BioBench - a workbench for biologists

Alain EMPAIN

National Botanic Garden of Belgium

Summary

This page centralizes the project information :

April 2007 : discovery of the hi-resolution (1.3 Mpix) webcam

May 2007 :
first success with Logitech's Quickcam for Notebook Pro 1.3 Mpix, at the resolution of 960x720
(UVC driver, Laurent Pinchart and luvcview, Michel Xhaard)

definition of Taxbench GUI, written in Qt4-ruby

new version of luvcview (20070512), allowing to take 1280x960 pictures

September 2007 : Advance at the BioBench front (PDF)
layout restructuration

more working parts

multiple measurement objects on the same image (counts, segments, length/width...)

interface to the 'R' statistical package (start)

The camera

First results

First images taken at a lower resolution (960x720); for the microscope, the optical axes are not yet well aligned : this is the first steps of a work in progres.	Hires Webcam for macrophotos; 960x720
May, 14th : Michel Xhaard email me the new version of luvcview, supporting now to save pictures with the highest resolution (1280x960) !	Hires Webcam for microphotos; 960x720
First pictures taken at the highest resolution (1280x960), owing to the newest luvcview release	the reference mid-gray grid background (white squares = 1 cm2), webcam objective heavily screwed off
	the same grid background (black squares = 1mm2), at the printer wax drop level (Tektronix 850) : the webcam objective is fully unscrewed, then gently screwed for 1/4 turn
	a leaf of bryophyte (moss), seen with a microscope and the objective-less webcam

A 'good-enough' approach

As a rule, the research apparatus are very expensive : the camera provided with our Olympus research microscope is not an exception : costing more than 6000 euro with its application program. Moreover, the program requires a Microsoft platform.
It would be a pity to switch to Microsoft just to use a camera.

As the new webcams have a higher definition (1280x960 pixels), it makes now sense to use them for the day-to-day work.

The image taken with the cooled professional camera is surely less noisy, but I hope to cope with this problem when needed with the averaging of a few snapshots.

Requirements to mount a camera on a microscope or a binocular

The camera must be lightweight and the objective MUST be removable. I choose the Logitech's Quickcam for Notebook Pro 1.3 Mpix for its exceptional small size, and despite seemingly fixed, its easily removable objective.

General view; the objective seems fixed
one small screw is securing the two parts of the camera enclosure
the enclosure opened : the objective is screwable, and will allow to make close-ups (see the first examples) and to be easily removed for use on a microscope; see the thin screw in the background REM: the special background is designed to provide a scale with color references, and a mid-gray average to help the white balance of macrophotographies
the objective is now unscrewed
the CCD chip is now visible
proof-of-concept : first try to position the CCD chip along the optical axis of the microscope : the carboard piece
and a cylindrical cardoard part (fastened with duct tape) to position the ensemble above the vertical camera tube of the microscope
or even in place of the regular eyepiece : ordinary microscopes do not have the vertical camera well, and the especially lightweight quickcam allows to transform any microscope into a microphotography workplace	Very rough try !

BioBench, the GUI

First working steps :

September 2007 : Advance at the BioBench front

(PDF)

Goal

The BIOLOGIST worbench (version 0.1.3 : 2007-05-11) is designed to interface the images taken from a biologist's day to day work.
In fact it may be of more general usage, but I begin to solve my own problems first !
It will allow to :

interface directly with the connected acquisition(s) device(s) [+]
identify the sample (collector-collnum : 'BR:Empain.A-1789' ) and the magnification (CAM/BIN/MIC: mm/1000pixels) [~]
label what it is(4 hex digit hierarchical code : say 4552='branch: leaf: apex: cells' ) [~]
rename the image as name+code (fi. easy retrieval of all apical cells images, or any branch-related images...) [~]

On a list of images, must allow to :

browse the new pictures acquired from the camera/macro/microscope [*]
zoom, crop, enhance the image [+]
define, label and export ZOI (zone of interest) [+]
insert an explicit scale (linear or as a grid) [*]
make counts and measures [*]
output statistics and interface them with R for subsequent analysis [+]
bundle the acquired information, statistics and subsamples (ZOI) in a bag file 'a la OpenOffice', perhaps keeping the annotations and ZOI simply as locations/text and rectangle definition, to allow to build on-the-fly the derivated images -- keeping permanently only the untouched master image [~] build true RDBS records owing to the embedded information [~]
support scriptable default behaviour for repetitive tasks [~]
can be distributed on a bootable CD

[*]:working, [+]:in development, [~]:project

Alain EMPAIN -- 2007

Implementation

In previous developments, I used perl and Qt3; I discovered Ruby last year and gave it a try with its Qt3 bindings. Now, I decided to begin with Qt4-Ruby to develop my knowledge of this nice couple.