From: David Basiji [basiji@amnis.com] Sent: Tuesday, July 17, 2007 1:37 PM To: Cytometry Mailing List Subject: RE: analog/digital and offline compesation Marty, You're right, at the beginning of the signal detection chain, the CCD is an analog device. To elaborate a bit, the pixels collect photoelectrons, they are converted to a voltage during readout, and the preamp is connected to A/D converters which are off-chip in our camera. As long as the camera is designed to operate with the CCD well within the appropriate range of its photon dose vs. pixel charge curve, linearity can be very good. Since the analog stages up to the A/D are on-chip, their performance is generally stable over time. It's also pretty straightforward to perform gain and offset correction for every pixel to compensate for inherent variation across the chip. As a result, digital compensation of image cytometry data can work very well. Of course, image cytometry has its own set of design issues that standard flow cytometers don't have to worry about, mainly having to do with maintaining consistent optical performance over a wide spectral range. Without that, you can't be sure that pixel(x,y) in any given image is actually measuring the same location in the cell as pixel(x,y) in all the other images. If that base truth isn't maintained, the effectiveness of pixel-by-pixel compensation degrades rapidly. Fortunately, this is truly a design issue and not something that varies operator to operator or day to day. Best, David David Basiji, Ph.D. President and CEO, Amnis Corporation 2505 Third Ave., Suite 210 Seattle, WA 98121 (206) 374-7165 direct (206) 919-3342 mobile (206) 576-6895 fax www.amnis.com This email and any attachment contain information which is private and confidential and is intended for the addressees only. If you have received this email in error, please notify the sender and delete it. > -----Original Message----- > From: Marty Bigos [mailto:mbigos@gladstone.ucsf.edu] > Sent: Saturday, July 14, 2007 9:59 AM > To: Cytometry Mailing List > Subject: Re: analog/digital and offline compesation > > Larry- > > A case could be made that the Amnis Imagestream, with its CCD > detection system (like a digital camera) is a full digital flow > cytometry system. > > Then again, on the CCD chip, are preamps (to set the ISO) and ADCs (to > convert the amplified CCD signal to n-bit digital values), so one > can't get away from the fact that we live in an analog world. > > Marty > > > > On Jul 12, 2007, at 12:49 PM, Larry W. Arnold wrote: > > > From: lwarma@med.unc.edu > > Subject: Re: analog/digital and offline compesation > > Date: July 12, 2007 12:49:44 PM PDT > > To: cytometry@flowcyt.cyto.purdue.edu > > > > Marko > > > > The main problem with doing it on the machine is - Will you do it > > right - i.e not overcompensate? Overcompensation is a problem for > > most users when they do it "manually" (i.e. by eye). > The Calibur > > is analog but of course the data has to in the end become digital > > for the computer to understand it. "Digital" machines, > like a B-D > > LSRII, have no log amps (or linear amps) - the data is taken > > directly from the pre-amp to high resolution ADCs - the log > data is > > generated from a log lookup table. All machines are of course > > hybrids - up through the pre-amp is analog electronics. What > > digital machines do primarily is get rid of the log amps although > > there are other things the "digital" part does like > allowing you to > > trigger on lasers other than the first laser the cell encounters > > since the digital data stream is preserved one can look back (or > > forward) in time to find the other pulses. In addition the > > triggering part of the machine is different in that rather than > > having the older "hardware discrimination circuits" the machines > > now trigger on individual pulses - starting and closing the window > > on each pulse rather than just opening the window for a defined > > length of time. You can always do "off-line"/digital compensation > > on any data. However, when that data generation > involves log amps > > (unless you do something very specific to correct it) there will > > be issues because log amps are not perfectly logarithmic. Log amp > > data fully compensated digitally (i.e. perfect mathematics) will > > reveal the imperfect log aspects of the log amp and you will see > > (after compensation) wavy/serpentine aspects to the data as you > > compensate it if you have data that is smeared across the entire > > voltage range. The problem is that if you have discrete populations > > of cells you may not realize that the data is not correctly > > compensated. Remember digital/software > compensation only > > reveals the issues that already existed - it does not create them. > > > > On the Calibur what we recommend is that the user compensate with > > the analog circuitry part way being sure not to > over-compensate and > > then finish the compensation off-line digitally (although I don't > > think very many take my advice). The log amp problems will be > > minimized by the "messy" analog compensation. > > > > Hope this helps. > > > > Larry > > > > > > > > At 06:14 AM 7/12/2007, you wrote: > > > >> Dear all, > >> > >> I use FACSCalibur for all my cytometry needs, which are mostly > >> simple but create magic for me :) Not that long ago I started to > >> use SUMMIT software (thanx Denis > >> P.) for offline compensation of my data (usually annexinV/PI) and > >> a colleague of mine uses FlowJo for her two color > purposes. And it > >> seems to be working just fine comparing to direct compesation on > >> the machine where I have to compensate FL2/FL1 almost 100% to get > >> a text book picture, and not a banana lookin-like dot plot :) I got > >> 2 questions: > >> > >> 1. I heard that FACSScan and FACSCalibur are analog machines and > >> that it is not possible to do offline compensation and that it > >> should only be done on the machine? (btw I thought that all > >> machines are analog and just have AD converters..?) is that true?? > >> > >> 2. What cause could produce an effect that I need to compensate > >> that much while doing it on the machine? (geometry of the laser and > >> mirrors/detectors, age of the machine...?) > >> > >> Thanx , Marko. > >> > >> -- > >> Marko Marjanovic, B.Sc. > >> Laboratory of Functional Genomics > >> Division of Molecular Medicine > >> Rudjer Boskovic Institute > >> Bijenicka 54, Zagreb, HR-10000 > >> Phone: +385 1 4561111 (1772) > >> E-mail: mmarjano@irb.hr > >> http://www.irb.hr/en/str/zmm/LABS/LFG/djelatnici/marko/ > >> > > > > Larry W. Arnold, Ph.D. > > Research Professor and Director, Flow Cytometry Facility Department > > of Microbiology and Immunology CB# 7290 University of North Carolina > > Chapel Hill, NC 27599 > > Phone: 919-966-1530 > > FAX: 919-962-8103 > > > > ########################################### This email and any attachment may contain information which is private and confidential and is intended for the addressee only. If you have received this email in error, please destroy it and notify the sender by return email.