一種新型自動影像分析系統(tǒng)，能夠?qū)Σ煌愋偷陌准?xì)胞進(jìn)行定位以及預(yù)分類。

這種分析是通過人工神經(jīng)網(wǎng)絡(luò)進(jìn)行的，這種網(wǎng)絡(luò)是根據(jù)大型細(xì)胞數(shù)據(jù)庫進(jìn)行培訓(xùn)的。分類前細(xì)胞能夠清晰地呈現(xiàn)在屏幕上，便于鑒定。這套系統(tǒng)叫做CellaVision DM8，是公司CellaVision DM96型的半自動版本。

這套系統(tǒng)瞄準(zhǔn)中型實驗室，系統(tǒng)具有半自動切片處理功能，一次可以處理多達(dá)8張切片。這套系統(tǒng)還能夠與其它安裝在同一地點(diǎn)或者邊遠(yuǎn)位置的CellaVision DM產(chǎn)品共享數(shù)據(jù)庫，允許通過局域網(wǎng)（LAN）遠(yuǎn)程進(jìn)入檢查結(jié)果和圖像。這套系統(tǒng)是瑞典CellaVision公司(Lund, Sweden)產(chǎn)品。2005年3月在阿姆斯特丹實驗室自動化會議上CellaVision DM8在歐洲開始投放市場。

CellaVision的DiffMaster能夠通過自動化顯微鏡以及先進(jìn)的影像處理功能來對血細(xì)胞進(jìn)行定位、鑒別和預(yù)分類。顯微鏡連接在安裝了基于大學(xué)細(xì)胞數(shù)據(jù)庫軟件的計算機(jī)上。軟件控制顯微鏡和一臺發(fā)現(xiàn)并記錄血細(xì)胞圖像的照相機(jī)，然后給出細(xì)胞類型的分類。在絕大多數(shù)情況下，操作者只需要確證分類，以及側(cè)重于那些處于邊緣的分類即可。

Automated Analysis of White Blood Cells
 
A new automated image analysis system locates and pre-classifies the various types of white blood cells.

The analysis is performed by an artificial neural network, trained on a large database of cells. The pre-classified cells are clearly presented on screen for easy verification. This system, called CellaVision DM8, is a semi-automated version of the company’s CellaVision DM96.

The system is directed at medium-sized laboratories with its semi-automated slide handling of up to eight slides at a time. It can share databases with other CellaVision DM products installed at the same or remote locations, allowing remote access to results and images via a local area network (LAN). The system is the product of CellaVision (Lund, Sweden). The CellaVision DM8 was launched in Europe at the Laboratory Automation conference in Amsterdam in March 2005.

CellaVision’s DiffMaster locates, identifies, and pre-classifies blood cells by means of automatic microscopy and advanced image processing. The microscope is connected to a computer equipped with software based on a large cell database. The software controls the microscope and a camera that finds and photographs images of blood cells, then suggests classifications of the cell types. In most cases, the operator merely confirms the classifications and can focus on borderline classifications.