12/10/98

Hello all again,

It seems that cost cutting and space saving have finally started attacking

us! Our darkroom, which I admit is rather large, is wanted by another more

influential person. It has been suggested that we attach a digital camera

to our Hitachi H600 TEM to take over from the film camera we currently use.

I have suggested that digital images are not yet able to produce images as

good as our film can and that the digital image can not be increased to a

much larger size without appreciable loss of resolution. Can any one please

help us find a solution as this same influential person also has their beady

little eyes on our laboratory!

We need to know

1. How easy is it to digitise the TEM without losing the ability to

still turn to film.

2. What sort of camera/system should we be looking at?

3. What sort of software is needed?

4. How much do these things cost?

Sarah Ellis

Trescowthick Research Centre

Peter MacCallum Cancer Institute

Locked Bag #1 A'Beckett Street

Melbourne 8006 Victoria

Australia

Phone +61-3-9656 1244

Fax +61-3-9656 1411

Email s.ellis@pmci.unimelb.edu.au

Sarah

I use a Hitachi H7000 and, although I don't have your urgent requirement, I

have looked at the convenience issue of digital photography on a TEM.

My understanding is that you would probably need to insert some sort of

fluorescent screen and high resolution digital camera at either the 35mm

camera port or STEM detector port position of the microscope. At present

this would either involve Hitachi or a specialist company that you could

trust. By the time that you have paid for all the specialist kit the numbers

I have been quoted don't come much below 30,000 UK pounds. It's probably

cheaper to get the system integrated into a new microscope if and when you

upgrade or have some spare petty cash you're not using.

However - have you considered a simpler and much cheaper option. Develop TEM

film in a simple darkroom (in an emergency any dark ventilated area will

do). Then scan negatives with the highest resolution negative/transparency

scanner you can get. I would guess that you could pay anything from 2,000 to

10,000 UK pounds depending on what you wanted and this would include

scanner, fast computer , storage, software and printer. The added advantage

is that from a typical negative and high resolution scanner it should be

possible if necessary to scan at much better resolutions than with the best

digital camera. I would guess that you would be able to print at up to at

least 5x the original negative magnification if you used the right scanner.

This is my master plan because you still have the archival and universal

properties of film but most of the convenience of digital plus you could at

some later date do a full up-grade. I'm still awaiting money to do this but

for the outlay the benefits greatly outweigh the costs.

Good luck and let me know how you get on.

Malcolm Haswell

Electron Microscopy

School of Health Sciences

Fleming Building

University of Sunderland

SUNDERLAND SR1 3SD

Tyne and Wear

Tel (0191) 515 2872

e-mail: malcolm.haswell@sunderland.ac.uk

One possibility is to use photoplates. I don't have any personal

experience with these, but my understanding is they are basically digital

film. The microscope does not need to be modified. Photoplates are used

in the same fashion as film. I understand they have a better dynamic range

than a CCD. They are also much higher resolution than a CCD. The down

side is cost. A reader costs in the $15,000 USD range I believe, and the

plates are close to $100 USD each. Also each image is 25 - 30 megabytes so

disk storage and backup for the images can add up in a hurry.

All of the above is from memory of a conversation with someone who had this

setup. The numbers may be misremembered. Hopefully someone who actually

has bought and used photoplates will respond.

Tom

Thomas Mullarkey Murray email:tm8a@virginia.edu

Thornton Hall - MSE phone:(804)982-5659

University of Virginia Fax: (804)982-5660

Charlottesville, VA 22903

We have been digitizing our images since 1995. Our solution to the massive

resolution loss in our biological digital images (301K) is to take films as well

(fewer) which serve as archives. My experience is you need 3 digitized images at

different magnifications to get out the same information contained in one film. Films

may be scanned as 3-4MB images and printed on a good 1200 dpi dye-sub printer or

printed the old-fashioned way.

Kate Connolly

Katherine.S.Connolly@dartmouth.edu

We only use a very small darkroom to develop the film. Basiclly a closet

next to the scope. Then we scan the negs at either 600 dpi or 1200dpi,

stor the files on CDs and print on good printers. We have gotten addicted

the enhanced information we get by viewing the images on the monitor, with

image enhancement and analysis tools at the click of a mouse. For us it

seems to be the best solution for now. We want to bypass film eventually

but the cost and drawbacks seem too great for now.

Bob

Derm Imaging Center

U of W

underwoo@u.washington.edu

Hello Sarah,

You've asked an interesting question (at least to me that is).The

organisation I work for may be buying an old TEM in the near future

which could be upgraded to digital. So I'd be interested in hearing

all the responses. For example, do digital cameras really offer a

chance for us to get out and even stay out of the dark room?

(apologies if this question sounds simple). I know that one

would need a slow scan CCD rather than a TV rate digital

camera for suitable quality images.

I'd be interested in all your thoughts on the subject.

Regards

Martin Roe

MLURI

Craigiebuckler

Aberdeen

Scotland

U.K.

mi596@mluri.sari.ac.uk

To all concerned:

I am also in the process of digitizing our tem. We have a Zeus 902 cem,

fortunately it has a side mount for a 35mm camera already in place. Even

with a very good camera, there must be a loss of resolution compared to

film. Electrons exposing the film emulsion are highly resolved, whereas a

fluorescent image is inherently less resolved. We have always said that you

can see more on the negative than the view screen. Has anyone evaluated

different resolution cameras to see at what point pixel array size becomes

irrelevant when imaging a fluorescent screen?

William McManus

Supervisor

Electron Microscopy Facility

Department of Biology

Utah State University

Logan UT 84322-5305

billEMac@cc.usu.edu

435-797-1920

Sarah,

Since you raise questions that probably are of interest to many people,

I will try to provide some information. Please keep in mind, that my

company, Soft Imaging System Corp, produces and sells these systems and

I might therefore be a little "biased" towards digital imaging. I will

also forward your email to our agent responsible for Australia so he can

provide you with more specific information.

1. How easy is it to digitise the TEM without losing the ability to

still turn to film.

That is usually not a problem. The digital cameras are either

side-mounted on a 35 mm port or bottom mounted below the film chamber.

in most cases (if not all), you can retain the film camera and take

images on negatives if that is required.

2. What sort of camera/system should we be looking at?

This and the following questions require a bit more thought. The answer

usually depends on what you want to to. As I mentioned above, there are

basically two different types of camera: side-mounted and bottom

mounted. One clear differentiation is the resolution and field of view

you can get from these cameras. Since the side-mounted cameras are

mounted above the viewing chamber, they "see" about the same area as a

negative, sometimes more. On the other hand, the bottom mounted cameras

usually see only a small part of the negative area. The side-mounted

cameras are better suited for applications where field of view is an

issue, the bottom mounted ones are better suited for applications where

resolution is the most critical.

Let's have a quick look at side-mounted cameras. They come in various

types and configurations. The simplest type is a simple TV camera, where

the CCD chip is either lens or fiber coupled to a small scintillator

that intercepts the beam (removable). This type of camera is great for

teaching and finding areas on a sample, as you get a real-time image on

a screen, but the low resolution (640x480 for US standards) makes it

less desirable for actual image acquisition (but see below, Multiple

Image Alignment).

Another type of cameras are digital cameras for side-mount. They come in

many possible sizes and configurations. Usually they have a better

resolution and a higher dynamic range (12 bit as opposed to 8 bit for TV

cameras), but often they require specific frame grabbers, and the frame

rate (numbers of images per second) can drop significantly. That does

not seem to be to important at first, but try to adjust astigmatism with

a camera that delivers less than 10 frames per second. I personally find

it impossible to do. I would say, that a good camera must deliver a

resolution of at least 4 times TV (i.e., 1280 x 1024), have a dynamic

range of at least 12 bit and a frame rate of 10 or more at full

resolution. It should also be able to accept exposure times of several

seconds for those darkfield images, which proabably means cooling the

camera.

Then there are specialty cameras: high sensitivity cameras, extremely

fast cameras, etc. Those can be used, but they are normally not required

for "normal" TEM.

Now come the bottom mounted cameras. Almost all of them are fiber

coupled for better sensitivity. They are usually square and come in

1Kx1K and 2Kx2K chips. The 2K chips are MUCH more expensive than the 1K

chips. They ususally do not have the readout speed of side-mounted

cameras, and are basically replacements for negatives. Their lower read

out frequency allows better digitization and they are frequently

digitized to 14 bit or better. they do only see a part of the negative,

and together with the lower readout speed, it can be hard to find the

correct area. Most of them, however, have some features (binning, etc.)

that allows faster image repetition rate at a reduced resolution. these

cameras are often liquid cooled and either hook up to the TEM cooling

system or have their own chiller.

Of course you can have both cameras attached at the same time. That way

you can pick what camera to use depending on the application.

3. What sort of software is needed?

That again depends on what you want to do. As a minimum I would think,

that you need image acquisition from the camera you select (obviously),

some gray scale manipulations, non-destructive overlays for

documentation, an image database for archiving and retrieving (should be

network capable and compatible with other databases), printing

capabilities that include Microscopy specific things like scale bars,

true magnification, etc. The software should be able to maintain

calibration data for different camera setups, if possible communicate

with the TEM, and, again if possible, allow for motorized stage control.

Important for TEM is also online shading correction to compenstate for

fluctuations in the scintillator and illumination, and real-time

contrast maximization. This is especially important for 12 bit cameras,

as you may have to look for the "right" 8 bits to display and illuminate

the sample all the time. Auto-contrast cuts down on the dose. Depending

on how you work, a single screen system (with the operating system and

the images sharing one monitor) or a dual screen system (with the images

displayed on a separate monitor) is best. If you want to use the digital

system for astigmatism correction, you need a real-time FFT feature. If

you need particle detection Fourier filters, Programming, etc., this

should be available also, but is not really necessary for simple image

acquisition. You want to keep this option open, though, by selecting

software that offers an upgrade path.

4. How much do these things cost?

Ahh, the most important question. The answer is: From anywhere from

several thousand $US to hundreds of thousands of $US, depending on

manufacturer, quality, software, hardware, cameras, and your demands. I

know, this answer is not very satisfying, but it is just as impossible

to answer as the question: how much does a car cost. As I said, I will

forward your email to our person in Malaysia, and he can answer those