers the optical signal of normal tissue as a :
reference. It is recognized that there are several mathematical forms for signal processing techniques which compare S re~ere~ce!signal with a signal obtained by a sensor system. The mathematical form chosen is that found to give best performance from analysis of optical signal data.
In order to implement the method of invention it is necessary to determine the composite healthy tissue .

-11- Z~s35 ., .
reference spectrum from signals at many healthy tissue sites from many different donors. The composite normal tissue fluorescent spectrum is mathematically derived by obtaining the individual fluorescent spectrum of healthy -tissue from multiple donors. Each spectrum is then normalized and shifted so that the intensity peaks of each individual spectrum occur at the same wavelength and has the same intensity. The wa~elengths of the peaks for each normal tissue spectrum are found by fitting a parabola to peak of the normal tissue fluorescent spectrum from each source. The intensity at that peak wavelength is then normalized to equal the in~ensity at the peak of the composite healthy tissue spectrum. The individual spectra are then aligned for best fit with the composite by a shifting process that mathematically amounts to finding the smallest least square~ residual for a series of dif-ferent wavelength shifts. The shift that results in the ;~;
best fit for a particular spectrum is then used to shift all intensity points on that spectrum to fit the 20 composite. The intensity at each wavelength for all the ~ ~ -shifted curves is then averaged to determine the shape of the composite normal tissue spectrum. ;
During this process of mathematically deriving the composite healthy tissue reference spectrum, the range of peak intensity wavelengths for normal tissue is recorded. In other words the maximum wavelength of the peak and the minimum wavelength of the peak of any of the normal tissue spectra is recorded for purposes of perform-ing th6 peak position test mentioned above.
; After the composite healthy tissue spectrum is mathematically computed, the relative intensity test thresholds must be determined before treatment can begin.
This test provides information to distinguish plaque from healthy tissue based upon the fact that normal tissue has 35 a higher fluorescent intenslty by a factor of two than `
. ~'.

.. ' . ;.
.: - ~ ... ...

~' "".. ','' , ''. ~, ~ 3~3 most kinds of plaque. This test cannot be used alone however because some types of plaque such as calcified plaque fluoresces as brightly as normal tissue. To determine these thresholds, the fiber is butted against known plaque and an excitation light pulse is emitted by the diagnostic laser. A lower limit intensity threshold is then set to prevent the treatment laser from being fired at noise. The higher limit intensity threshold is then set by recording a value approximately two times as -10 intense as the fluorescent return from the plaque, since it is known that normal tissue usually fluoresces more than two times as brightly as plaque. The actual upper limit is se~ according to actual readings of intensity Yalues from normal and diseased tissue taken during - ~ ;
analysis of diagnostic data. Accuracy can be improved if the diagnostic data for relative intensity and from which ;; ; -`
thè normal tissue reference ~pectrum is computed comes from ths very patient to be operated upon. -~
Finally, a curve shape threshold is determined ;~
by which the shapes of the return light spectrum and the healthy tissue reference spectrum may be compared. ThiS
curve shape threshold is determined by finding the great~
est least squares residual from all the individual healthy tissue reference spectra used to generate the composite and using this largest residual as an indicator of the worst fit between a normal tissue spectrum and the composite. This largest residual is used as the curve shape threshold to make sure that normal tissue is not ;~
mistaken for plaque.
i The data and ~hresholds so collected are now ready for use by the computer in controlling the treatment laser. In the preferred embodiment, 50 cycles of il-luminating the opera~ion situs with pulses from the diagnostic excitation laser and reading the fluorescent return light and making a fire decision are performed .,. .':
-~

~ 13~
535 ; - ~ :
before the fiber is moved and the system is read~usted.
After each excitation pulse, the return light is analyzed ;~
by the three tests mentioned above. First, the intsnsity of the return light is compared to the maximum and minimum S intensity thresholds. If the intensity is within the range defined by these two threshold, the first test is passed indicating that the fiber may be pointed at plaque.
Next, the wavelength of the peak intensity of the return ~ ~ ;
light i8 determined and a determination is made a~ to~ ~ ;
whether this wavelength is within or outside the range of wavelengths of peaks for normal tissue. If it is outside the range of wavelengths in which the peaks of normal tis~
sue spectra were found, then the second test is passed and the probability that the fiber is pointed at plaque rises.
Finally, the shape of the return spectrum is compared to the shape of the reference healthy tissue spectrum. This ~;
is done by normalizing the return spectrum and computing the fit to the reference spectrum using the least squares residual technique. The maximum least squares residual is ~0 then compared to the curve shape threshold. If the fit is wor~e than the fit of any of the individual healthy tissue spectra to the composite reference spectrum computed from these individual healthy tissue spectra, then the conclu-sion is drawn that the fiber is pointed at plaque, and the computer commands the treatment laser to fire a pulse through the fiber. If any of the tests indicates that the fiber may not be pointed at plaque, then the treatment laser is not fired.
It is also important to obtain signals from ab~
normal substances. IThe~deviation of individual healthy tissue signals from the reference signal are used to select the threshold for firing the treatment laser. The devia~ion of individual signals from unwanted signals are used to select de~ails of the mathematical form of the ~ -~
35 algorithm. ~ `
'..~."'.,' :"',:., ;,~ ~". ,` '':,' ' '"': 'i ~'`'''`'''"'' ''''`'`' ' : .

~0~3S
The method of the inven~ion in a broader ap-plication comprises steps for clinical diagnosis and treatment of a variety of types of medical or other tasks ~ , where action must be taken on the basis of data qenerated ;
from the patient or object to be acted upon. This method includes the development of hardware for the task in ques~
tion; the collection of diagnostic data; the selection of techniques to solve the problem in question; the develop- - -ment of algorithms to implement the chosen techniques; the ~ ~
10 testing of the algorithms and revision thereof if neces- , " ~-sary; and compensation for changes in hardware -characteristics and tissue types.

Brief Description of the Drawings Figure 1 is a block diagram of the apparatus in ;
which the teachings of the invention are employed.
Figure lA is a diagram of a linear array of pixels generated by the apparatus of Figure 1. ;~
Figure 2 is an illustration of the composite reference spectrum generation process.
Figure 3 is a flow chart,of the process of generating the composite reference spectrum.
Figure 4 is a flow chart of the process of mak- ~ `
ing the fire decision for the treatment laser given the 25 characteristics of the return fluorescent light. -~ ~
Fig. 5 is a plot showing a typical optical ` `
signal of healthy tissue and an optical signal of healthy tissue with hemoglobin absorption on the left side of the ;-curve.
Figure,6,is a graph of a,curve fit between the composite reference spectrum and a normalized fluorescent return light spectrum.
~ Figure 7 is a~flow chart of the preferred embodiment of the treatment process according to the teachingF o~ the invention.

'".,,,.,:-.",.:

Z~ 53~

Figure 8 is a flow chart of an alternative ; ;
embocliment of the method of the invention, incorporating a ;~
series of hierarchical, iterative, automatic decision- ;
making steps.
Figure ~ is a flow chart of a generalized problem-solving method based upon the method of Figure 8.
Figure 10 shows a formula utilized by the method of Figure 8. ;
Figure 11 is a block diagram showing an alterna-tive ~o the embodiment of Figure 1.
Figure 12 is a block diagram showing an alterna-tive to the embodiment of Figure 1.
, .. .
Detai ed DescriPtion of the Preferred Embodiments ~ Figure 1 is a block diagram of a typical system in which the teachings of the invention may be employed.
The description immediately below will be with reference -to Figures 1-7, following which is a detailed description of a preferred embodiment of the invention relating to the flow charts of Figures ~ and 9.
The e~act details of the optics shown in Figure 1 are not critical to the invention, and any conventional optics to selectively couple the outputs of two lasers to a single fiber and to coupled light returning from the 25 patient through the fiber to a spectrometer will suffice ; ;
for purposes of practicing the invention. The excitation - ;
laser 11 has a shutter (not shown) which is controlled by `
a computer 13 via a control bus 15. The laser 11 may be, ~ -for ins~ance, the HeCd laser made by Omnichrome of Chino, `
30 l California~. The co!ntrql bus 15 carries signals to control -~
the shutter so that ~xcitation light from the laser 11 is allowed to be coupled into the fiber only during the period prior to firing a treatment laser 17 when the fir~
ing decision is being made by the computer 13. The treat~

,, ,:`' ',..;~': ' ~ -16- ;~
2~ 5~S
ment laser may be the flash lamp pumped dye laser avail-able from Candela of Wayland, Massachusetts.
The treatment laser 17 includes a shutter or some other mechanism that can be controlled by the -5 computer 13 via a control bus 19 so that a treatment pulse ~ `~
to destroy obstruction material may be generated under the control of the computer.
The excitation laser 11 and the treatment laser 17 are coupled through a beam splitter 21, a lens 23, and a holed mirror 25 into a fiber optical waveguide which has been threaded by angiography to the position of an obstruction. As will become clear from the discussion below, the excitation laser is pul~ed by the computer 13 to cause a pulse of excitation light at 325 nm to il-luminate the tissue in front of the fiber inside the bloodvessel being operated upon. The material in front of the fiber 27 emits fluorescent light which is guided back by the fiber, to then exit the fiber and impinge upon the holed mirror 25. This light is reflected by the holed mirxor through a lens 29 to a spectrometer 31 which includes a diffraction grating. This diffraction grating spreads the various wavelengths of fluorescent light out at different angles to impinge upon various pixels of a linear photosensitive array in a detector 33, such as the array of pixels 1-512 shown in Figure lA. The spectrometer 31 and detector may be such as those made by `
Princeton Applied Research (EG~G) of Princeton, New i~
Jersey. The detector 33 detects the intensity of all light in a p.redetermined frequency band, which in the `
30 ; prefe~rejd embodiment islthe band f~rom 375 to 650 nm. Each ;~
of the pixel~ in the detector 33 may be individually read by the computer 13 through conventional interface circuitry. An operator interfaces with the computer 13 through a terminal 3S, through which commands regarding the number of cycles between readjustments, the minimum ::;, .".,;,:

:' ~ ,' '. ' ` :,'.',", Vj ~ 2~S35 ::
,,, ~, . . ...
~" . ~ :
and maximum thresholds of intensity, and the treatment laser firing rate in Hz may be issued to the computer 13. ;
Although the description below of the embodiments of the invention will be by reference to 5 return light which is fluoresced by the treatment tissue ;~
of the patient, such return light may also be generated by reflectance of the irradiating laser beams. Thus, the method should be understood also to encompass the alterna-tive embodiments wherein such reflected return light is used, in which case the details of analysis of the return spectra will need to be adjusted.
The method of the invention will be as described hereafter is implemented with the apparatus shown in Figure 1. However, alternative apparatus may also be used to implement the method, such as that shown in Figure 11, which shows an alternative to the embodiment of Figure 1 wherein the trea~ment laser 17 also acts as a diagnostic or excitation laser, thus eliminating the second laser 11 which was used in the configuration of Figure 1. This also eliminates the need for the beam splitter 21. In this embodiment, the laser 17 is energized, and sends a beam or pulse to the treatment site of the pa~ient, thereby ablating a portion of tissue. The tissue ~ , `~;;
fluore~ces or reflects due to the treatment light, and the ` `
25 return light is then received by the detector and analyzed ~- in the same manner as if two separate lasers 11 and 17 - `
were used. The analysis of the return light is then used ~ -to decide whether to fire the laser again at the site. "~
For a description of a system utilizing a single-laser ;~
30i apparatus, see Lau~qr, Guenther, et al., Excimer La~er~
Induced Simultaneous Ablation and Spectral Identification of Normal and A~herosclerotic Arterial Tissue Layers, ~ ~ -Circulation 1988; 78: 1031-1039, which is incorporated ~--herein by reference.
; ~ `

:.' ""~ ' ::'' ~ ` : :.: :`': ":;

:: ........
2~53~
The operation of the above system will be ~ -described utilizing the example of distinguishing atheromatous plaque from healthy tissue by their fluorescent spectra from excitation at 325 nm and emission between 375 nm and 650 nm. Before the system can operate ;
to make distinctions based upon return fluorescent light, two things must be establi hed: first, a healthy-tissue, normal reference spectrum; and secondly, certain thresholds for use in comparing the spectrum of return fluorescent light to the reference healthy tissue spectrum. To do this it is necessary to obtain diagnostic opticaI signals from a representative sample of normal tissue and plaque.
In order to obtain such signals, the operative end of the optical fiber may be placed on various arterial specimens such as intima, media, atheroma, graft occlu-sions, thrombus, adventitia and blood, as well as in air. ~ -This diagnostic data is collected before any operation is performed by illuminating healthy tissue from multiple ;
human sources with excitation light from the diagnostic laser and coIlecting data regarding the intensity of the return fluorescent light at each of a plurality of -~
wavelengths. Further, diagnostic plaque fluorescent intensity data is also obtained by illuminating known ~ - ;
~5 atherosclerotic tissue, i.e., known plaqua, with axcita~
tion light at the same wavelength as will be used in the actual operation. Usually~ this process of collecting diagnostic data is done on both healthy and atherosclerotic tissue taken from in vivo measurements, 30 but improvements in accuràcy;can be obtained by taking ;~ ` ;
these readings on the actual patient to be operated upon.
The diagnostic plaque fluorescent intensity data from cadaver sources is ~upplemented during the operation by absolute plaque fluorescent intensity data from the patient being operated upon for purposes of adjusting the ' ,,`';`''`, : .

~Oll ~ii3~; .
maximum intensity threshold value to be described below.
The purpose of collection of the diagnostic data and the processing thereof to be described below is to provide healthy tissue reference values. These values are used to compare the actual return light from the situs of the operation for purposes of deciding whether the fiber is pointed at healthy tissue or plaque. The manner in which these decisions are made will be described in more detail below.
Either a continuous wave or pulse of light may be used ~at, e.g., 325 nm or 337 nm) from a laser, and transmit~ed through an optical fiber to a light detecting ~
system, which may comprise the spectrometer 31 in conjunc- ~ -tion with the detector 32, discussed above. Light intensities (e.g. photon counts) of the returning light may be divided into discrete wavelength ranges (e.g. every ~-1 nm), and a spectral curve may be generated for each specimen tested, i.e. intima, media, etc. The spectra are stored for later comparison with spectra generated at a ~ - `
20 later time from in vivo or othex ~pecimens, in a manner to ~ ~
be described in detail below. ~etails of generating such ~;
spectra are discussed below. `-Thus, the shape of the fluorescent intensity ;~
emi~sion spectrum versus wavelength of normal tissue is 2S preferably generated by constructing a composite of many ; ;
normal spectra. The composite spectrum is found by super~
impo~ing multiple normal tissue spectra and then averaging them. Thi~ process is illustrated in Figure 2 and the `~
method is represented by the flow chart of Figure 3.
30 Figure 2 shows thejproaess of normalizing and shifting onej ;
normal tissue fluorescent spectra to a reference spectrum, ;~
which may itself be a composite of several or many spectra. -; In one embodiment of the invention, a composite spectrum is generated for plaque or other abnormal tissue ~: '..:, : ,~ . ::~ ,:.:.::
. ,,, ~ .: ;., :. : ~ ,.. :. ::

~" 20 ~

in the same manner as the normal-ti~sue composite spectrum is generated. In this embodiment, the test tissue spectrum is compared to both the normal and abnormal composite spectra (in a manner to be described below relative to the normal tissue spectrum), thus increasing the reliability of the abnormal tissue identification.
Thus, Figure 3 is a flow chart showing a method ' of the invention for calculating the composite heàlthy tissue reference spec~rum and setting the various thresholds. The first step in this procedure of generat-ing a composite is to first generate a reference spectrum for healthy tissue and fit the spectrum to a parabola.
The peak may be normalized by dividing all the value~ of this first reference spectrum by the peak value, thereby preserving the shape of the spectrum while forcing the peak value to equal 1.
Another spectrum is then generated for the next -sample of healthy tissue, and this is also~ fitted to a - ` `
parabola, as represented by step 42 of Figure 3. The - ;~
20 curve fitting process is carried out by a least squares `~
analysis or other standard curve-fitting method, and identifies in rel.iable fa~hion the wavelength at which the peak~Lntensity of the curve 4Q occurs. This~process of curve fitting to find the peak intensity is represented by ~;
25 step~42 of Figure 3. The process of fitting a parabola to :
the spectrum 40 finds the wavelength of peak intensity more;accurately than searching for the peak intensity on the cilrve, since noise spikes on the curve could lead to a ;~
false answer in the latter case by erroneously indicating 30 i a highest value which iisinoise and not part of the acltual fluorescence spectrum.
This next spectrum is also normalized so that -:
the peak value is 1, as with the firs~ (reference) spectrum of the unknown test tissue of a patient at i~s 35~ peak so that it will have the same intensity at its peak 2C~ 3~;
as that of the partially complete composite spectrum to which it will be added. Oth~r methods of normalization may be used, so long as ~he same normalization method is used for all the spectra. In this embodiment, the posi- -tions of the peaks are of greater importance than the peaks~ absolute magnitudes.
In Figure 2, the spectrum 40 is thus a normal tissue fluorescent spectrum taken by exciting healthy tis- - ;
sue from some human source such as a cadaver. However, in vivo tissue is preferably used to generate the normal composite spectrum, because in vitro tissue becomes piqmented very quickly ~fter death, which skews the ;i: ;
spectra taken from it. If relatively old tissue is used (such as more than a few hours), then compensation must be made for the spectral skewing, which may be difficult to implement and lead to unreliable results. Indeed, a large ~ -increase in accuracy in makinq firing decisions for the treatment laser can be obtained if the healthy tissue ;; ; ~`
spectra from which the composite is generated are taken 20 from the patient to be operated upon prior to the -~
operation.
The process of normaliæing the curve 40 is repre~ented by steps 44 and 48 in Figure 3. The first step in doing this is to note the intensity of the curve 25 40 at it peak as symbolized by step 44. In the case of - ;` -spectrum 4n,: the peak intensity Il at wavelength lambda is noted. The peak intensity IREF for the first reference spectrum 46 (which for later steps will be a composite of numerous spectra) is then noted, an~ all intensity values on the spectrum 40 ,arejmul~iplied by thq fraction IREF/I
to normalize the curve 40 as symbolized by step 48 in Figure 3. This results in the curve 52 in Figure 2.
Note that the ratio used in step 48 uses the ~ -~
term I . This term refers to the intensity at the peak of the particular healthy tissue spectrum being operated upon : ." ,.""~..",,":",. .

: ' . ~' ." ~ :' : .

r~ . --2 2-- 20~535 ;

at the time since the same step 48 will be used to operate ;~
upon all the healthy tissue spectra that go into making up the composite reference spectrum 46. The same process of curve fitting for pea~ determination and normalization is carried out on each healthy tissue spectrum. The first healthy tissue spectrum that is operated upon becomes the initial composit~ reference spectrum. Thereafter, subsequent healthy tissue spectra are normalized and fit-ted to this first healthy tissue spectrum until all the healthy tissue spectra have been so processed.
The individual healthy tissue spectrum is then - ;
aligned with the composite spectrum by finding the least squares residual of the individual spectrum and the composite spectrum, by use of the formula shown in Figure 2. This process essentially amounts to shifting the spectrum 52 by several discrete steps ; of wavelength and - ;
- then, for each j, taking the sum of the squares of the ;
diffarences between ~he curve 52 as shifted and the ` - ;~
composite waveform for all wavelengths i (wavelengths will be referred to as either i or lambda interchangeably).
This~process determines the best fit by finding the ;
where the sum of the squares of the differences is at a mi~imum. In the presently-described embodiment,~ the proc~
e9S i9 carried out with the equation~

(1) R(~)= sum for all i of (I(j+i)-C(i))2 ' ;' , ;
where~
C(i) = the composite reference healthy tissue 30~ ~function comprising a spectrum of~;fluorescent in~tensity values at each of a plurality of wavelengths i (i = Iambda in Figure 2);
+i) = the individual healthy tissue spectrum component being averaged into the composi'~e reference : ~ - , " ~

~, , ,: ~, ~, : :: :,, ~ :.::

~ . :, :": ~ i:

2~ 535 : :
: ;
, ,.~ , :.
spectrum expressed in terms of a series of fluorescent intensity values at a plurality of wavelengths (j~
i = any of a series of offsat wavelengths, each of which results in a least squares residual number R(j) 5 when i is cycled through all its possible values and the ~ -expression of Equation 1 is evaluated for all i and that ; -;
particular j; ~ ' i i = the wavelength lambda at which to evaluate C(i) and I(~+i) with the current j to determine the ~ ,~
intensity values.
Thus, C(i) represents the intensity of the composite spectrum at pixel i. Pixel i refers to the intensity at a particular wavelength by virtue of the operation of the diffraction grating described above in 15 spreading out the spectrum so that different wavelengths -~
fall on different pixels of the spectrometer. I(i+j) represents the intensity of the individual spectrum at ~ ;
pixel i+j, where ~ is a parameter which introduces a shift ~
of I(i~. The value of ; is varied to find a minimum value ~ ~ `
o the least squares residual, R(;). The superposition of the individual spectrum and the composite (reference) spectrum i9 optimized at the minimum value of R(j). The value~ of the first reference spectrum and those of the next-generated (and now standardized, i.e. normalized and shifted) spectrum are then summed for each pixel.
As indicated by step 54A in Figure 3, the above ~rocedure is carried out on numerous healthy tissue spectra, each time utilizing the first reference spectrum as the standardizing spectrum, and each time summing the ; ,~
3~ succeeding standardlzed spectra with the sum of the fore~
going spectra. Finally, the sum of all the spectra is .,.
divided by the number of spectra added, thus generating a `~
composite spectrum. - . ~ ~
This~process may be repeated as often as ; ~-desired~, as indicated in step 54C, each time using the , - ~:

: .

~ -24~ S3S

previously-generated composite spectrum as the reference spectrum for the next cycle. The use of step 54C lessens any weight which may be given to the resulting composite spectrum by the use of the first reference spectrum since, ~
5 as described above, the first reference spectrum is used ~ ;
as ~tandard with respect to which the remaining spectra ;
are shifted by the least squares method. Thus, repeating~ ~;
the above steps, each time using the newly-generated ~ ' composite spectrum as the reference spectrum for the next ;~
sequence (using the same spectra data again in the subsequent sequence of steps) minimizes the artificial weighting effect. ~;
Step 50 shows a step relating to determina~ion ~ ~
of a range of peak positions for healthy tissue. In this ;;
step, the wavelength of the peak of each healthy tissue spectrum generated, such as spectrum 40, is compared to ~he range of wavelengths in which healthy tissue peaks oc~
curred for purposes of updating the range, if necessary.
The purpose of this is to determine the range of ;
wavelengths in which peaks for healthy tissue spectrum occurred for purposes of comparing against the wavelength ;
of the peak of the return light from the situs of the operation when actual patient tissue is treated, to aid in making the determination of whether the fiber is pointed at healthy tissue or diseased tissue. This step may be carried out after the composite spectrum is generated, as indicated in Figure 3, or it may be carried out as the compo~ite spectrum is being generated, in which case it would come between steps 48 and 54. In the latter case, 30 ~ the first two healthy~ti$sue spectra that are analyzed initially set the bounds of the healthy peak range. ~ ~ `
Thereafter, the wavelength of each healthy ~issue peak is compared to th~ current maximum wavelength and minimum ~wavelength range limits. If the wavelength is less than the minimum wavelength in ~he current range, the range :;: , . ."'' ' ' ' ~'''''``'`'`

."''` ~

~ -25- 20~53S ~ :

limit on the minimum wavelength end is updated with the new, lower wavelength. Similar updating of the maximum range limit occurs if the wavelength of the peak exceeds ; -~
the current range limit on the maximum end.
Thresholds for various tests are found after ;
normal tissue is charac~erized by generation of the composite healthy tissue reference spectrum. As noted earlier, one of the types of tests to be performed in mak- ~`
ing the firing decision is a determination as to whether the intensity of the return light from the tissue at which the fiber is pointed is within a certain intensity range.
Since it is known that normal tissue intensity for return light is generally at least twice as high as return light from m~ny types of plaque, the range of intensity values 15 to which the return light is compared is set so that if ~
the intensity of return light is in this range, the fiber ; ~;
is probably pointed at plaque. This test cannot be used ;~
alone, however~, since some plaques such as calcified plaque is almost as bright as normal tissue. The range of intensity values is set with a minimum intensity threshold and a~maximum intensity threshold. The minimum intensity -;
threshold is the minimum intensity required to give adequate signal-to-noise ratio. This threshold is used to ensure that the treatment laser is not fired when noise is ;
causing the apparent return light. This process is symbolized by step 56 in Figure 3. The computer retains the intensity data from the healthy tissue spectra used to compute the composite reference spectrum. This intensity data is examined to set the minimum intensity threshold.
30~ " The ma~imum ~ntensity threshold is determined from relative intensity ratios of plaque and hea~lthy tis~
sue found during analysis of diagnostic data and the absolute intensity of plaque measured during~surgery.
This process is symbolized by step 58 in Figure 2. The absolute intensity data is collected by the compu~er and , :': ': ;
: ':'~ ~ ';
~ , ~: :.
: ' ~ ` ~ ~: .;

:: . :.:: ~:
. -. ~ -: : ~ , .

::
~ -26-;35 : ~ ~
used to readjust the maximum intensity threshold, if `~ ~;
necessary, during the interval between cycle groups. The ~ -normal mode of operation of the instrument is in cycle groups of 50 cycles, wherein each cycle is characterized by illumination of the tissue in front of the fiber by a pulse from the excitation laser followed by analysis of the return light and a decision to fire or not to fire the treatment laser. After 50 such cycles, the sy tem may be readjusted, including adjustment of the maximum intensity 10 threshold, and 50 more cycles are run. ~;
The curve position thresholds are used to set a - ;~
range of a maximum wavelength and a minimum wavelength. ;~
These thresholds define a wavelength range within which all the peaks fell of the healthy tissue spectra used to generate the composite fall. This is determined by the upper and lower values of wavelengths of the peaks of the ~ ;
healthy tissue fluorescent intensity emission spectrum versus wavelangth.
The curve shape ~hreshold is determined by the greatest least squares residual R(~), computed from a representative sample of normal tissue spectra and the composite healthy tissue reference spectrum. The greatest residual is used as a threshold so the normal tissue will not be mistaken for plaque. That is, the greatest minimum R(~) of all the minimum R(j)'s computed during the process of computing the composite is stored as the curve shape ~ ~ ;
threshold. This threshold will later be compared to the minimum R(~) computed for the return light as compared to ;~
the composite as part of the analysis of the return light ;~
in making the f!ire decision for the treatmelnt laser.
In some embodiments, an additional step 59 is performed to determine a ~heme stain" threshold. The ef- ~5 fects of heme stain on normal tissue spectra is explained in more detail below, but suffice it to say that the pres~
35 ence of heme stain can lead to mistaking normal heme stain ;~
,-. ~ .,.:, .,;

s~

27 ~ ~
5i3~

tissue for plaque. The determination of the heme stain threshold step 59 symbolizes the process of determining the intensity ratio between absorption at 425 nm and the intensity at the peak for both heme-stained normal tissue and non-heme-stained normal ti~sue. These two ratios are then compared by subtraction or division or some other mathematical relationship to determine the heme stain threshold. The exact mathematical manner of comparison is not important as long as the same mathematical method of 10 comparison is used during actual treatment to compare ;
similar ratios to those de~cribed above for purposes of comparing to the heme ~hreshold. More detail on the calculation of intensity ratios during treatment is given -below in conn~ction with discussion of Figures 5-7.
Following the collection of diagnostic data, optical signaIs of normal tissue are analyzed by consider- ;
ing their intensities, the fluorescent intensity emission spectrum versus wavelength locations, and curve shapes.
Aspects of these features used to characterize normal tis- ;~
sue will be discussed, and then the way they are used by the system of the invention will be described.
Diagnostic data from clinical studies show that healthy tissue does not have a characteristic absolute intensity which can be used to control a laser system because some plaques are almost as bright as healthy tis-sue while others are much darker. However, the emission intensity of normal tissue is more than twice as great as that of mo~t unwanted substances found in the vicinity of normal tissue sites which have been probed.
~; 1 Diagnostic data ~rom clinical studies also show that for normal tis ue, the range over which the wavelen~th positions of the peaks of fluorescent intensity emission spectra occur is less than that of unwanted substancec in arteries. As noted above, the positions of these peaks is obtained by fitting a parabola to the emis~

.: " ~
:. ~.'."~, ~

-28- 5 ~

,:, ,; :
sion curve. This is a more reliable method of determining the peak than the maximum intensity. The fluorescent intensity emission spectrum versus wavelength has a ~ :;
relatively flat top so a small noise spike in this region 5 could give a maximum intensity that is not representative -~
of the curve's peak position, if parabola fitting were not used.
Figure 4 shows the method of the invention for probe-and-fire laser treatment after the thresholds have .10 been found. Step 60 represents the precess of causing the excitation laser to emit a pulse of excitation light at 325 nm and reading the fluorescent intensity spectrum of the tissue at the distal end of the fiber by reading the output of all pixels of the detector in Figure 1 to obtain the intensity of the return light at each wavelength.
This intensity data is stored in the CPU's associated -memory for analysis by subsequent steps.
In step 62, the peak intensity of the return light is found by any reliable method, and the intensity and wavelength at the peak are stored for future use. In the preferred embodiment, the peak ~s found by curve fit~
ting a parabola to the return light spectrum. ;;-~ ~;
Next, the intensity at the peak o~ the return Iight is compared with the minimum intensity threshold in 25 step 64. If the peak intensity is less than the minimum ;~
threchold, control returns to step 60, and the first cycle is over. If the peak intensity is greater than the minimum threshold, proce~sing continues to other steps ;
since ~he intensity indicates that an adequate signal to 30 'noise`ratio `exists. Step;66 represen~s thq process of comparing the peak intensity of the return light to the maximum intensity threshold, the threshold having been determined by the pre-operation analysis of diagnostic ~ ~;
data. During calibration, about 50 spectra are preferably 35 taken for plaque from the patient being operated upon. ~ ~j ', " " ~ .''~".'.,; ';"' ;~ - - 2~ S~S

If the peak intensity of the return light is greater than the maximum intensity threshold, the fiber may be pointed at healthy tissue, so no firing of the : -::.. ;~
treatment laser will be allowed, and this cycle i5 s terminated by returning processing to step 60. If the peak intensity of the return light is less than the maximum threshold, it is more likely that the fiber is pointed at plaque or some unwanted obstruction such as a ~.
blood clot. Step 66, the maximum intensity test, tests if ~
10 the intensity of the return light is significantly higher . .:
than the plaque absolute intensity measured during surgery :~
prior to the probe-and-fire laser treatment. If the ';.
intensity has increased significantly so that it exceeds the maximum intensity threshold, the laser will not be ;~
15 fired and another spectrum will be obtained for analysis :~
by the sensing system as symbolized by step 60. In the event the peak intensity of the return light is less than the maximum intensity threshold, the presence of plaque is - :: :
indicated, and processing proceeds to the next test ~ :~
symbolized by step 68.
Step 68 is a peak position test for determining whether the peak of the spectrum of the return light (i.e. ~ ~ :
the fluorescent light from the test tissue) is within the :
range of wavelengths encompassed by the peaks of healthy 25 ~tissue spectrum used to calculate the composite reference qpectra. If it is not in this range, the material at the ~: :
distal end of the fiber is an unwanted substance and the laser is fired as symbolized by the path 70. If the wavelength of the peak of ~he return light is within the 30 .~range of kealthy tissue peaks, healthy tissue may be in front of the fiber so a spectrum shape tes't must be ! ~ "
performed to further reduce the odds of firing the treat~
ment laser~at healthy tissue~ The details given relative :-to step 123 of Figure 8 may be used also for the peak position test of step 68.

. ~,: :~,. :.:

..: ., ~ .
'''' ''..'; ' -30- Z~S3S ~: ~

Step 72 represents the tissue index test for the ;~
test tissue, and is given in detail relative to step 130 below. In this test, the return light spectrum obtained by the sensor is normalized and i5 compared with the composite reference spectrum. The tissue index technique ussd to construct the composite reference normal spectrum is also used to determine the deviation of the return light spectrum from the composite reference spectrum. The tissue index for the test tissue is compared to the tissue index threshold determined from the normal composite spectrum. The tissue index threshold is generally given by the largest least squares residual of the composite and the healthy tissue spectra used to compute the composite.
This amounts to a determination of whether the best fit of 15 the test tissue spectrum to the composite is worse than ;~
the best fit of all of the known healthy tissue spectra ;
used to compute the composite reference spectrum.
If the tissue index for the test tissue is equal to or less than tissue index threshold, then no firinq of the treatment laser is done during this cycle, and processing returns to step~60 via path 74. If the fit is worse than the worst fit of any normal tissue spectrum (i.e., the test tissue index is higher)l then the presence ~ ; -of p}aque or other unwanted obstruction in front of the 25 treatment laser is indicated, and the treatment laser is `~
fired as indicated by path 76 to the firing step 78. `;~
After firing, processing returns to step 60 via ~ .
path 74 through steps 80 and 82. Step 80 is to determine if 50 cycles (or any other number of desired cycles) have -~ `
been performed.` !' In the' préferred embodiment, the operat~
ing physician is provided with a foot switch, and the computer operator has a keyboard for controlling the system ~not separately shown). The laser is interlocked with the foot switch such that it will not fire if the ~ ~ 35 physlcian takes his foot off the switch, and in addition : '. ,: ~ .::,`
" , ,i . ",. ',, ~": ,: ,.. ", ::

..
-31~ 3S

the laser will stop firing if the computer operator hits a ~ -predetermined key on the keyboard. If not, step 82 incre- -ments the cycle counter and step 60 i9 performed again.
If the requisite number of cycles have been performed, 5 step 84 is performed to stop the cycling process and ~
prompt the operator for any needed ad~ustments. ' A difficulty with thi~ technique is that a small fraction of normal tissue exhibits hemoglobin absorption as shown by the reduction in intensity on the left side of the peak in Fig. 5. The method of the invention therefore tests for this reduction in intensity by using a ratio test. If this heme dip shown in Figure S is found to oc~
cur, the least squares residual analysis is only carried out on the right side of the curve.
It is difficult to compare the algorithm discrimination results of one diagnostic study with another because of dif ferences which include 1) choices of excitation light sources, 2) methods of emission detec-tion, and 3) tissue siteq and tissue types. The data set of spectra used to construct the algorithm described here is much more extensive in size than any of those used in the prior art discussed. In addition, prior art considered only in vitro data, and no prior art known to ~ ;;
applicants teaches or considers the use of a laser angioplasty system to collect data in vivo for use in mak~
ing firing decisions. Much of the data used to develop the method described herein was obtained in vivo, and some data was obtained under actual clinical conditions. The shapes of in vivo plaque spectra tend to more closely resemble ~the shap,e of n,oXmal tissue spectra than the shape Qf in vitro plaque spectra. Also, spectra obtained during laser angioplasty surgery show peak shifts not observed in -`
in vitro data. In addition, spectra can vary from optical ~system to optical system. Thus, results obtained with an actual laser~angioplasty system which has an optical fiber ;~

" .-~ ."..:,, ::

2~535 ~:

and an optical multiplexer are more valid for determining the capability of a method to distinguiih tissue types. ~
Due to the difficulty of comparing results for ;
different diagnostic studies, and due to the greater ~ `
validity of the data used to determine the method described herein, only this more valid data set will be considered in discussing discrimination capability.
Alternative methods of discrimination considered in making the following comparison include line widths, areas, peak positions, intensities, ratios, centroids, correlation functions and least square residuals (i.e. tissue indices). These algorithms have been tested against the entire data set. Results show that absolute intensities and one or two ratios can give no better than 50~ correct identification of abnormal tissue, leading to erroneous identification af abnormal tissue as normal 50% of the time (although they ma~ properly identify normal tissue as such close to 100% of the time). Shape tests using the least s;quare residual identify abnormal tissue 66% of the 20 time (misidentifying abnormal tissue as normal the ather `~
34~, although again, normàl tissue may be correctly identified 100% of the time) (Leon, M.B. et al., Circ (1987) 76:IV-408; Leon, M.B. et al., J Am Co}l Card, in ;~
press)~ This discrimination percentage for abnormal tis~
sue is improved to 85~ when peak positions are used as a ;~ criterion. Analysis indicates that if a composite of a few normal spectra of a particular patient to be operated upon is used ini~tead of a composite of normal spectra from many patients, discrimination of tissue types for 30 i identifying abno~mal tissue can be increased to~over 90 For instance, in actual tests which have been conducted using the method of the inve~ntion, the peak position test and tissue index test used in conjun~ion have been found to result in 91% accuracy in identifying noncailcified 35 plaque, while maintaining 100% accuracy in identifying ; :. .. ,,'i i, . , .:, . ..:: :: .
" ' ~' ~; '..i, 'i ~
:: .: .;. .: ;;:
' ' ~i: . . ,"i 'i , :`,`,i. "
.': .' ''.'' ',:`,,'.`
:: , ~ ;::.:::. .
.. ,..:: ., ii 2~ 35 ::
...... ,. ,.. `, .. ::,:
-.......
noxmal tissue. Bartorelli, Al, In Vivo Coronary Plaque Recoqnition by Laser-Induced Fluorescence S~ectroscopY, Supplement II Circulation, Vol. 78, No. 4, October 1988, which is attached hereto as Appendix A (and in which the term ~specificity~' refers to percentage of successful identifications of normal tissue, and ~sensitivity" refers to percentage of successful identifications of noncalcified plaque).
In summary, the process according to the teach-ings of the invention takes advantage of the fact that the fluorescent spectra of normal tissue of all patients are similar, while the fluorescent spectra of plaques differ even for plaques in the same category. The teaching of the invention is to make discriminations based upon this fact by determining when the spectra of plaque differs by more than a specified amount from a composite normal tis-sue spectrum. The normal tissue reference spectrum used according to the teachings of the invention is derived from 75 in vivo spectra obtained from approximately 25 patients.
There is one additional complication in the discrimination picture. Normal tissue occasionally -~
exhibits absorption by hemoglobin at approximately 425 nm or on the left side of the peak of the normal tissue composite spectrum 46 of Figure 2. The spectrum which results by "heme stain" is shown at 90 in Figure 5, and a non-heme-stained spectrum is shown at 92 of that figure.
This phenomenon occurs about 5% of the time, so the normal 30 tissue composite spectrum has the shape shown at 92 in `~
Figure 5. Since the process according to the teachings of `~
the invention compares the spectra of return light to the ~-composite normal tissue spectra, heme-st~ined normal tis~
sue might accidently be mistaken as plaque and fired upon by the treatment laser because of the differences in shape of the heme-stained normal tissue spectrum. To prevent -~ ~

..............

:~ ..: :: :
: :: .: ..

-34~
~0~535 this from happening, in the preferred embodiment of the ;
invention, an additional test for absorption at 425 nm is performed. ;~
This test can be best understood by referring to ;
. .
Figure 6, which shows the composite normal tissue spectrum at 46 and a typical return light spectrum at 48. The heme stain absorption test is simply a comparison of two intensity ratios. The first of these two ratios i5 the ratio of the intensity of the composite normal tissue spectrum at wavelength A, 425 nm, divided by the intensity of the composite normal tissue spectrum at wavelength C. ~ ; ~
The inverse ratio could also be used. The second ra~io is ; ~ - ;
the intensity of the return light spectrum at wa~elength A '"''~'~.. "","'"'~,'',""4., divided by the intensity of the return light spectrum at -; ;
wavelength D, which is the peak of the return light. The inverse ratio could also be used. If the difference ; between these two ratios is greater than a predetermined ~ -~
heme threshold set in advance~by analysis of diagnostic data regarding spectra of heme-stained and non-heme~
20 stained normal tissue, then plaque is indicated, and the;~
treatment laser may be fired.
Figure 7 shows a method of controlling the treatment and excitation lasers incorporating this ad-ditional heme stain absorption teist. The method of Figure 7 is s~milar to the method shown on Figure 4, except that an additional test step 84 is present, and Step 72 has been split up into two steps, 72 and 72A. Step 84 i~
calculates the ratios defined above and compares them ~o the heme stain threshold. If the heme stain threshold is ;
30 i exceeded, thenlthere is~ noiheme stain, and the program branches to Step 72, whereby the entire treatment tissue spectrum is used to compute the tissue index. If the heme ;~
stain threshold is not exceeded, then there is heme stain, ; ~ ~
and the program branches to Step 72A, whereby only the ~; ;
35 right side of the spectrum is used to compute the tissue ;~
,, : . .. ;, .:.
.: .: :
: , .
:. ', ;:.:
,','", "'.,' '~

:

~ _35_ 535 :

index. Both Steps 72 and 72A test branch to Step 80 if the tissue index for the return light spectrum is found to be within the predetermined range, and branch to Step 78 if not.
By placing the distal end of a fiber on tissue and analyzing the return spectrum of the tissue, the systems shown in Figs. 1 and 11 can determine the kind of tissue which would be ablated if the treatment laser were to fire. However, these systems utilize a very small field of view, so the location of the fiber in the artery is not well known, and the distribution of unwanted tissiue is not well known. The addition of a technique having a wider field of view would assist in positioning the fiber to increase patient safety and to increase the proportion ~ ~;
15 of unwanted tissue which could be ablated. -~
Wider field of view techniques than sensing fluorescent spectra can be added as shown in Fig. 12.
These methods could be usied as adjuncts to the fluorescent-tissue type discrimination, or they could be 20 used alone without fluorescence. The preferred method is ~ ~ -to use a wider field of view technique in conjunction with a ~luore~cence technique, because a wider field of view technique detects the tissue type at the distal end of the treatment fiber with lower certainty than fluorescence ;~
25 sensing ~ ~ -Wider field of view techniques may include ultrasound, magnetic resonance, angioscopy, or other proc~
esses. In the casie of ultrasound, device 12 in Figure 12 is an ultrasound transducing crystal, and is positioned in 30 an appropriate~housing~ In the case of magnetic '! l'~' ' ~ ~',.'!' ,.,"'`,."~"
resonance, device 12 is a magnetic resonance transducer.
In the case of angioscopy, device 12 represents a lens and an optical transducer.
Item 16 in Figure 12 represen~s an appropriate ~ -35 wire or cabla, which carries electronic signals to and ~

, ~,.

~ -36-Z0~53'~

from supply 18 in the case of ultrasound or magnetic resonance. In the case of angioscopy, item 16 represents an optical fiber bundle which carries the optical signal for an image from the body, and also carries light ener~y from supply 18 to the body. Item 16 generally will run parallel to the optical fiber 27, at least in the region close to the patient. In some usages, items 16 and fiber 27 will be in a catheter in the body.
Electronics unit 18 includes an electronic power supply for the device, including items 16 and 12 in the case of ultrasound and magne~ic resonance. Unit 18 may ;
include the light source in the case of angioscopy. Unit ;~;~
18 may also include necessary image processinq electron~
ics. The power supply, light source and the image -processing apparatus may actually be located in different physical units.
Item 8 is a screen showing the image returned by the wide field of view device. This image can be used by a physician to position fiber 27, so that its distal end is placed against unwanted tissue. The image on screen 8 will show where remaining unwanted tissue is located in the ar~ery, and it may be able to show the physician the type of unwanted tissue which remains in the artery.
Item 9 is an electrical communication wire which links the wide field of view image processing electronics unit 18 and the system control computer 13. Item 9 is optional when using a wide field of ~iew subsystem. The ; -wide field of view image processing electronics 18 is preferably linked to the system control computer 13, so 30 ~ that the program may utilize information from the wide field of view image. An example of such a usage would be ~;
that an image is obtained before a treatment laser ~;~
sequence after the fiber 27 and items 16 and 12 are cor-rectly positioned for treatmen~. The program would use a :~
mathematical technique such as a correlation method to '.~ ' ~37~ ~ ' 3 ' make sure fiber 27 and items 16 and 12 remain correctly positioned during a treatment laser sequence. If the position change~ by more than a predetermined amount, the treatment laser is prevented from firing. Thus, the embodiment of Figure 12 improves the accuracy of the abla-~ion method.
Following is a description of an alternative, preferred embodiment of the invention, wherein numerous factors, including those described above such as heme lO s~ain~ spectra ratios, and the like, are taken into ac- ;~
count in an iterative analysis which greatly reduces the likelihood of firing the laser onto normal or healthy tis~
sue. This method is depicted in the flow charts of Figures 8, and is utilized in software which i~ stored in 15 the computer 13 shown in Figure 1. `
The following description should bs understood on two levels: first, as an alternative embodiment to the above described method of the invention, involving ad~
ditional steps and techniques, as in Figure 8; and 20 secondly, as a generalized problem-solving method which in ~ `~
the present implementation is used for the specific puxpose of laser abla~ion of abnormal tissue, but whose implications and app}ications are in practice much ;~
broadert as in Figure 9.
The method implemented by the software depends on a hierarchy of factors. In general, if there is a change in a factor at a particular level, factors at lower levels are affected. The factors in this hierarchy are: ~ ;
1) purpose of the control system; 2) criteria for choosing ~ ~ -~ the system hardware; 31 basic components of the hardware;
4) collection of data for algorithm development; 5) analysis of data to determine the best recognition technique and thresholds; and 6) detailed determination of the decision making flow. Factors 4 through 6 are carried 35 ou~ in an iterative fashion. As with the embodiment `~

' ~ 38 discussed above, data is first collected from sample normal tissues--preferably from several hundred samples for a good statistical sampling--and the data generated is later used in a variety of tests relative to data col- ;~
lected in vivo, for determining whether the laser should ~ -be fired in the carrying out of angioplasty.
More generally, as mentioned above, the present invention also teaches a method for problem-solving, based upon the application of a series of tests to be performed, with the application of certain tests dependent upon the ~ -~
outcome of earlier tests. This method has numerous ap- ;
plications, including but not limited to use in the medical areas discussed herein.

1. Purpose of the control systems The purpose of the control system in the present embodiment is, as discussed, to distinguish between the signals of healthy and diseased tissue so that a treatment laser will fire only at diseased tissue, and further to distinguish between signals of diseased material and other material which should not cause the laser to fire, such as -blood and air. In the present inventi.on, the control system includes the apparatus depicted in Figure l~
including the software used in the computer. The control system also has features which increase patient safety and ~-protect components of the system. Examples of increasing patient safety are closing a shutter in front of the ~ ;
ultraviolet diagnostic laser when patient exposure is not ;;
30 necessary, and lowering the treatment laser energy level ~ ~;
when soft diseased material is detected. An example of protecting system :, :

, -`.", ' ' '~
, ,.,: :, - . -: ', '~ ':: . ., :, ' :"

-39- z~ 35 - ~ ~

hardware is closing a shutter in front of the detector ~ -~
before firing the treatment la~er.

2. Criteria for choosinq hardware:
There are several criteria for choosing hardware which affect the diagnostic signal in the present inven~
tion. The hardware must support a technique which ef~
fectively discriminates between healthy and diseased tis~
sue. The technology on which the hardware is based ~ust be capable of operating effectively in a clinical setting.
These features include speed, reliability and size. The method used must be safe for the patient and the clinical operators of the system.

3. Basic Hardware~
The basic hardware in the present system are those components shown in Figure l.
In the case of the present system the excitation ;~
source is a laser which emits ultraviolet light. A
continuous wave HeCd laser emitting at 325 nm may be used, but~experiments show that a pulsed nitrogen laser emitting at 337 nm gives very similar results. The treatment source is~a pulsed dye laser which is operated at 485 nm.
The computer preferably comprises an optical multichannel ~- ; `
analyzer, such a3 the OMA III model 1460 made by EG&G
Princeton Applied Research of Princeton, New Jersey. The diagnostic laser light and the collec~ed tissue -~
~fluorescent light are transmitted by the same optical components including an optical fiber or fiber bundle, such aslbundlel32 shown, iinlFigure 1. The sensor ! or detec~
tion system is a spectrometer coupled with a detector hav~
ing~an intensified photodiode array. Aspects of the hardware which may affect the implementation of the prcsent method are discussed below.
.. :

'`' - `'.:; :~

~ ~40~
;~OC~;~535 ~ ;:
The carrying out of particular steps in the method change due to changes in the hardware which cause the diagnostic signal to change, or they may change due to 5 changes in the hardware which alter aspects of the system ' -' ' which can be controlled by the computer.
Features of the diagnostic source which affect ~'~ ''- '~' ' the diagnostic signal include the type of signal such as ~' ' ;
electromagnetic waves, ultrasound or magnetic resonance.
The wavelength of the source may affect the signal. For example, tissue exposed to an ultraviolet laser source' '~"
yields a signal which differs from that of the tissue '; '~
exposed to white light. The diagnostic signal produced by ~ ''';'' a pulsed source may depend on the pulse width and ' '' 15 frequency in some cases. Also, the source intensity can'~;' affect the signal. Thus, the preferred embodiment of the ~; ' ;'~
system controls the pulse rate, the period of exposure,~ ' ''~; ;
and~the intensity. If there are two diagnostic sources,''-' the capability of varying the wavelength by alternating between the two should be provided, such as between laser light and white light. In addition, the capability of ~' varying between two different diagnostic systems, such as ; ;'-' laser light and ultrasound, may be provided. Other '' ~' diagnostic media besides light and ultrasound may be used.
The method of the invention preferably also controls the ablation depth of the treatment laser. Fac- ' tors which may affect the ablation depth include the wavelength, pulse width, pulse rate and energy per pulse. " - '~
The pul~e rate and energy are variable, but should be 30 great enough that progress in a clinical case is adequate. ;-~
However, the depth ablated should not exceed the tissue ' depth'at'which fluorescënce o'riginates. When possible', ablation of diseased tissue should be adequate and abla~

'..''''"''"'"'';"' ~,",.

i :', '', ' ' ~,: .: '::' -. .: .` ' .

~ "'' '" .~ .` "'';': ;' : ' ~; ~. ( .:

; -41- Z~ 3 ', "~
tion of healthy tissue should be poor. The pulse width and wavelength may be varied as necessary.
The capability of the computer affects the method of the invention. The greater the speed of the computer, of course, the more complex the method which can be accommodated by the software stored therein, especially when utilizing a real-time in vivo ablation implementation of the invention. This is beacuse in such a setting, the computer's decision of whether or not the treatment laser is to be fired must be made in as short a period as pos~
sible following the collection of the diagnostic spectrum, so that the fiber tip does not move significantly, which ~-could cause the firing of the laser upon healthy tissue. -~
Other computer factors affecting the complexity of the method which may be implemented include the memory size, number and type of input/output ports, and the capability of synchronizing timing. Currently the computer controls a pulse which fires the treatment laser and it has output control of shutters both in the treat~
ment laser cavity and in front of the diagnostic laser and spectrometer. An input signal from a foot switch can start and terminate the sequence which determines if the treatment laser fires.
The collection optics can affect the spectral resolution and the spectral transfer function or wavelength region over which the system is sensitive. The efficiency of the optical system affects the intensity of the detected signal. Shapes of tissue spectra obtained with this new optical system differ from spectra obtained 30~ previous;ly, because~an,a~ditional wavelength region contributes to the spectrum.
; Changes in the sensor or detsctor system can also cau e spectra or images to change. In the case of ~`~`;-images~these changes include the spacial resolution, the wavelength region over which the system is sensitive, and "~
. .....
` ,',`.,`,`,.

3S ~: ~

the contrast. In th0 case of spectra these changes include the spectral resolution and the wavelength region over which the system is sensitive. The sensitivity of the detection affects the intensities measured. The 5 present system uses a spectrometer whose grating is blazed ;
for optimum operation at 300 nm. The detector has a ~ -wavelength resolution of two pixels per nm and a sensitiv- -;
ity of about 0.2 counts per photon.
;: :. :" ~
10 4. Data collection: ~ ;
Generally data collection begins with spectra of in vitro tissue. We have found that the age of the tissue can have 2 significant effect on the shape of the fluorescent spectrum. Spectra taken from tissue harvested over a day after death can look very different from these of fresh tissue. Spectra of fresh tissue looks similar to spectra of tissue obtained ln vivo. This result is important because articles in this field which statisti-cally test methods for tissue type recognition (Anderson P.S. et al, Lasers in Med. Sci., V2:261 1987, and Casala et al, Circulation Abstracts, V76SuppiV:524 1987) base their analyses on spectra which look like spectra of old tissue and not spectra o~ in vivo tissue. Ti~sue types are easier to discriminate in older tissue, but results of methods tested with these spectra are misleading when considering the capability of a given method for diqcriminat~ng between tissue types in clinical in vivo treatment cases. ~~`~
In vitro analyses are, however, useful for 30 determining the ablation;ldepth of a laser pulse. As ~ ;
explained above, the ablation depth must be known for the ~
method to choose the optimum pulse energies and rates. ` ~-In vivo diagnostic spectra are generally ~ ;
acquired after preliminary in vitro studies. Unlike 35 actual clinical cases where the treatment laser is used, ~

: ' ' ,,' '..~ "' ;''''''~`';.' ,,' ~' ' :' ;~'' . `, ~, ' .

~43~ 2 ~ ~ S3 S ~ ~
~ -" . ~., the tissue type of spectra taken during a diagnostic case can be identified with relatively high accuracy. These spectra are generally ob~ained during open heart surgery and balloon angioplasty cases. Although physicians cannot actually see the tissue in a balloon angioplasty case, the location of a fi~er indicated by angiography is often of whether the distal end of the fiber is against plaque or ~ ' normal tissue. Comparison of angioplasty spectra and ~ :
surgrary spectra indicate that iden~ification based upon angiography alone is relatively accurate. Therefore, these in vlvo diagnostic spectra are used to develop the method. ~ ;;
In vivo spectra are also obtained during `
clinical cases. There are materials inside arterial ~ -obstructions whose spectra appear different from those obtained during in vivo diagnostic cases. For this reason a method cannot be based on spectra found in diagnos~ic cases alone. An example is thrombus, whose spectra are frequently encountered in clinical treatment cases but rarely in diagnostic cases.
All of the spectra for a given case are col~
lected at one time, so that changes in spectra--preferably ~`
~due to factors ~uch as blood in tha field and the varying distance between the fiber tip and tissue caused by tissue ablation--can be examined.

5. Analysis of Spectral Data Spectral features can be identified by their ;
shape, wavelength position, variability and intensity.
30 iThere`are a varietyloflsi~mple mathematicalltechniques~
which describe these features. Shapes can be described by ratios, slopes, widths, areas (of normalized curves), and minimum least square analyses compared with a reference `~;
shape. Curve positions can be d~scribed by their curve fitted peaks, true peaks, and centxoids, as mentioned in `''' "' .~`~'.`'",'`'`' '`'' ' ~ '''~'`.,.'`,'''`'''.',';, -44~
2~ 3S ` ~ ~
the above discussion of the embodiment of Figure 5. ~ ~;
Variability can be described by chanqes in area and the `
intensity at particular points.
A trial and error approach is used to determine `
5 the best method for distinguishing between the spectra of ~;
, .
different tissue types. The spectra of different tissue types are put into arrays. There are arrays which include ``
all healthy tissue, all diseased tissue, particular types ~ `
of diseased tissue and normal and diseased tissue with ~10 blood in the field. The analytic msthods are put in a second ~ype of array. The analytic methods are applied to the spectra in the tissue arrays. The result of these calculations is distributions of values for healthy and unhealthy material for each technique. There are several criteria for choosing the best discrimination technique.
The treatment laser must not fire at healthy tissue and healthy tissue with blood. The best analytical techniques for discriminating tissue types are those which yield the lowest percentage of values for diseased-tissue spectra in 20 the range of values found for spectra of healthy tissue. -~
Another desirable criterion for selecting method techniques is tha~ not only should the distributions of ;
the spectra of unhealthy and healthy tissue have minimum ` ``
overlap, but the separation between values for each tissue type should be large. Since the distributions for healthy and unhealthy tissue values generally overlap, the bound-ary where they overlap is the threshold where the laser ` - ` ;
will and will not fire.
In some cases the capability to distinguish -~
betweenitissue ~yp~s may~be improved by combining method techniques. For example, consider a particular type of diseased tissue whose spectra have widths which are `~
greater than the widths of all normal tissue spectra fifty percent of the time. Suppose that sixty percent of ~he 3s fifty percent of diseased tissue spectra which are in the : , . -, ,:~

, : ., : .: ~:, : , ~ 45~
2~ i53~ i~

range of normal tissue have peak positions which are outside of all normal tissue. Eighty percent of the diseased tissue can be distinguished from the healthy tis-sue with a method which instructs the laser to fire at all 5 tissue whose spectra either have widths greater than all normal tissue spectra widths or have peak positions which are outside the range of all normal tissue spectra peaks.
Different tissue types will have different spectra. Therefore, different recognition criteria are 10 needed. For example, fibrous plaque spectra have peak~ at wavelengths shorter than those of normal tissue and thrombus spectra have peaks at wavelengths longer than those of normal tissue.

15 6. Method Davelopment:
Figure 9 is a flowchart showing the generalized method of the invention, and is discussed in detail fol-lowing the discussion of Figure 8. Generally, changes in the method are initiated by the development of new 20 hardware, the occurrence of tissue types not currently considered, and the development of new method techniques.
If changes in the hardware affect the diagnostic spectrum then new data must be acquired unless the effect is small or unless a transfer function can be generated which ac-25 curately predicts how the spectrum will be changed. Other changes in the hardware affect the features of the system :
which ~he method can control. An example of a hardware `;
change which changes the spectrum is one in which an improvement ~n the optics which increases the wavelength 30` bandwidth of the,sys em has requi$ed changes in the method because the shapes of the fluorescent spectra emitted by tissue appear to be different. An example of a hardware change in which the method can control a feature of a component which could not be controlled previously is the ~ ;~
35 treatment laser power. Thrombus is an unwanted material " : : :.

~ ~'"'''` ' ,, ~,, ,~ ~ , , ,:,,, " ",,,., ,, . : .::, : -46-Z~S35 .:
, .. . ......
with a double peaked spectrum which can be ablated with the laser set at a low power. A thLn layer of blood between the fiber and tissue also produces a double peaked spectrum. Lowering the laser intensity permits the abla-tion of thrombus, but reduces the damage to tissue behinda thin blood layer.
Changes are initiated within the loop at Step 124, which considers if the method performs adequately during clinical cases. Many changes in the method have been based on results of clinical cases. The need to consider peak positions in addition to curve shapes was found from clinical treatment cases when a diseased material which has a spectrum shape similar to that of normal but has a peak shift was frequently detected.
Aueomatic grouping of pixels shown in Step 120 of the cur~
rent method box was introduced when too much time was spent~during clinical cases changing the grouping outside of the loop.
Referring now to Figure 8j a flow chart is shown 20 for a specific application of the invention relating to ;~
laser ablation of athero~clerotic lesions in arterial ves- ~ ;
sels. The method will be described in conjunction with the flow chart boxes by reference to the letters thereof. ~ ;

25~ Step 100: Information obtained prior to fiber insertion. - `~

~he in~ormation which is obtained in advance, as ~ -discusqed above relative to test cadavers or other~tissue, includes the range of peak positions of a parabola fitted ;; ~ `
30i to a sp~ctrum for normalland abnormal tissue, thus generating a threshold value for normal tissue without the ;~
presence of blood. Also,~ the positions of the spectrum used~to compute the parabola are given. A second threshold supplied àt~this time is the hemoglobin or blood stain ratio, as described in further detail relative to : . ::: ~ ~ ::
, , :, ,;:
;; : .: ~:
. ; :,:~:
.,~

:: : ;~:: ::

,", , : :

-47- ~
3~ :~

step 128 below. Positions of the spectrum where this~ ~ ;
ratio is computed are also supplied. This factor is described below in greater detail relative to step 126. A ;~
third threshold which is supplied is for the least squares deviation between data spectra and a spectrum referred to as the standard normal which is a composite of normal tis- -sue spectra obtained in vivo. Also supplied are the points on the spectrum used to compute the least square deviation. Default minimum and maximum intensities are also given. A detector background must be subtracted from every spectrum. The background is obtained at the begin-ning of a clinical treatment case.
In addition to the above information, preferably a relative pixel gain calibration value is also generated, in order to compensate for the fact that the response of each system to a s~able broadband source is different from the responses of other systems. Improvements in the system can reduce these differences, but there are limita~
tions to the improvements which can be made.
The present method is ba3ed on the shapes of spectra obtained with a single or a small number of ~ `
systems, i.e. the physical arrangement of apparatus such ;
as that shown in Figure 1. Other systems, and in fact other identical apparatus, will measure slightly different shapes when observing the same material or source, simply because of inherent differences between the instruments, even lnstruments made to the same specifications.
~herefore, it is preferable to calibrate all systems with -a standard source, such as a light source traceable to the 30i Naticinall Bureau ofjStandards. Such a lamp is a~ailable, for instance, from General Electric as calibrated by Optronics Laboratories of Orlando, Florida. By comparing ~ `
the white light spectrum measured by the diagnostic system used to acquire spectra on which the method is based, a 3S calibration file can be computed. This calibration file . ' :' ' ' :

:~ :, :
~:, 5~ ~5 is used to modify every spectrum measured by the treatment system ~o that the method treats the modified data as though it had been obtained by the original diagnostic system.
S The present method when used in other applica-tions may require different thresholds obtained from diagnostic testing. Thresholds for images would be based on values obtained from image recognLtion methods in which an image found during treatment is compared with a refer-ence image, which might be acquired ~ust prior to begin-ning a treatment cycle. Background and gain correction files can also be implemented in the system of the inven-tion. ~;
: ,'. ., ,,',.'.:', ',': ' Step 102: Phosphor reference test.

Prior to clinical treatment, diagnostic ~`~
tests may be run to test the system. At this time, the ~ ;
software and diagnostic system hardware are tested by comparing a signal obtained from a reference phosphor(obtained when the system wen~ through quality con~rol) with the signal measured from that same phosphor prior to - ~ `~
clinical treatment. The difference be~ween the data ~ ~ -spectrum and the reference ~ignal must not exceed a ~
25 threshold based on whether or n~t the tissue will be able - ~-to successfully identify tissue types.
Also at this time, the background signal due to the eloctronic and optical apparatus itself is subtracted.
In order to do this, the equipment is activated, and read~
ings are!taken,!but withlthe ~hutter 36 closed. Any data generated by the detector 33 under these circumstances must be due to background noise of the equipmen~, and the program stores the amount of apparent intensity for later subtraction from any counts that are generated from actual specimens or patients.
~' ',"',"~, ',';';
~- - ,, " ......

, . :: .., .: ":,.........

_ 4 9 _ 3~

Step 104: Insert fiber.

Once the phosphor reference test is completed, ~ ~;
S the optical fiber is inserted into the patient.

Step 106: Probe and display.

After the fiber has been inserted into the patient's body and the distal end of the fiber is placed on the area of diseased tissue the probe and display mode ;;~
i5 used. At this time in this mode the diagnostic source excites tissue fluorescence and method parameter values and the spectrum are printed on the computer screen ~t a `~
future date probe and display might also include images and other information.

: Step lOR: Information obtained with the fiber in situ. ~ ;

~ At this time intensities found from material in ;~
the arterial obstruction are used to determine the intensity range in which the laser will be allowed to ire~. Diseased material generally fluoresces more weakly than~normal tissue, so that if the intensity of material -.
is found to be low then the intensity range will have low ;~ ~ values,~with the result that diseassd material will be abIated without ablating normal tissue.
`~ Other factors used by the method in the closed loop part of the flow chart of Figure 8 (see the box fol-3o~ lowing Step~(ilO8) c~uldlalso lbe obtained while the fibjer isin situ. -Under some circumstances, it may be desirable to obtain~a spectrum of a patient~s normal tissue as a refer~
ence~corve for material which should not be ablated.
Referenc~e;images could be obtained to indicate the or;ientation of the fiber and position of the fiber ~ip. A

; :: : :,::

2~53S
-so-marker on the distal end of the fiber would be used to indicate the tip position. ;

Steps 110 through 136 comprise a l'probe and 5 fire" portion of the inventive method, in a closed loop ~
which may be carried out for a predetermined number of -times or until the operator causes an exit to the loop.

Step 110: Store data. '~' ; ' Ths program of the invention determines and stores how many times the laser fires in a closed loop cycle, and in the preferred embodiment stores method -parameters and complete spectra. Other data may, of lS course, be stored as desired.
~, . . .,, ,, :-.
Step 112: Diagnostic source(s) on. Adjust if necessary.

At this time the continuous wave ~eCd laser ~o which emits at 325 nm is used to excite a fluorescent s;ignal in tissue. Although the laser is on continuously, tissue is exposed to the HeCd beam only during the period while the detector is collectiny the fluorescent spectrum (which is then used by the method to determine if the treatment laser should be fired)~ The computer program controls a mechanical shutter 10 which is in front of the ;
~laser 11, which is opened when data collection begins and ~ r,' is clo~ed when data collection has been completed. The shutter 36 is closed when treatment is being conducted, -~
.
30 i and~is open~when!,diagnastic light L~ beingjused.~
Conversely, shutter 34 is closed when diagnostic light is being used (in the case of a two-laser embodiment), and is open when traatment light is used. `
`When a pulsed nitrogen laser is used, the method determines when a pulse is to be triggered. Two or more . :;. ::
::; ::

sources could be used in the same system. The method could open a shutter in front of an ultraviolet laser, obtain a spectrum, close the shutter and then open a shut-ter in front of a white light source and obtain a second spectrum. It could control an ultrasound or magnetic resonance device and obtain images. In the case of light -emitting devices the method might control the intensity with an electrical or an optical device.

Step 114: ~ead detector(s). Adjust if necessary.
.- . . ,. " ;,: ,', At this time an intensified photodiode detector wi~h 512 pixels is used. The fluorescent spectrum is measured between about 375 nm and 650 nm. The method can control the integration time for one spectrum. It controls another shutter 36 positioned in front of the `
detector 33,~as shown in Figure 1, and the shutter 36 is closed before the treatment laser is fired so that the -detector is not damaged. In an alternative embodiment, the dynamic range of the detector may al~o be controlled.

Step 116: Pre-in situ determined data corrections.
~ ::
At thi~ time, spectra are corrected by subtract~
ir.g a background. In the future gain calibrations as di cu~Yed in Step 100 will be used. Corrections for bad `~
pixels, bad detector areas or signal contamination by components in the system may have to be removed.
Step 118: Signal intensity adequacy determination.
At this time, adequate signal intensity is indicated by peak intensity. The maximum allowable intensity is the intensity at which the detector becomes saturated (typically indicated by a flattening out or 35 othex distortion of ths output spectrum), and the minimum ~

: :,. :~';:, -52- ~ 3~ ::

, . . .
allowable intensity is the intensity at which the signal- -to-noii3e ratio i3 not adequate to interpret spectra. ~;
These maximum and minimum intensities are the extreme values which can be used in an intensity test. The S criteria for determining an intensity range discussed in -~
Step 108 will generally produce a more restrictive intensity range than the criteria discussed in this sec- ~ , tion. Other signal quality features may be considered, such as contrast of images.
If the intensity is of the signal is in the cor-rect, predetermined range for analysis, the program proceeds to Step 124, which includes the test for whether the leas~ squares parabola fit is within the desired - -~;
range. If the signal intensity is not within the desired ;~` -range~ the program branches to Step 120.

Step 120: Signal correction possible.
.. ~ :,: ,;.- ' -If the peak intensity lias in a certain pre~
20 determined range of low values, then the signal-to-noise ~ ~;
ratio can bé improved by summing ad~acent pixels, i.e. the pixels 1-512 shown in Figure lA. If the peak intensity is below this range, it is deemed too weak for analysis, even ~-by the pixeI-adding technique of this step, and the program branches back to Step 110. The range for correct~
able intensiities for the peak value can be empirically determined, and depends upon such factors as the relative magnitudes of the pieak intensity signal and the signal-to- `~
noise ratio, the contrast in the case of images r the -~
30 i~sensitivity of the!detector, and other factors which may -~ -~
be found to adversely affect the reliability of the ~;
signal, especially as it ffects the ultimate computer determination of whether given tissue is normal or ab~
normal. -;
.
35~ :~

~' ; ' ,:

;",: ~:

~53~ 2~S35 `;:
."
The purpose of this step is, therefore, to ~ ~ ;
determine if the peak intensity lies in the range in which ~
the signal-to-noise ratio can be improved. In the ~ i preferred embodiment, a signal correction is made if the ~ ~ -peak intensity is between the minimum threshold intensity and one-eighth of the minimum threshold.
:~'.~ ,",'..
, . , .: .:
Step 122: Correct signal.
:~;' .''''' ',':
By the action of the diffraction grating (not -separately shown) carried by the spectrometer 31 shown in Figure 1, the spectrum of fluorescent light obtained from in vitro specimens or a patient is diffracted such that each pixel 1-512 in the linear array 38 (see Figure lA) 15 receives a narrow band of wavelengths of the spectrum, and ~;
the computer then stores 512 signals, each having an intensity corresponding to a given band of wavelengths.
The graphs of Figure 5~`are representation9 of such data.
In t~he present embodiment, since the spectrometer detec~
~20 tlon range chosen is about 375 to 650 nm, this leads to a -~
spectrometer resolution of 512/(650-375) pixels per nm, or approximately 1.9 pixels per nm. Other resolutions may be utilized by changing the 9ize of the array 38 (including the size of the individual pixelY~ or by changing the dif~
25 ~fraction grating. ~ ;~
If the signal received by a given pixel has in~
ufflcient~intensity, it is then corrected by grouping or summing signals received from ad~acent pixels. This reduces the wavelength resolution, but improves the ` `;
30 l!signal-~o-noisélratio.~ The mçthod first groups two plxels; ~ ~ ;
~uch as pixels 1 and 2, pixels 3 and 4, etc., in Figure lA)~,~adding their intensities, and tests if the new peak ~ ~ ;
intensity is adequate. If the intensity remains in- ;~
adequate it then groups 4 pixels (e.g., pixels I-4, pixels 3s 5~-8, etc.),~and again makes the intensity test. If the i 3; ::9 :",:,: .: ~...

';~ ;',,'"""' peak intensity is still below the predetermined threshold, `~;
the program group~ 8 pixels and once more makes the intensity test. ~ -Applicant has found that the capability of the method to discriminate tissue types is not significantly diminished by the loss of wavelength re301ution by a fac~
tor oi 8 when pixels are summed or grouped by 8, which is -at least in part due to the fact that the cpectra gener-ated tend to be rather smooth, with few or no steep slopes between intensities, because the intensity responses of similar wavelengths are relatively close. The effective result of grouping a 512-point spectrum eight pixels at a ~ ;
time is to produce a spectrum with 64 points which is eight times as intense. Other signal corrections may be 15 needed; for instance where images are being generated, ~ ~
enhancement of the images may be desirable, such as by ~ ~ ;
using known, computer-controlled image improvement techniques.

20 Step 124: Parabola fit te~t. ;~

Th~ intensity range set in Step 108 is tested in Step 124. At this time the method tests the peak position to see if it is in a range in which both diseased and healthy tissue peaks are observed. Also at this time the lea t squares deviation of the parabola fitted to the data to find the peak is compared with the data. If the iit is not good te.g., if the r value in standard least squares iormulae is not acceptable), then the curve does not have 30 i the general shape found for healbhy and unhealthy tissue. r~
In an alternative embodiment, the system is pxovided with an imaging-capable system, such as ultrasound or~magnetic resonance, and in such an embodimènt, a standard image correlation technique is used ; ;

'' .'.'' .,' ,"'' ,;`.` '.;
: :: -:-:,.:

2~535 .,, ,...' ~,, ., .. ", .. . .
, . ", .. .. .
to compare an image taken at this time with a referenceimage, to test for proper orientation of the fiber.

Step 126: Test for blood absorption.
, ~.
Blood has a strong absorption feature in the wavelength range of the tissue fluore cence feature used to discriminate tissue types, as discussed above and as shown in graph 90 of Figure 5. Therefore, blood alters 10 the tis~ue flllorescence spectrum. There are some types of ;~
tissue which have hemoglobin absorption in them. The purpose of this step is ~o determine if the spectrum being analyzed is altered by a factor such as blood. This step may be referred to as the heme test, because the absorp~
tion feature of blood is due to the heme portion of hemo~
globin.~
; The test for blood absoxption first involves the ~determination of a number derived from a composite of -`~-normal tis;sue spectra obtained from in vivo diagnostic -; 20 cases. ~T!his~number~ which may be referred to as a standard normal composite, thus represents an empirically~
determined~ factor which is based upon past data and is used in a gi~en instance to determine how likely it is that the ti~sue being tested is normal.
2S ~ ~ ~The value for the standard normal composite ; ~;
depends on the system used to take the spectra because, as oxplained above, dlfferent~systemq may obtain different ~ ` ;
appearing spectra from the same source due, for instance, to equipment specification;differences. Typically, standardinormal spectra which have been de~eloped are constructed from about 50 spectra obtained from abqut 20 ~ ;
patients.~The spectra used~;to~compute a standard normal - compos1te appear quite similar and show little bIood absorption. First; the~spectra are aligned by the posi~
35 tions of their peaks and summed together to form a `~

`, .; . ' ,,, ,' , ~0~535 ,~ , " . .:

: . , ~, ":
preliminary composite reference spectrum. Next, each ~ ;
normal spectrum is shifted wi~h respect to the reference spectrum until a least square residual is found. Using the position where the best least square fit occurred, the spectra are summed to form the standard normal composite.
The center of the blood absorption feature usu-ally occurs at about 412 nm. This can be seen in the graph of Figure 5, which shows an absoxption dip on the left side, which begins rising sharply at about the 70th -~
pixel (which is about the expected value, since 1.9 pix/nm times (412-375=37) nm equals 70.3).
The ratio used to test for blood absorption is given by the ratio of: (1) the data peak intensity divided -~
by the data value at 412 nm for the tissue being tested, 15 to (2) the standard normal peak intensity divided by the ~`
standard normal value at 412 nm. If this ratio is below a threshold ratio, then there is blood absorption in the spectrum which the method must take into account. If blood a~sorption i~ indicated, then the program branches ~ -to step 132; if not, then it branches to step 128. - -~-The details of the method which tests for blood -~
may be different from those described in this test, but in ganeraI will have the feature in common that the presence of the strong absorption by blood at about 4I2 nm is ~ ;~
detected. Also, in alternative embodiments tests for othar error-introducing factors may be introduced, such as for th~ presence of compounds which may be present that ~ ~-influence the fluorescence spectrum, or for any other fac~
tors which may affect the accuracy of the spectrum ~i-30 i de~ecteq from the patiqlnt.
As an additional portion of this test, a spectrum variations test may be employed. In this test, each test tissue spectrum (except the first one generated in~a gi~en procedure)~ is compared with the preceding test ;
tissue spect`rum, to see if there is a great variation -.'''~ '` '',`` ' '.'`

Z0~535 between the two. The comparison may be done by a lea~t- -squares analy~is, by taking a ratio of the two spectra, or by other equivalent methods of detecting variations. If a variation is found and is higher than a certain pre-determined threshold (which is empirically determined)~
this is an indication that blood has come into the field of view of the optical fiber, and the program branches to Step 132. Alternatively, a high variation may cause the program to branch back to Step 110 (not separately shown in Figure 8).

Step 128: Peak position test (no blood present).

This step tests for whether the detected peak for the test tissue is in an empirically-determined normal range. The range chosen -~hould be based upon the study of many normal tissue spectra. Applicants have determined that a range of approximately the 140th pixel to the 180th pixel i8 to be expected for normal tissue, as reflected in graph 92 of Figure 5. This correspond~ approximately to a wavelength range of (140xl.9 + 375) to (180xl.9 + 37s) nm, or about 638 nm to 727 nm. There are thus two peak posi-tion thresholds, one at the lower boundary for normal tis-sue and one at the upper boundary.
As explained in Step 100, a parabola is fitted to the ~pectrum and the peak of the parabola is used as the ~pectrum peak, becauqe the true peak shape is more sensitive to the effects of noise. Materials whose spectra have peaks outside the range found for normal tis- ~ -sue are ablated by the treatment laser.
If the detected peak is outside the normal range, then the program branches to step 136, and the laser is caused to fire at the tissue. If the detected peak is in the normal range, this still doas not necessar-ily mean that the test tissue is normal, and the program 2~535 ;: ~
-58- ~
, ,. - ~
~ . .,,,," .:",,.

branches to step 130 for the tissue index test, discussed ;~ -below.
Variations on this step may be implemented. For instance, it may be found that a certain type of abnormal tissue typically has a peak which is higher than the normal range, but rarely or never has a peak which is lower than the normal range. If it is desired that only t~is particular type of tissue be removed (and other ab-normal tissues, having lower-than-normal peaks, be allowed , to remain), then the program can be adjusted to test not only whether the peak is in the normal range, but also ;~
whether it is on the high side or the low side. If it ic on the high side, then the program would branch to step 136 as before; but if it is on the low side, then the~
program would branch to step 130.
:
Step 130: Tissue index test (no blood present).
: '~:, ' ~'.
This step tests how closely the shape of the 20 spectrum~for the test tissue matche the shape of the `
standard normal spectrum. This is done by generating a va}ue which may be referred to as the tissue index, which is a least-squares residual, for the tissue being tested.
The tissue index value is based upon the least-square 2S deviation of the test tissue spectrum from the normal tis-. ~
sue spectrum.
The tissue index test, which may aiso bereferred to as the standard normal test (since it tests for deviation from the standard normal spectrum), involves 3d ~ Iseveralltypes ofi,parameters, including: the tissue indqx threshold; (2) the cursor or pixel positions used to make ~ `
the~calculations; and (3j shift alignmeht factors. These are~discussed~below.
Figure 6 explains part of the tissue index test 35~ proc~edure. ~A, B, C and D in Figure 6 are pixel positions, `~-~
: : ,. ~ .. .
, "~,;:' ~` '`.' '' : : ''.; ` ." ~ ,' ,~`

: ,: .~. ,: . : .

2~ 535 . ~:,',~...' and the fixed curve is the standard normal curve which is compared with the data. The standard normal curve to be ~ -~
used is the composite spectrum generated as shown in the flow chart of Figure 3 above, and preferably numerous spectra (on the order of 100) are ukilized in generating the composite spectrum.
The test tissue data changes every time the detector is read. The formula used to compute the least square residual (tissue index) is given. The greater the difference between the data and the fixed tstandard normal) curve, the greater the likelihood that the data is ~ ~
a spectrum of diseased tissue. ~ ;
The tissus index threshold is calculated by the formula shown in Figure 10, as follows. First, the test ;;~
tissue data is normalized by dividing the entire set of data by the value found for the peak intensity for the test tissue spectrum. Then,;the numerical differences ; ~; ~
between the test tissue spectrum and the standard normal ~ ~ -spectrum~are found at five predetermlned pixel locations, the five pixel locations being chosen such that the numerical differences may be expected to be maximized.
This is done by observing many sample spectra of the type ; ;-of abnormal~tissue being treated (such as plaque) from `. :`' ''`~,'~1 '`
numerous patients, and by observation, averaging - ;; ~
;techniques or other statistical analysis, determining the ` ;; `
frequencies (and hence the pixel locations) at which the - -` `
test tissue spectra differ most from the normal spectrum.
~These pixel values are stored in the memory of the - ~ "
computer. . ~ J;'.''~
In one implementation of the invention, the ,~
pixel positions (referred to`in Figure lO by the variable Dli) are not stored as absolute pixel positions; rather, -they are stored as differences between the pixel position `~
35 of ~he peak of the curve in~question and the desired pixel ` -`
,: ~, : ,~

~ -60- ~ O ~ 53 S ~ ;~
.,... :
' :' ., .,' , "::, .. .
position. For instance, if the peak for the normal spectrum i9 at pixel 170, and one of the empirically- ;
determined maximum curve differentials (between the curves for abnormal tissue and normal tissue) is at 70, then Dl (i.e. Dli where i=l) is stored as -100. Likewise, if the other four pixel locations which maximize the curve dif-ferentials are found to be 80, 250, 310 and 370, for ~ -example, then the values of D12 through D15 are -90, 80, 140 and 200, respectively. The variable D1 may be -referred to as a pixel differential variable.
Although for the sake of illustration only two ~
pixeL locations (A and C) other than the peaks (B and D) ~ ;
are shown in Figure 6 for maximizing the curve dif-ferentials, in the preferred embodiment, as discussed above, five such pixel locations are chosen. Other numbers of curve differential-maximizing pixel locations may be chosen, either more or fewer, depending upon how ~ -; the operator determines that the choice of isuch pixel locations affeats the reliability of the tissue tests~
For instance, for a given type of tissue it may be found that~there are three particular pixel locations relative - ;
~ to the peak intensity where the test tissue spectra reli-;~; ably;differ by a large amount from the normal tissue spectrum. In that case, those three pixel locations alone 25 would be utilized. On the other hand, it may be faund ~ "
that there is a broad range over which the test tissue spec~tra differ from the normal tissue spectrum, in which case a larger number of relative pixel locations may be chosen for the tissue index test.
30i In the preferredlemboldiment,~two pixel positions on the left side of the spectrum and three pixel positions on the right side of the spectrum are used to compute the least~
square residual. In the formula of Figure 10, therefore, ` ~ `
N=S for the preferred embodiment.

: ~ ': .::.'.
`''-, ,,",~

`':

2~ i3~
Once the pixel differential variables are empirically chosen, they are stored in the computer for use whenever tissue ablation is conducted on the type of tissue from which the pixel differential variables were determined. Another set of pixel differential variables, D2~, where j runs from 1 to 7 in the preferred embodiment, must also be chosen. These are preferably small numbers, such as -6, -4, -2, 0, 2, 4 and 6, and each tissue index `; `~
value TIj is determined for one such value of j. Thus, I0 seven values for TI are generated (TIl through TI7) by the formula of Figure 10. The use of the variable D2, which may be referred to as the fine shift adjustment variable, compensates for possible misalignments between the peaks of the test tissue spectrum and the standard normal spectrum. The program then chooses the smallest TIj value of these seven, and this is stored as the value for the ;~
tissue index of the test tissue under treatment. The `~ `
reason for choosinq the smallest TI value is to minimize ~`
the possibility that normal tissue will be ablated as ab~
20 normal tissue. ; - `~
One method for selecting a set of values for ; ` -each of the pixel differential variables is to first ;-~
generate a composite normal spectrum, as discussed relative to Figure 3 above, and then to compare successive abnormal tissue spectra with the composite spectrum, either physically or mathematically, and arrive at one, several or many sets of pixel differential variables. A `~
tissue index value can then be generated, using these `~
pixel differential variables, for each of the abnormal ;~
30 It~issue s~ectra, and the,results of the tissue inqex tests for the abnormal tissue spectra are scrutinized for reli~
ability in predicting the presence of abnormal tissue. ;`; ~-The~set of pixel differentiaI variables which yields the ~largest percentage of correct identifications for the ab~
normal tissue spectra should be chosen. In addition, the .:.: ;. ~, ~ ,. ,-:,...
:~ . ,. :
:
:, l:
; -62~ 535 : :.

normal tissue spectra from which the composite spectrum was generated should be subject to the tissue index test for each set of pixel differential variables, and the results scrutinized for reliability in identifying normal tissue as such. There may be no single set of pixel dif-ferential variables and threshold values for which 100~ of each of the normal and abnormal tissues are correctly identified, in which case a balance must be reached; in making such a balance, it is preferable to err in favor of identifying tissue as normal, rather than abnormal. Most preferably, the threshold will be chosen that 100~ of the normal tissue will be identified as such, while maximizing the percentage of abnormal ticsue which is correctly -~ ;
identified.
In applicants' experience, a value for TI which yields reliable results for atherom~tous tissue is about ~-0.005. If the TI value yielded or test tissue under treatmant is gxeater than 0.005, then the program branches ~ ;
to Step 136; and if not, then the program branches to Step ~ -~
110. Thus, the value of 0.005 comprise~ a threshold for TI. This threqhold should be chosen such that it is highly unlikely that normal tissue will be identified as diseased tissue. Generally, the large~t tissue index ~computed for a normal tissue spectrum is chosen as the threshold. This choice, combined with the choice by the program of the smallest value of TI; which i5 generated as ` -~
the value for the tissue index for the test tissue, maximizes the likelihood that there will be an overlap -~
between the tissue indices of the normal and abnormal tis~
30i sue. Thus, thellikellihood!is minimized that, folr a given tissue, erroneously large TI values will lead to ablation of normal tissue. ~ -Each of the differential variables Dll-D15 and D21~D27 is determined in advance of actually carrying out an operation on a patient. With experience, the operator -~
~ ':: , ` . . . ;' -: :.:. ~::.
i".:.:, -63- 20~535 may wish to alter the values of these variables for a qiven type of abnormal tissue, in which case the tissue index threshold (given as 0.005 in the above example) will have to be altered accordingly. The details of this step may also need to be altered if the optics of the system are altered.

Step 132: Peak test (blood present).

In the case of the peak test of this step, where blood has been determined to be present in the field, the larger of the two peak thresholds is set to a slightly ~;
hiqher value than the threshold for spectra having no blood absorption. Otherwise, this step is identical to -step 128, which is the peak test where no blood is present. Thus, instead of an expected pixel range of 140 to 180 for the normal peak, the upper threshold is raised to 185; and this may be ad~usted as normal-tissue data may indicate. The upper threshold is rai~ed because it has been found that the presence of the blood absorption dip (as in graph 92 of Figure 5) skews the peak slightly to the right, such that it appears to be occurring at higher frequencies, corresponding to higher pixel numbers.
If the test tissue peak i3 within the prescribed range, in this case from the 140th to the 185th pixel, then the program branches to step 134. If the test tissue peak i5 outside this range, the program branches to step ;
136 for ablation by the laser. ;

Step 134: Tissue index test (blood present).

Step 134 is identical to Step 130, except that the tissue index value is generated only for data on the ;;
right side of the graphs of Figure 5, i.e. to the right of the peak for the spectrum of the test tissue, because the :.,. ~
.:, :'' , ' , i , .

,. :, .

2~ 53S
: -64-blood absorption dip distorts the data corresponding ~o the lower pixel number~. It will be appreciated that in this case, the v~lues of the curve differential variables Dli and D2j are only positive. The tissue index test is thus modified in this way to account for the effect of blood absorption on the spectrum, and may be otherwise modified as necessary or empirically determined to avoid -misleading factors in the data. If the t$ssue index test yields a tissue index value for the test tissue which is outside the range for normal tissue, then the program branches to Step 136 (as was the case with the tissue index test of Step 130). Otherwise, the program branches to Step 110.

Step 136: Open treatment laser shutter and fire.
At this time, the shutter 34, which may be car-ried in the cavity of the treatment laser 17, must be opened before it can be fired. The shutter 34 serves as a safety feature, so that~the treatment laser does not fire if the laser is accidentally triggered. Also, the shutter prevents contamination of the diagnostic signal by light which may be emitted by the treatment laser flashlamp even when the laser is not firing. In a preferred embodiment, 25 the la~er output energy is also ad~usted; in the case of a --~. Z ;`
pul~ed laser, the total energy or the power of each pulse ~ X :
may~be ad~usted, depending on the depth of ablation desired, the intensity of the abnormal tissue spectra be~
ing detected, and other factors. The treatment laser i5 . ,~' ~ ;.".'~.:-' fired after any~ such adjustments are made.

The above me~hod is carried out repeatedly and -~ ;
quickly, as controlled by the computer, for a pre-~ detèrmined amount of time or number of cycles. Thus, if method i5 carried out, for example, 50 times and then 2~53~

stops, the operator is given an opportunity to assess the r~sults b~fore beginning treatment again.
When the iterative method of the invention is utilized, each s~ep in the iteration (such as Steps 124, 126, 128/132, 130/134, etc.) increases the probability that the determination made by the computer as to whether the tissue is normal or abnormal will be correct. As mentioned above, the thresholds are preferably chosen such that substantially all normal tissue is correctly identi-fied as such, and these thresholds result in a given percentage of abnormal tissue (usually less than 100 being correctly identified in each of the tests represented by the above steps. ;~
By way of example, assume that Step 132 yields a 70% correct discrimination rate for plaque (i.e., identi~
fies plaque correctly as such 70% of the time), and that Step 134 yields a 50~ discrimination rate for plaque. In ;~
carr~in~ out Step 132, then, 70% of the time that plaque is under observation the program will branch to Step 136, ~ ~`
and the other 30~ of the time the program will branch to ~ :
Step 134. When the program branches to Step 134, if plaque is present then 50% of~the time the program will then branch to Step 136, and the other 50% of the time it ~`
will branch to Step 110; thus, half of the plaque which ; ~
escaped Step 136 will be caught by Step 134. The total ~;
percentage o plaque caught by either Step 134 or 136 (and ~ ;
thus treated due to Step 1363 will therefore be 70% ~ (50% ;;
of 30%) = 85%, a 15~ improvement over the test of Step 132 alone, and a 55% improvement over Step 134 alone. The percentage of successful id~ntifications of abnormal tis-sue is thus increased for each additional step in the iteratlve~process.

Referring now to Figure 9, a generalized ap-~plication of the foregoing method is shown, which may be ~ ~

, ':

- , . -: .:

-66- 2~ 3~
! .

used in specific embodiments for many types of situations, including other types of medical treatment than those discussed above, and including other situations wherein a particular action (in the above example, the firing of the laser) may be called for.
Step 150 relates to whether new hardware is required for the task at hand. This may be, in the case of laser angioplasty, the determination of whether a given ~ -~
type of treatment will require a specific range of laser power; or, in a general sense, this step relates to the type of equipment to be chosen for particular application of th~ method. ,~
Step 152 reflects the fact that the choice of equipment may affect the types of results which are obtained. Again using laser ablation of abnormal tissue as an~example~, this step may require the adjustment of the diagnostic signals to be generated from tissues, so that they match those generated earlier or are consistent with one another. ~For instance~, if new optics are utilized and -20 the same set of standard~normal spectra are used for reference for a given patient, the newLy-generated data ! }i;;
; for test;tissue may have to be ad~usted to compensate for variables in the new optics, including chromatic aberra- ~i^`i`;
tion, absorption and reflection characteristics, focusing : 25 characteristics, and the like.
Step 154 relates to the actual collection of diagno~tic data, which in the embodiment discussed above ~ `
refers~to the~generation of the normal tissue spectra, the composite spectrum, the abnormal tissue spectra, the pixel b Idifferential varliablles,jand so on.
Step 158 relates to the generation of an `~
algorithm for the utilization of the data generated in Step~154. A~given technique~is first introduced (as indica~ed by Step 164), and the da~a generated in Step 154 ~ls~then utilized to develop an algorithm to implement the , : .

. . .

2~ 35 -67~

technique. For instance, if it is determined to be , advantageous to qtandardize test tissue spectra to the , spectra generated from normal tissu~ (a new technique), i for the purpose of identifying atheromatous tissue, then a particular curve-fitting formula or method for generating pixel differential variables (algorithms) may be developed ~ ~
for implementing the technique. ; ~-Once th~ algorithm for the given task is ;~
developed, it is tested to determine whether it is adequate to make the decisions necessary for the situa~
tion, as represented in Step 166. In the above example, ;;~
the end resulk is the proper or improper determination of ~ ;
whether the test tissue is normal. If the chosen algorithm leads to incorrect results, it must then be ; ~ :~
inspected and perhaps modified and retested to determine ~-15 whether it~ is even capable of yielding adequate ~
discrimination between the decisions necessary to be made. ~ -Thus, if inadequate discrim1nation is found, the method of Figure 9 branches to step 156, wherein it is inspected whether an~adequate new (or revis~d) al~orithm can be ~
developed. If not, then the method branches to step 150, ~ ;;
which requires the development of now hardware; if so, than step 158 is again reached, to introduce a new technique or develop a new algorithm for ~he task. ~ ~;
Once the algorithm is found to yield reliable -results, the method reac~hes step 168, wherein the new ~ algorithm ii implemented (and new thresholds, in the case ;~
;~ ~of the above axample)~ Then, in step 170 the method calls for the determination of whether there are any new hardware control parameters which can or should be im-30~
plemented, which is accomplished in step 172 if that i9 the case. This may relate, for example, to the utiliza~
tion o~f different laser pulse energies.
At step 174, the actual clinical case is reached, and ~he algorithm and hardware parameters chosen ; ..: -. : :..::': '.

2G1~53~

are put to the test. In the example at hand, the laser is fired if so indicated, and in other examples, other ac~
tions may be taken. Step 176 questions whether the outcome was satisfactory; if, for example, the physician finds that all of the abnormal tissue in a patient has been ablated, without the ablation of an undue amount of normal tissue, then he may decide that the operation has been successful, and step 178 (finish) is reached.
If the algorithm developed in Step 158 is found ~ ~
10 to be unsatisfactory in practice, then Step 180 i5 . ', reached, at which it is determined whether a change in the hardware control will suffice to solve the inadequacy. If so, then Step 172 is reached for such a change. If not, then the clinical data generated in step 176 is analyzed, the method branches to Step lS8, and a new technique is introduced as indicated by Step 164. The method is then repeated until satisfactory results are achieved. ~s`; ~
~he foregoing shows that the particular method ~ ,;
of the invention relating to laser anqioplasty has implications in a broader sense, such as to clinical treatment of medical problems in general, where actions must be taken -particularly in the area of surgery--based upon the collaction and analysis of data relating to the patient. Other variations on the foregoing are possible to accommodate different types of tasks.
Although the invention has been described in terms of the preferred and alternative embodiments described herein, those skilled in the art will appreciate other modifications which could be made without departing ~;
30 from the true spirit andlscope of~the invention.~ All s~uch ~ -modifications are intended to be included within the SCGpe of the claims appended hereto.

: ' ' ;, :''~ '"

-'.'','`'`',`"

Claims (37)

1. An apparatus for unblocking blood ves-sels, comprising:
first means for guiding excitation light to the site of blood vessel restriction and for analyzing fluorescent return light caused by said excitation light to determine if the peak fluorescent intensity is within a predetermined range and to determine if the wavelength of said peak is within a predetermined range of wavelengths;
and second means coupled to said first means for guiding treatment laser light to the tissue causing said fluorescent light if the analysis by said first means indicates that said tissue is likely to be diseased tis-sue.

2. The apparatus of claim 1, further comprising means in said first means for comparing the shape of the spectrum of said fluorescent return light to the shape of a composite healthy tissue spectrum and for determining if the shapes are similar within a predetermined measure of similarity and for preventing said second means from guid-ing said treatment laser light to the site of said surgi-cal procedure if said shapes are similar within said pre-determined measure of similarity.

3. The apparatus of claim 2, further comprising means in said first means for computing a ratio of fluorescent intensity at a predetermined wavelength less than the peak wavelength to the fluorescent intensity at the peak wavelength of a composite reference spectrum for normal tissue and for computing the same ratio for said fluorescent return light and for comparing said ratios to derive a factor and for comparing said factor to a pre-determined factor threshold, and for preventing said second means from guiding said treatment laser light to the site of said surgical operation if the comparison of said factor to said factor threshold indicates that the fluorescent return light may have originated from healthy tissue.

4. A means for unblocking blood vessels, comprising:
means for storing a composite reference spectrum of fluorescent light intensity for healthy tissue and for storing at least some of the peak intensity values for healthy tissue fluorescent spectra used to compute said composite reference spectrum as lower and upper intensity thresholds, and for storing the wavelengths at which at least some of said peaks for healthy tissue fluorescent spectra occurred, and for storing a shape threshold indicating the measure of similarity of shape between said composite reference spectrum and the worst fitting one of the healthy tissue spectra used to generate said composite reference spectrum;
first means coupled to said means for storing for guiding excitation light to the site of blood vessel restriction at a surgical procedure site and for analyzing fluorescent return light caused by said excitation light to determine if the peak fluorescent intensity is within a predetermined range of intensities defined by said lower intensity threshold which is set to ensure adequate signal-to-noise ratio and by said upper intensity threshold which is set according to the peak fluorescent intensity of known diseased tissue from the patient oper-ated upon and further based upon known peak fluorescent intensity values from healthy tissue stored in said means for storing, and for determining if the wavelength of said peak intensity of said fluorescent return light is within a predetermined range of wavelengths defined by the maximum and minimum wavelengths of the peak fluorescent values from healthy tissue spectra used to generate said composite healthy tissue reference spectrum stored by said means for storing, and for determining if the shape of the spectrum of said fluorescent return light is similar to the shape of said composite reference spectrum of fluorescent light from healthy tissue within said shape threshold, and wherein said composite reference spectrum is derived from healthy tissue excited to fluoresce by excitation light of the same wavelength as that used to excite tissue at said surgical procedure site;
second means for guiding one or more treatment laser light pulses to the tissue causing said fluorescent light if the analysis by said first means indicates that said tissue is likely to be diseased tissue.

5. The apparatus of claim 4, wherein said means for storing stores the data defining said composite refer-ence spectrum from healthy tissue in non volatile memory.

6. The apparatus of claim 4, further comprising means in said first means for computing the ratio of fluorescent intensity of said composite reference spectrum at a predetermined wavelength less than the peak wavelength of said composite reference spectrum to the fluorescent intensity at the peak wavelength of said composite reference spectrum for normal tissue, and for computing the same ratio for said fluorescent return light, and for comparing said ratios to calculate a heme stain factor and for comparing said heme stain factor to a predetermined heme stain factor threshold, and for preventing said second means from guiding said treatment laser light to the tissue generating said fluorescent return light if the comparison of said heme stain factor to said heme stain factor threshold indicates that the fluorescent return light may have originated from healthy tissue.

7. An apparatus for ablating diseased tissue in the body by irradiation with laser light, comprising:
means for exciting to fluorescence tissue in a body by irradiation by laser light;
means for comparing the spectral characteristics of the fluorescent return light to compare the peak intensity, peak wavelength and spectral shape to known standards and for determining whether or not to fire a treatment laser at the tissue which emitted said fluorescent return light; and control means in said comparison means for controlling a treatment laser to emit light energy when a decision to fire said treatment laser is made.

8. The apparatus of claim 7, wherein said means for comparing includes means for storing a composite fluorescent light reference spectrum for healthy tissue computed as the average of a plurality of fluorescent light spectra of healthy tissue which have been normalized to said composite reference spectrum peak intensity and shifted to the position of best fit as indicated by the smallest least squares residual.

9. The apparatus of claim 8, wherein said means for comparing includes means to store predetermined upper and lower intensity threshold values and means to find the peak intensity value by curve fitting a parabola to said return light spectrum, and for comparing the peak intensity so determined to the range defined by said upper and lower intensity threshold values and for preventing said control means from firing said treatment laser if the peak intensity of said return light is either less than said lower intensity threshold or greater than said upper intensity threshold.

10. The apparatus of claim 9, wherein said means for comparing includes means for determining the wavelength of peak intensity for said fluorescent return light by fitting a parabola to the spectrum of said fluorescent return light, and wherein said means for stor-ing includes means for storing the range of wavelengths encompassing all the peak intensity values of said individual healthy tissue fluorescent spectra which were averaged to compute said composite reference spectrum for healthy tissue, said means for comparing including means for determining if the wavelength at the peak intensity of said return light spectrum is within the range of wavelengths in which all the peak intensities of said individual healthy tissue spectra were found and means for causing said control means to prevent said treatment laser from being fired if said peak of the return light spectrum is found to be within the wavelength range of peaks for the healthy tissue spectra.

11. The apparatus of claim 10, wherein said means for storing includes means for storing a shape threshold defining the worst fit of any of said healthy tissue spectra averaged to form said composite healthy tissue reference spectrum, and wherein said means for comparing includes means for computing the least squares residual for the best fit between the fluorescent return light spectrum and said composite reference spectrum and for comparing the least squares residual so computed to said shape threshold and for causing said control means to fire said treatment laser if the best fit between said fluorescent return light spectrum and said composite reference spectrum is worse than the worst fit of a healthy tissue spectrum used to compute said composite reference spectrum.

12. The apparatus of claim 11, further compris-ing means in said means for comparing to determine the ratio of intensity at a predetermined wavelength less than the wavelength of the peak intensity to the intensity at the peak for said fluorescent return light and to determine the same ratio for said composite reference spectrum and for comparing said ratios to each other and to a predetermined heme stain threshold and for causing said control means to prevent said treatment laser from firing if said comparison of ratios indicates that the tissue emitting said fluorescent return light may be healthy tissue with heme stain absorption.

13. An apparatus for discriminating between normal tissue and diseased tissue and for firing a laser pulse at tissue deemed to be diseased tissue, comprising:
excitation means for guiding excitation light to the site of a surgical procedure and for guiding fluorescent light caused by said excitation light to a return light output;
first means coupled to said light output for comparing the peak fluorescent intensity and the wavelength at said peak intensity for said return light from said surgical procedure site to predetermined criteria;
second means coupled to said first means for comparing the shape of the fluorescent light spectrum of said return light to the shape of a composite healthy tis-sue spectrum to determine if the shapes resemble each other within a predetermined degree of similarity if said first means determines that said predetermined criteria are met; and means for guiding treatment laser light to said surgical procedure site if said first and second means determine that said predetermined criteria are met and that said shapes do not match each within said pre-determined degree of similarity.

14. A method of generating reference data for use in making discrimination decisions for control of a treatment laser in firing at diseased tissue but not at healthy tissue, comprising the steps of:
exciting a plurality of samples of healthy tis-sue to fluorescence with excitation light;
detecting the intensity spectrum of fluorescent light emitted from each said sample of said healthy tissue at a plurality of frequencies;
finding the peak intensity of each said spectrum;
assigning one said spectrum as an initial composite reference spectrum;
normalizing each said spectrum to the peak intensity of said initial composite reference spectrum;
computing a plurality of least squares residuals for each spectrum at a corresponding plurality of shifted wavelength positions for each spectrum to find the best fit between each said spectrum and said initial composite reference spectrum;
shifting each said spectrum to the position of its best fit; and averaging all of the spectra with said initial composite reference spectrum to derive a final composite reference spectrum.

15. The method of claim 14, wherein the step of finding the peak intensity of each said spectrum comprises the step of fitting a parabola to each said spectrum using the least squares residual method.

16. The method of claim 14, further comprising the step of determining the range of wavelengths which includes all the wavelengths at which peak intensities occurred for the spectra used to form said final composite reference spectrum and storing said range as peak position range.

17. The method of claim 14, further comprising the step of determining the minimum least squares residual for the best fit of the worst fitting spectrum used to compute said final composite reference spectrum and stor-ing same as a shape threshold.

18. The method of claim 14, further comprising the steps of computing the ratio of intensity at 425 nm to the peak intensity at the peak for said final composite reference spectrum and computing the same ratio for heme stained normal tissue and comparing the two ratios mathematically to derive a heme stain threshold.

19. A method of distinguishing between healthy tissue and diseased tissue, comprising the steps of:
illuminating the tissue to be distinguished with excitation light;
detecting the fluorescent return light spectrum;
comparing the peak intensity, peak wavelength and shape of the fluorescent light spectrum to pre-determined reference data; and guiding treatment laser light to said tissue if said comparison indicates that diseased tissue caused said fluorescent return light.

20. A method of distinguishing between healthy tissue and diseased tissue, comprising the steps of:
(1) guiding excitation light to illuminate the tissue to be distinguished;
(2) guiding fluorescent return light resulting from said excitation light to a detector, (3) determining the fluorescent intensity spectrum for said return light at a plurality of wavelengths;
(4) determining the peak intensity and the wavelength at said peak intensity for said fluorescent return light spectrum;
(5) comparing the peak intensity to a pre-determined intensity range;
(6) if said peak intensity is within said pre-determined intensity range, comparing the wavelength at said peak to a predetermined range of wavelengths, and, if said peak intensity is not within said predetermined intensity range, returning to step (1);
(7) if said peak wavelength is within said pre-determined range of wavelengths, comparing the shape of said fluorescent return light spectrum to a composite reference spectrum for healthy tissue to generate a shape similarity factor and comparing this shape similarity fac-tor to a shape threshold, and, if said peak wavelength is not within said predetermined range of wavelengths, guid-ing treatment laser light to ablate the tissue causing said fluorescent return light;
(8) if the shape of said return light spectrum fits the shape of the composite reference light spectrum such that said shape similarity factor is within said shape threshold, returning to step (1), and, if the shape threshold is exceeded, guiding treatment laser light to ablate the tissue causing said fluorescent light.

21. The method of claim 20, wherein steps (1) and (2) are carried out by guiding both said excitation light and said return light, respectively, over at least a first optical fiber.

22. The method of claim 20, wherein the step of determining the peak fluorescent intensity of said return light comprises the step of fitting a parabola to said fluorescent return light spectrum by a least squares residual method.

23. The method of claim 22, wherein step (5) comprises comparing the peak intensity of said fluorescent return light to determine if a minimum intensity threshold, set to ensure adequate signal-to-noise ratio, is exceeded and comparing the peak intensity of said fluorescent return light to determine if the peak intensity is less than or equal to a maximum intensity threshold, set to ensure that fluorescent return light, which is normally of greater intensity than fluorescent return light from diseased tissue, is not mistaken for diseased tissue and ablated.

24. The method of claim 23, wherein step (6) comprises the step of comparing the wavelength at the peak for said fluorescent return light to a range of wavelengths which encompasses all the wavelengths at the peaks of a plurality of healthy tissue fluorescent spectra which are averaged after normalization and curve fitting to form said composite reference spectrum.

25. The method of claim 24, wherein step (8) comprises normalizing said fluorescent return light spectrum to said composite reference spectrum and finding the best fit between the two spectra using a least squares residual method and comparing the minimum least square residual to a curve shape threshold, said curve shape threshold being indicative of the worst fit between any of the normal tissue spectra used to compute said composite reference spectrum and said composite reference spectrum.

26. A method for optimizing clinical treatment of a patient suspected of having a given condition, comprising the steps of:
collecting from the patient diagnostic data expected to relate to the given condition;
selecting a first technique for treatment of the given condition;
analyzing the data and developing therefrom a process for clinical treatment based upon the technique selected;
applying the process to the diagnostic data and obtaining a first analysis result;
comparing results of the step of applying the algorithm to known cases of the condition for determining whether the algorithm discriminates between presence and absence of the condition;
if the comparing step indicates insufficient discrimination, altering the algorithm such that satisfac-tory discrimination is achieved.

27. The method of claim 26, wherein the diagnostic data includes data relating to both presence and absence of the condition in actual clinical tests.

28. The method of claim 27, including, before the step of collecting diagnostic data, the step of selecting hardware for the treatment.

29. The method of claim 28, including, before the step of collecting diagnostic data and after the step of selecting hardware, the step of testing whether the selected hardware requires manipulation of the diagnostic data, and if so, manipulating the data to compensate for biasing of the data due to the selected hardware.

30. The method of claim 26, including, after the comparing step and before the algorithm altering step, the step of altering hardware parameters for resolving the insufficient discrimination.

31. A method for the ablation of abnormal tis-sue in a patient, comprising the steps of:
(1) irradiating normal tissue with a first diagnostic medium for producing first return light from the normal tissue;
(2) generating a first spectrum from the first return light;
(3) irradiating a sample of tissue with the first diagnostic medium for producing second return light from the tissue;
(4) generating a second spectrum from the second return light;
(5) standardizing the second spectrum to the first spectrum;
(6) comparing the shapes of the first and second spectra, for generating a first variable relating to how closely the shapes coincide;
(7) comparing the first variable with a pre-determined first threshold;

(8) based on step (7), determining whether the tissue sample is normal;
(9) if the tissue sample is determined to be normal, ablating at least a portion of the sample.

32. The method of claim 31, including, after step 5 and before step 6, the step of:
(10) comparing at least a portion of the first and second spectra for determining whether blood is present with the tissue sample.

33. The method of claim 32, wherein, if blood is determined be present with the tissue sample in step 10, step 6 comprises comparing the first and second spectra at areas of the spectra where influence to their respective shapes due to the presence of blood is minimized.

34. The method of claim 31, including, after step 2 and before step 3, the step of:
(11) normalizing the first spectrum by determin-ing a peak position value for the first spectrum and dividing each value in the spectrum by the peak position value.

35. The method of claim 34, wherein step 5 includes the step of:
(12) normalizing the second spectrum by determining a peak position value for the second spectrum and dividing each value in the spectrum by the peak posi-tion value.

36. The method of claim 35, including, after step 6 and before step 7, the step of:

(13) shifting the second spectrum along an axis representing frequency of the return light.

37. The method of claim 36, including, after step 13 and before claim 7, the steps of:
(14) repeating each of steps 6 and 13 for a pre-determined number of times, each time shifting the second spectrum by a different amount and each time generating a new value for the first variable; and (15) determining which of the values of the first variable so generated is smallest of all the gener-ated values; and (16) shifting the second spectrum by the amount represented by the smallest generated first variable.