Apparatus and method for multidimensional characterization of objects in real time

• US 4,987,539 A
• Filed: 08/05/1987
• Issued: 01/22/1991
• Est. Priority Date: 08/05/1987
• Status: Expired due to Fees

• Current Assignee: Stanford University
• Original Assignee: Stanford University
• Inventors: Breitbard, Gary, Nozaki, Tom Jr., Stovel, Richard T., Parks, David R., Moore, Wayne A.
• Primary Examiner(s): Lall, Parshotam S.
• Assistant Examiner(s): Melnick, S. A.
• Application Number:
  US07/082,254
• Time in Patent Office: 1,266 Days
• Field of Search: 364/555, 364/413.08, 364/478, 209/3.1, 209/3.3, 209/579, 436/63, 436/172, 436/164, 422/73, 356/318, 356/72, 356/39, 382/6, 382/37
• US Class Current: 356/39
• CPC Class Codes: B07C 5/3427 by changing or intensifying the optical properties prior to scanning, e.g. by inducing fluorescence under UV or x-radiation, subjecting the material to a chemical reaction
• CPC Class Codes: G01N 15/14 Optical investigation techniques, e.g. flow cytometry
• CPC Class Codes: G01N 15/149 specially adapted for sorting particles, e.g. by their size or optical properties

Abstract

A system for characterizing objects to be sorted in real time and making sorting decisions using lasers to excite fluorescing dyes used to tag objects such as cells which flow through the laser beam in a stream of conductive fluid. The scattered light and fluorescent light at various frequencies define a multidimensional space. These light characteristics are sensed and digitized. The data so gathered is buffered and then compared to a binary classification tree. The individual parameters cause a certain path through the binary decision tree to be taken while the cell is passing from the point in the stream where the light characteristics are gathered to a point where electrostatically charged electrodes deflect individual statically charged drops containing cells to be sorted into various sorting bins. When the decision tree has been traversed, the sorting decision for the drop containing the cell which had those parameters is done. This sorting decision is buffered until the time the drop reaches the charged electrodes. The sort decision is then accessed and used to properly charge the electrodes so that the drop is diverted into the proper bin.

Claims

1. An apparatus for making classification decisions in real time comprising:

means for gathering and buffering digital data in multidimensional space having greater than three dimensions describing the characteristics of one or more events including FIFO storage means for storing said digital data in first-in-first-out order along with an event number for use in correlating said data from each said dimension from the same event; and

means for comparing selectable dimensions of said data in real time to a binary classification tree and for making a classification decision in real time based upon said comparison; and

means coupled to said means for comparing for receiving and storing said classification decision until such time as an event is to be sorted and then for using said classification decision to sort the event in real time to which said classification decision applies into the proper category.

2. An apparatus for making classification decisions in real time on events and sorting in real time said events based upon said classification decisions comprising:

means for asynchronously collecting multidimensional data about an event through a plurality of sensors providing analog signals at analog signal outputs, each said analog signal from one of said sensors having a peak for the corresponding event, said multidimensional data defining a space having more than three dimensions, each said analog signal output being coupled to an input of a corresponding channel means, each said channel means including means for sampling the corresponding analog signal and converting said corresponding analog signal to a digital data sample, each said channel means having a separate means for sampling each said analog signal corresponding to an event at the peak of said analog signal corresponding to said event and for converting each said sample to a digital data sample;

first means for temporarily storing said digital data samples in a FIFO buffer along with data by which data samples corresponding to any particular event may be correlated to that event, and where said FIFO buffer includes a separate FIFO buffer for each said channel means, said FIFO buffer storing said digital data samples in first-in-first-out fashion, and wherein each said separate FIFO buffer includes means to generate a control signal when data is present in said separate FIFO buffer;

means for synchronizing the access of all said digital data samples which pertain to a particular event by retrieving said data from said FIFO buffer when said control signal indicates that data from selectable ones of said channel means is present in all the FIFO buffers for the selected channels, and for comparing in real time said digital data samples so accessed to the node values of a binary classification decision tree and for generating classification decision data in real time based upon said comparisons;

second means for temporarily storing said classification decisions for a predetermined time; and

sorting means for retrieving from said second means the classification decision pertaining to each event at a predetermined time and for sorting said event based upon the classification decision pertaining to said event.

3. The apparatus of claim 2 further comprising event detection means for detecting when an event occurs and for generating an event signal, and further comprising an individual, programmable delay means corresponding to each said channel means for receiving said event signal and for delaying it by a programmable amount and then generating a sample signal and transmitting it to said means for sampling to cause said means for sampling to sample said analog signal.

4. The apparatus of claim 3 further comprising attenuation means in each channel means for attenuating the analog signal in each channel by a programmable amount and for outputting said attenuated analog signal as a compensation signal, and further comprising compensation means in each said channel means for receiving said compensation signals from predetermined ones of said other channels and for summing the compensation signals and decreasing the analog signal in the channel of interest by an amount proportional to the sum of said compensation signals.

5. A cell-sorting decision-making apparatus comprising:

a laser for generating a light beam;

a plurality of scattered light and fluorescent light detectors;

nozzle means for causing a stream of conductive liquid in which have been entrained a plurality of cells to which fluorescing dye has been attached to flow through said light beam, said cells to be sorted after they pass through the beam of said laser and for causing said stream to break into a plurality of droplets containing said cells to be sorted;

means for sensing when a cell has passed through said laser beam and for generating a first signal indicating the time of said passage for each said cell so detected;

means for receiving said first signal for each cell so detected and for generating a second signal in response to said first signal, each said second signal being delayed by a programmable amount from the time of occurrence of said first signal;

a plurality of channel means, each coupled to one of said scattered light or fluorescent light detectors, each said channel means comprising means for sampling and digitizing the analog signal generated by the corresponding detector as a cell passes through said laser beam, and each said channel means further comprising attenuation means for generating a compensation signal which is an attenuated copy of the signal being sampled by said channel means wherein said attenuation means is programmable, and each said channel means further comprising compensation means for receiving said compensation signals from selected ones of said other channel means and for compensating the signal being sampled for crosstalk between said channel means which could generated errors, said means for sampling and digitizing further comprising sampling means for receiving a corresponding one of said second signals for each event defined as a cell passing through said light beam passage of said cell through said light beam being detected and for sampling the analog signal generated by the corresponding one of said light detectors caused by each said event at the time of receipt of the corresponding one of said second signals, and, said means for sampling and digitizing further comprising digitizing means for converting the samples analog signal from said corresponding light detector to a digital value representing one characteristic of said event;

a FIFO storage means coupled to each said channel means for asynchronously receiving said digital value corresponding to said one characteristic of each event detected by the light detector coupled to said corresponding channel means and for storing said digital data in first-in-first-out order;

means in said plurality of channel means to generate a third signal when an event has been detected and for storing said third signal in said FIFO storage means with said digital data characterizing the event which caused the generation of said third signal, said third signal including event number data and channel number data;

means for receiving said third signal from each said FIFO storage means and for comparing said third signal to sort mode data supplied from an external source indicating which channels are involved in a particular sort mode, and when said third signal has been received from each said FIFO storage means for all the channels involved in a particular sort mode, for sequentially reading the digital values from only the FIFO storage means for the channels involved in a particular sort mode and outputting said digital values on a bus;

second FIFO means for receiving and storing the digital data from said means for receiving said third signal;

classification means for reading the data from said second FIFO means and for comparing the digital values representing all the characteristics of an event that were gathered for the current sort mode and for comparing said data to the nodes of a binary classification tree and for traversing through said binary classification tree on a path based upon the results of said comparisons and for making a sort decision based upon the path taken through said binary classification tree.

6. The apparatus of claim 5 further comprising:

sort means for receiving said sort decisions from said classification means and for generating a drop charge signal and further comprising means to couple said drop charge signal to said conductive stream for charging the conductive stream to a potential related to the sort decision and for electrostatically deflecting said droplets into a bin corresponding to said sort decision.

7. The apparatus of claim 5 further comprising means for loading binary classification tree data into said classification means.

8. The apparatus of claim 7 further comprising means coupled to said multiplexer means and to said classification means for merging the digital values representing the characteristics of a particular cell with the sort decision corresponding to said cell and for displaying said digital values on one or more two dimensional displays.

9. The apparatus of claim 8 further comprising user interface means for prompting the user to enter parameters to control the operation of said apparatus for making sort decisions such as the amount of attenuation to set on each channel to generate said compensation signal and the amount of delay between said first signal and each corresponding second signal, and for generating control data to control said apparatus for making sort decisions in accordance with the parameters entered by said user.

10. A method of sorting objects comprising the steps of:

gathering light scatter and fluorescence data in a multidimensional space having more than three dimensions by passing said objects through one or more laser beams and collecting scattered and fluorescent light using a plurality of detectors and an event detector coupled to one of said detectors to generate an event signal and a plurality of signal conditioning channels, each said channel including a sample and hold circuit and a FIFO buffer, with each said channel coupled to one of said detectors, by generating a sample signal using said event detector and delaying said sample signal a selectable delay interval for each said channel, the amount of delay for each channel being selected to correspond to the time delay from the time an event is detected at the detector to which said event detector is coupled to the time when the signal peak occurs in the signal corresponding to the same event as detected by a detector coupled to that channel, and for coupling said peak in said signal using said sample and hold circuit as triggered by said delayed sample signal, and converting the sampled signal value to digital data, and storing all said digital data corresponding to a particular event in the same relative position in all said FIFO buffers in said plurality of channels along with data indicating the event to which each piece of digital data in said FIFO buffers corresponds;

resynchronizing all said digital data corresponding to a particular event by retrieving all said data corresponding to a particular event from the same relative locations in said FIFO buffers and comparing said light scatter and fluorescence data in real time to a binary classification tree having node values which define the various classes of events to be separated and making a sort decision in real time regarding substantially all events; and

using said sort decisions to sort said objects in real time.

11. A method of processing data elements describing a plurality of characteristics of a plurality of events for purposes of sorting said events into classes comprising the steps of:

asynchronously gathering a plurality of data elements regarding a plurality of different characteristics of each of a plurality of different events to be sorted using a plurality of different channels of signal processing circuitry where each channel gathers data elements regarding one characteristic of each event to be sorted by sampling a signal corresponding to said event when the peak of said signal occurs to generate a data element corresponding to one particular characteristic of the event, and electronically compensating said data elements for crosstalk between all other channels that cause errors in the data elements by receiving compensation signals in each channel from all the other channels and subtracting in each channel the sum of these compensation signals from said data element generated by said channel;

temporarily storing said data elements as they are gathered until all the data elements regarding a particular event are gathered; and

reading the data regarding an event and comparing the data to the nodes of a binary classification tree and making a sort decision based upon the path taken through said binary classification tree based upon the results of each comparison.

12. The method of claim 11 further comprising the step of displaying on one or more two dimensional displays the data from which each sort decision is made.

13. An apparatus for making classification decisions in real time regarding a plurality of events comprising:

data gathering means including a plurality of light detectors for asynchronously gathering a plurality of digital data derived from signals generated by said light detectors from scattered and fluorescent light caused by objects tagged with fluorescing dyes which have been passed through a plurality of laser beams using a plurality of channel means for storing said data, said digital data defining a multidimensional space describing each said event and said data gathering means including means for temporarily storing said digital data in a separate FIFO buffer for each channel means, said data gathering means including an event detection means for detecting when an event occurs which will cause a pulse signal to occur in each said channel means, and for generating a sample signal upon occurrence of said event, and, after a delay tailored to match the individual delay for any particular channel before said pulse signal in that particular channel peaks, for sending said event signal to that particular channel, and for repeating this process for each said channel means, each said channel means including means to sample said pulse signal upon receipt of said event signal and for converting said sample to said digital data and storing said digital data in said separate FIFO buffer for that channel means, all said channel mean storing digital data pertaining to the same event in the same relative positions in said FIFO buffers, said data gathering means including attenuation means for attenuating the data gathered by each said channel means by a programmable amount and for outputting said attenuated data from each said channel means as a compensation signal, and further comprising compensation means in each said channel means for receiving said compensation signals from predetermined ones of said attenuation means in the other channel means and for summing the compensation signal received and further attenuating the data gathered by said channel means by an amount proportional to the sum of said compensation signals received from said predetermined ones of said attenuation means in said other channel means;

comparing means for resynchronizing the digital data pertaining to a particular event by retrieving the digital data pertaining to one event from selected ones of said FIFO buffers in said channel means and for comparing in real time said digital data regarding each said event to the nodes of a binary classification tree and including means for making a classification decision based upon the comparison results at each said node.

14. The apparatus of claim 13 further comprising a buffer means in said data gathering means for temporarily storing said digital data until it is needed by said comparing means.

15. The apparatus of claim 14 wherein said buffer means comprises a plurality of FIFO buffers.

16. The apparatus of claim 14 wherein said events being sorted correspond to said digital data derived by said data gathering means from scattered and fluorescence light from living or dead cells which are tagged with fluorescent dye, said cell being passed through at least one laser beam, and further comprising means for entraining said cells in a stream of conductive fluid, and further comprising means for passing said stream through the beam of at least one laser having a wavelength selected to excite said fluorescent dye, and further comprising at least two scattered light detectors and at least two fluorescent light detectors for detecting light emitted from said fluorescent dye and light scattered from said cells.

17. The apparatus of claim 16 wherein each cell has attached thereto at least two fluorescent dyes, and further comprising a second laser for emitting coherent light in a beam through which said cells pass, said light having a frequency to excite fluorescence in a different dye than was excited to fluorescence by the light from said first laser and wherein at least some of said light detectors are fluorescent light detectors for detecting fluorescent light emitted by said dyes.

18. The apparatus of claim 17 wherein said means for gathering includes a nozzle for guiding a stream of electrically conductive fluid containing said objects, and further comprising a drop charging electrode in electrical contact with said stream, and further comprising means for vibrating said nozzle so as to cause said stream to break up into droplets sized to carry at least one said object, and further comprising means for converting said classification decision into a drop charging signal and applying said drop charging signal to said drop charging electrode as the object to which said classification decision pertains reaches the point where said droplets are formed.

19. The apparatus of claim 18 wherein said attenuation means comprises means for reducing noise in the signals from said light detectors caused by sources other than the light to be detected by any particular detector.

20. The apparatus of claim 19 wherein said data gathering means includes a matched filter for each said light detector, said matched filter having filter characteristics matched to the characteristics of the signals generated by the corresponding said light detector in said data gathering means.

21. The apparatus of claim 20 wherein each said channel means further comprising an amplifier with a logarithmic transfer function and an amplifier with a linear transfer function, said apparatus for making classification decisions further comprising a user interface computer including means for allowing either said logarithmic or said linear amplifier to be selected to amplify the signals from the light detector corresponding to said channel means and further comprising means for selectably setting the gain of said linear amplifier through said user interface computer.

22. The apparatus of claim 21 further comprising means for sensing signals from said light detectors only if they are above a desired threshold, and for detecting the peaks of said signals and outputting an event signal at the time of occurrence of said peak.

23. The apparatus of claim 22 wherein each light detector corresponds to one said channel means, and further comprising a sample and hold amplifier in each said channel means for sampling the signal from the output of the selected amplifier and further comprising digital-to-analog conversion means for converting the sample value to a digital value and wherein the digital value from each digital-to-analog conversion means is stored in a separate FIFO buffer in each said channel means, and further comprising a separate delay means in each said channel means for each corresponding light detector, each said delay means having an input coupled to receive said event signal from said means for sensing and peak detecting and for delaying the output of said event signal by a programmable amount for each said light detector, and wherein said user interface computer includes means for selectably setting the desired amount of delay for each light detector, and wherein said event signal for each light detector is coupled to the corresponding sample and hold amplifier for that light detector to cause said sample and hold amplifier to sample the signal from said light detector.

24. The apparatus of claim 23 wherein said means for vibrating said nozzle is coupled to said user interface computer and wherein said user interface computer includes means for selectably setting the drop formation rate.

25. The apparatus of claim 24 further comprising two separate event FIFO buffers selectively coupled to the output of said FIFO buffers for storing the digital data from each light detector, one of said event FIFO buffers for storing the event data temporarily until said means for comparing can read the event data and make a classification decision, the other said event FIFO buffer for storing said event data until the classification decision is made and said user interface computer requests said event data and the corresponding classification decision, said user interface computer further comprising means for displaying said event data.

26. An apparatus for making classification decisions in real time comprising:

means for collecting multidimensional analog data about an event;

means for subtracting out noise from said data, and filtering said data to maximize the signal to noise ratio;

means for converting said data to digital data by sampling said analog data at its peak and doing an analog to digital conversion on it;

first means for temporarily storing said data while awaiting a classification decision;

means for retrieving said selected data from said storage means and for comparing it to a classification decision tree and for generating a classification decision based upon said comparison;

second means for temporarily storing said classification decisions until they are needed; and

means for retrieving said classification decisions from said second means and for classifying said event based upon the classification decision retrieved from said second means pertaining to said event.

27. The apparatus of claim 26 wherein said data gathering means includes two lasers emitting coherent light at two different frequencies and a plurality of scattered light detectors and a plurality of fluorescent light detectors, each detector including means to block light from all but a single range of wavelengths tuned to the emission spectra of a particular dye in a plurality of dyes used to stain objects being illuminated by said lasers.

28. The apparatus of claim 27 further comprising means for compensating the signals from said light detectors such that the signal derived from each light detector after compensation has an amplitude which is due substantially solely to emissions from a single selected dye.

29. A method for classifying events in real time comprising the steps of:

asynchronously, electronically gathering a plurality of data about an event, said data defining a multidimensional space having more than three dimensions;

resynchronizing said asynchronously gathered data by storing said data in a plurality of first-in-first-out buffers, one such buffer being dedicated to each dimension of the multidimensional space such that all said data pertaining to a single event in said multidimensional space can be retrieved at substantially the same time when said event is to be classified;

reading data about said event out of said first-in-first-out buffers and comparing said data in real time to data organized in the form of a binary classification tree;

formulating in real time a classification decision based upon the results of the comparison against said binary classification tree;

storing the classification decision in a buffer such that said classification decision can be retrieved later when said classification decision is needed to take action regarding the event to which it pertains; and

retrieving in real time the classification decision regarding a particular event to be acted upon, and taking the action specified in the classification decision in real time.

30. The method of claim 29 wherein the step of gathering data includes the steps of shining light of a predetermined wavelength on objects having one or more fluorescent dyes on them capable of being excited by light of said predetermined wavelength and detecting the amount of scattered light at predetermined angles and the amount of light emitted by said dyes.

31. The method of claim 30 further comprising the step of filtering light emitted from said dyes such that a plurality of light detectors each receives light only in a specified band of wavelengths corresponding to the emission band of a particular one of said dyes.

32. The method of claim 31 further comprising the step of compensating the signals derived from said light detectors after filtering to eliminate the effects of light emitted in the passbands of the various filters generated by dyes the light from which is not desired in said passbands such that the signal derived from each light detector is caused substantially by only one dye.

33. The method of claim 32 wherein each said light detector emits a pulse when an object passes through said light, and further comprising the step of filtering the output of each said light detector so as to remove the average noise level, and of maximizing the signal to noise ratio by matched filtering of the pulse emitted from said light detector.

34. The method of claim 33 wherein the step of taking action on the classification decision includes the steps of interpreting the classification decision, converting each said classification decision to a pulse having a predetermined magnitude and polarity.

35. The method of claim 34 further comprising the steps of entraining objects to be sorted into a train of electrically conductive liquid flowing through a nozzle, vibrating the nozzle so as to break the stream into droplets some of which contain objects, shining coherent light on said stream of objects, using said pulses generated from each classification decision generated from data gathered about each object to charge the droplet containing said object as it breaks from said stream, and deflecting said charged droplet into a sorting bin designated for objects of that type.